POSITIVE TRENDS

The 2004/05 crime statistics

Antony Altbeker
Institute for Security Studies

Published in Crime Quarterly No 14 2005

In September 2005 the SAPS released its latest batch of crime statistics for the country, covering the period April 2004 to March 2005. The trends that emerge from these figures suggest that South Africans should be encouraged by the successes shown in managing down the levels of crime. This article describes some of the key trends and explains what can and cannot be read into them.

Spring is South Africa’s annual crime statistics season, the time of year when the South African Police Service (SAPS) releases figures on the number of cases it recorded in the financial year that ended in the preceding March. The release of these data coincides with the presentation of the SAPS Annual Report. This document – one of the most impressive of the annual reports put out by any government department – contains far more information than just the crime statistics; however, for obvious reasons, it is these on which the press and public fixate.

The public obsession with crime statistics is perfectly understandable. These are, after all, among the most important measurements of the health of our nation and the state of its social fabric. There are, of course, problems with using police crime statistics as the sole measure of the level of crime. The essential problem here is that what is reflected in official statistics are only those crimes that were brought to the attention of and recorded by the police. It is by no means a full and complete accounting for the actual amount of crime committed in the country in that year. Before looking at this, however, the basic facts as we have them from police statistics will be reviewed.

How much crime was reported in 2004/05?

Although the SAPS provides data on 25 crimes (four of which are sub-categories of aggravated robbery), it is not unreasonable to say that there are eight forms of criminality which dominate public concern about crime: murder, attempted murder, serious assault, and rape (collectively, ‘violent crime’); and aggravated robbery, common robbery, car theft and housebreaking (collectively, ‘property crime’). As Figure 1 depicts, the aggregate of these crimes rose from about 855,000 in 1994/95 to nearly 1.1 million in 2002/03, before falling to 980,000 in 2004/05.

This meant that in 2002/03, the sum of all these crimes was 28% higher than it was in 1994/95, with violent crime being 20% higher, and property crime 32% higher. Since then, violent crime levels have fallen by 8% and property crime levels have fallen by 11%. All of this is good news. Indeed, on closer examination, the news is even better. Figure 2 shows that murder (which has fallen fairly consistently for ten years) is well below the level of 1994/95 and attempted murder (having fallen steeply in the last two years) is, for the first time, also lower. Rape and assault GBH account for over 70% of incidents of serious violence.

A similar picture emerges when we look at serious property crime (Figure 3). Car theft numbers in 2004/05 were lower than in any previous year, burglary (which includes both business and residential housebreaking) is off its peaks, as are both common and aggravated robbery (though the last has fallen only in 2004/05).

The bottom line, then, is that crime levels, as revealed by police statistics, appear to be improving, and to be improving across a broad front. The obvious objection to this conclusion, however, is that police statistics do not capture all the crimes committed in a given period. The question that arises is how seriously we ought to take this conclusion.

Are the statistics right?

It is well known that police crime statistics tell only part of the crime story. This is so because for a crime to be recorded in the statistics three things must happen: it must be noticed, it must be reported, and it must be accurately recorded. The last two steps mean that doubts are often expressed about the reliability of official statistics.

The under-reporting of crime by victims is a worldwide phenomenon. It arises principally from the fact that many victims of crime do not believe there is much purpose in reporting a matter to the police. This is either because they believe the incident was minor, or because they don’t believe much purpose will be served by making the complaint, since the perpetrator cannot be readily identified or, indeed, because they know who the perpetrator is and they would prefer not to subject him or her to criminal sanctions.

The result is that in a survey of 5,000 South Africans conducted in 2003, 71% of those who reported having been robbed said they had not reported the matter to the police, with the same being true of 64% of victims of stock theft, 59% of victims of theft, 45% of victims of assault, 43% of burglary victims and 36% of those who had something stolen out of a motor vehicle.1 This pattern is not unique to South Africa, however, as Figure 4 reflects.2

Nor is under-reporting the only problem. Under-recording is also a relatively common phenomenon – and one that is often badly misunderstood, since it is often assumed that when police officers fail to record a crime that has been reported to them, they must have done so in order to deceive the public into believing that crime levels are lower than they really are. This may well be one factor in explaining the disparity between reported crime levels and recorded crime levels, but there are other, less malevolent reasons why some cases may not get reported.

One example of a crime which comes to police attention but does not get into the official statistics are the thousands of relatively petty domestic disturbances to which officers are called every month but which are dealt with (they are not ‘resolved’) at the scene. These are cases in which a crime may well have been committed – most will involve one or more of crimen injuria, common assault and malicious damage to property, but some might also involve theft – and has come to police attention, but has not made it into official statistics, for reasons other than the desire to deceive the public.

The upshot of all of this is that there are real reasons to doubt that police crime statistics reflect every crime committed in any given period. Indeed, it is doubtful that this is the case anywhere in the world. It does not follow that police statistics should be dismissed, however. This is for two reasons.

One reason why it is useful to look at police statistics is that, even if they may not fully account for crimes committed, it is possible that the trend reflected is not dissimilar to the underlying reality of crime levels. Thus, if the rates of under-reporting and under-recording are relatively stable, it may be that the trend we observe in the statistics reflects a real trend in crime levels.

It is not, of course, possible to show that reporting and recording rates are consistent over time unless surveys are conducted regularly. Indeed, although surveys conducted by ISS are not done as regularly as would be ideal, they do show that there have been some changes in the reporting rates for some crimes.

The 2003 victimisation survey, for instance, found that reporting rates had generally risen in South Africa from the levels recorded in a 1998 survey: theft out of vehicles was reported 64% of the time compared to 57% in 1998; assault was reported 55% of the time compared to 38%; and the reporting rate for theft was 41% compared to 28%.

Since reporting rates for these crimes appear to have risen it follows that any changes in the recorded levels may overstate any rise or understate any fall in the actual number of crimes committed – assuming that recording practices have not changed at the same time.

In contrast to this generally positive trend, the reporting rate for robbery fell from 41% to 29%, while the rate for housebreaking was unchanged at 57%.3

A second reason why looking at police crime statistics can tell us something about the real crime levels is that some crimes are not under-reported or under-recorded to any significant degree. The two most obvious examples of crimes falling in the category of well-reported and well-recorded are murder (because of its seriousness and because there is a body which needs to be accounted for),4 and car theft and carjacking (which are well reported for insurance purposes). It turns out that when we look at these two crimes, they both reveal significant declines over a reasonably long period (Figure 5).

The number of murders peaked in South Africa in 1995/96 at 26,877 and has since fallen by 30% to 18,793 in 2004/05. In per capita terms, because the population has grown, this represents a fall of over 40%. Car theft numbers peaked later, in 1998/99, at 107,500 and have since fallen by 22% to under 83,900. This trend is broadly matched by car hijacking numbers, which have also come off by 21% since their peak of nearly 15,800 in 1998/99.5

Although it need not follow that declines in these well-reported and well-recorded crimes necessarily imply that the downward trends of other crimes reported in the statistics are also an accurate reflection of crime trends, they do suggest that real progress is being made in key areas. In this regard, we should look at those few crimes for which increased numbers of incidents have been recorded.

Increases in some crime categories

Although the majority of the most serious crimes saw decreases in 2004/05 relative to 2003/04 (with some having seen continuing declines over a number of years), there were a few categories of crime in which the recorded number of incidents rose. These include rape (which rose by 5% to 55,114), indecent assault (which rose by 9% to 10,123), common assault (which rose by 5% to 267,857), drug offences (34% to 84,001), drunken driving (20% to 29,927), truck hijacking (3% to 930), cash-in-transit robberies (15% to 220) and bank robbery (7% to 58).

There is no single explanation that accounts for the increases in the recorded number of these crimes. In some cases – truck hijacking, cash-in-transit robbery and bank robbery – this is likely to reflect increased incidents rather than increased reporting, since these are well-reported crimes. It should be noted, however, that the absolute number of incidents in all three cases is pretty small and the total increase in the number for all three categories of crime amounts to only 61 additional cases.

In the case of drug crime and driving under the influence, on the other hand, the increased number of cases reflects increased police activity, since these are cases that are opened by police officers themselves. It probably reflects the increased number of police officers on our streets, as well as improved targeting of problematic locations and people.

Finally, in the case of rape and indecent assault, the increased number of recorded incidents could reflect either a change in the level of reporting or an increase in the underlying level of victimisation. Indeed, so great is the presumed under-reporting of these crimes, it is not inconceivable that an increase in reporting rates combined with a decrease in the number of actual incidents, could have led to an increase in the number of recorded cases. It is, therefore, impossible to tell from these figures what, if anything, has happened to the actual number of incidents of these crimes.

Arrest and detection rates

While there are numerous sets of figures in the SAPS Annual Report to which attention should be paid, two of the more significant relate to the number of arrests made by the SAPS, and the detection rates achieved by the Service’s detectives.

In the 12 months to April 2005, the SAPS arrested nearly 2.5 million people – more than one person in 20 in South Africa, and nearly one person in every ten over the age of 19 (Figure 6).6 It is an extraordinary figure, made more so by the fact that it is 40% higher than the figure for the number of arrests made four years earlier.

In 2004/05, 27% of all arrests were made in the Western Cape and 23% in Gauteng. Twelve percent were made in the Eastern Cape and 10% in Mpumalanga.

The enormous number of arrests made by the police reflect, in part, the large-scale operations that they conduct, and will, therefore, include a very large number of people arrested for minor crimes. At the same time, assuming that arrestees are processed properly, the SAPS’s Automated Fingerprint Identification System makes it possible to link people arrested for minor crimes to more serious cases. In addition, it also helps to build up the system’s database. This might help improve detection and conviction rates in the future.

Having said this, there are also legitimate human rights concerns that arise when a society arrests as many people as we currently do every year.

Arrests, as anyone who reads the literature on police performance assessment knows, are a very rough measure of police productivity, because the hurdle that must be cleared is pretty low. To make any sense of these numbers, then, it is necessary to look at the rate at which serious crimes go to court and the rate at which convictions are obtained. While the police provide no information on conviction rates – arguing only half-accurately that this is something over which the National Prosecuting Authority has more direct control – they do provide information on the rate at which cases are taken to court.

A couple of obvious points need to be made about the rate at which cases are referred to court. The first is that there is a pleasing rise in the proportion of cases going to court for all serious crime categories. The second is that those cases that are typically committed by strangers – robbery, car theft and burglary – are solved much less frequently than are cases of violence where the perpetrator is linked to the victim. A third, slightly less favourable point to make, is that for these eight crimes, the number of cases taken to court fell from 293,012 to 291,679 between 2003/04 and 2004/05 (see Figure 7). The lower number of recorded crimes, however, meant that this represented a higher proportion of all cases.

A less obvious point to make is that, in calculating these ratios, the SAPS have done themselves a severe disservice: especially in the case of murder, the rates at which cases are referred to court are actually better than they appear here.

The SAPS calculates the referral rate of cases to court by dividing the number of cases going to court in any year by the number of cases reported plus the number of cases on hand at the start of the year. This, they argue, reflects the fact that the cases on hand also received investigative attention during the year and should, therefore, be factored into the assessment. There are a number of problems with this approach, however:

• The first problem is that comparing a flow (the number of cases going to court) with both a stock (the number of cases on hand) and another flow (the number of cases reported) is not methodologically sound. Flows can be compared to stocks or to flows, but not to stock plus other flows, as this is not comparing like with like.
  
  

• The second problem is that the approach leads to anomalies if the rules governing the disposal of cases differ across crime types. If, for example, it is much more difficult to close murder cases as undetected than it is to close burglary cases in this manner, you find that the stock of murder cases on hand at the start of the year is a much greater proportion of cases reported in that year than is the stock of burglary cases. And this, in fact, appears to be the case. At the start of 2004/05, the SAPS had more than 32,000 murder cases on hand. This meant it had over 70% more cases on hand than were reported in 2004/05. By contrast, on the same date, the SAPS had more than 53,000 burglary cases on hand. This was 80% fewer than the number of cases that would be reported in the course of the year.
  
  

• Finally, and perhaps most significantly, we know of no other police agency that would calculate a comparable rate in this fashion. Thus the New York Police Department, which this year expects to record the fewest murders since 1961, has seen almost no change in the proportion of murders that are solved – this has hovered at 60% for decades.7 Interestingly, when we recalculate the rate at which cases are referred to court by the SAPS, using more standard methods, we find a ratio of almost exactly 60% (Figure 8).
  
  

Having recalculated the SAPS referral rates, it would be desirable to compare these to those achieved by police agencies in other jurisdictions. The trouble is that getting comparable data on these rates seems to be impossible, since the standards set for sending a docket to court and even for obtaining a conviction, differ greatly across countries. Nor does there appear to be a central repository of these data.

It is also quite possible that even if we had comparable data for some steps in the criminal justice system (such as the number of cases that go to court as a proportion of the number of cases reported), in the absence of data on subsequent stages (such as the rate at which prosecutors reject police cases), these data may resist interpretation.

Still, a referral rate of 60% does not seem unreasonably bad unless a large proportion of those cases are, in fact, rejected by prosecutors or fail for other reasons. Sadly, some admittedly dated research suggests that this may be what happens in a large number of murder investigations. By contrast, conviction rates in other matters seemed more comparable to those achieved in England and the United States (see Figure 9).8

Conclusion

Although doubts are frequently expressed about the reliability of police crime statistics, an initial look at the figures suggest that progress is being made on a broad front. Most serious crimes are down on the number recorded in 2003/04, with many having fallen for a number of years. Car theft and murder, two of the better reported and recorded crimes, are substantially down, suggesting that other downward trends may be plausible.

In addition, figures relating to arrests and the initiation of criminal justice processes suggest that steady improvements are being made in the rate at which cases are taken to court, while the number of arrests made has grown exponentially.

All of which is to say that progress is being made in the fight against crime, even if much work still remains to be done.

Acknowledgement

This article is part of the ongoing work of the ISS’s Criminal Justice Monitoring project.

Endnotes

1. P Burton, A du Plessis, T Leggett, A Louw, D Mistry and H van Vuuren, 2004, National Victims of Crime Survey: South Africa 2003, ISS Monograph 101, Institute for Security Studies, Pretoria, 2004, p 107.
   
   

2. Data are sourced from Unicri, A Alvazzi del Frate, Victims of Crime in the Developing World, 1998, Unicri, Rome, p 88.
   
   

3. Burton et al, op cit, p 107.
   
   

4. I have argued previously that there are many countries in the world in which even murder is not well reported and recorded (see A Altbeker, Puzzling statistics: Is SA really the world’s crime capital?, in SA Crime Quarterly 11, 2005). This does not appear to apply to South Africa.
   
   

5. Note that no carjacking figures were kept until 1996/97.
   
   

6. Figures for population over the age of 19 calculated from mid-year population estimate of Stats SA.
   
   

7. S Roberts, Where killers are out of style, New York Times, August 7, 2005.
   
   

8. The figures in the following graph come from T Leggett, South Africa’s Conviction Rates in Perspective, SA Crime Quarterly 5, 2003.