Michigan State University Press
  • Performance measures for the South African Police Service:setting the benchmarks for service delivery

Ted Leggett looks at international experience and argues that there is a need to return to a set of performance measures that align with what the police actually do, and makes several suggestions for indicators that do not create perverse incentives.

South Africa's remarkably smooth transition to democracy was made possible in part by the retention of the civil service of the prior regime. Surprisingly, given the ideological nature of the apartheid government, this retention has not resulted in overt acts of sabotage. But the new government quickly came to realise that the implementation of progressive legislation was reliant on the performance of government employees whose professional behaviour was ingrained during service to a state with a completely different ethos and agenda.

This conflict was perhaps most evident in the South African Police, the agency that formed the front line of conflict with the revolutionaries who later became the executive and legislature. In response, a programme of aggressive organisational transformation was undertaken which involved everything from quota-driven affirmative action to the de-militarisation of the rank structure (South African Police Service Act, 1995; White Paper on Safety and Security, 1998).

Unfortunately, in the South African Police Service (SAPS), as in other departments, the transformation of the civil service has not been able to keep pace with the policymakers. Sweeping legislation has been stalled, sometimes for several years, by the problems of practical implementation.1 The public, which long ago reached the limits of its ability to defer gratification, has placed considerable pressure on its elected representatives to deliver the goods, and this has prompted a series of actions designed to compel civil servants to conform with the popular will. [End Page 55]

One of the first concrete manifestations of this drive was the White Paper on Transforming Public Service Delivery, known as the 'Batho Pele' (People First) White Paper. This document laid out a set of principles for all public servants to follow, and was aimed at converting the civil service from a bureaucracy hinged on secrecy and regimented hierarchies to a responsive and accountable service organisation. Among the 'Eight Principles of Batho Pele' are:

  • • the right of citizens to have access to information on the services to which they are entitled;

  • • the right to openness and transparency on all issues, including how public money is being spent; and,

  • • the right to 'value for money' (Batho Pele White Paper, Section 3).

In order to ensure the principles are put into practice, the paper declares:

National and provincial departments must publish standards for the level and quality of services they will provide... Standards must be precise and measurable, so that users can judge for themselves whether or not they are receiving what was promised (Batho Pele White Paper, Section 4.2.1).

The White Paper was given legislative force with the passage of the Public Finance Management Act (Act 1 of 1999). This law builds reporting on 'performance against predetermined objectives' into the financial accounting process (Public Finance Management Act, 1999, Section 40 (3) (a)). The onus has thus been laid on all government departments to itemise the services they provide to the public, set goals for service delivery, and monitor progress towards these goals. While not explicit on the point, the implication is that performance measures should be quantifiable in such a way that they can be compared to expenditure, thus allowing objective determination of whether value for money is being realised.

What quickly became clear after all this top-level activity was that most departments had neither the mechanisms nor the expertise to be able to establish and measure performance benchmarks. According to an August 2000 Public Service Commission Survey: 'It was found that there is a general lack of practical skills in the public service to apply Batho Pele principles' (2000:148). Thus, the drive to get the state bureaucracy to perform by requiring it to account for service delivered has itself been stymied by the lack of capacity within the public service to measure its own performance. [End Page 56]

Of all government agencies, the Department of Safety and Security, under which the SAPS falls, has been under the most intense pressure to perform and most desperately in need of democratic legitimisation. Not surprisingly, the Department was among the first to attempt performance measurement and goal setting. The Police Services Act of 1995 required the National Commissioner to develop a policing plan, including setting priorities and objectives. In response, a set of 'Police Priorities and Objectives' (PPOs) were documented in a series of small blue books that were distributed to all police officials with their pay slips. Performance monitoring against these standards was assigned to a civilian Secretariat.

It is generally agreed, however, that this exercise was largely unsuccessful, for a variety of reasons. According to Shaw (2000), in order to avoid public criticism and to include all the diverse components of the SAPS in the process, far too many priorities were identified. Over 100 indicators were tracked at station level at one point, and evaluation of this mass of information did not result in concrete consequences for police managers or their staffs. Data quality was poor and subjective, and target setting did not relate to any independent set of criteria about optimal performance. Furthermore, the involvement of the civilian Secretariat itself may have led to the impression that the process was essentially an external audit, while, as Weatheritt argues, performance measurement, 'can only be achieved if officers themselves come to believe it is important and necessary' (1993:21). The end result was an exercise on paper, which had little effect on police operations (Shaw 2000).

One of the problems encountered in evaluating the PPOs was that individual station areas in South Africa have access to vastly different resources, a maldistribution rooted in the geographic segregation of the apartheid era. One attempt at performance measurement that took these factors into account was 'Project Lifeline', a cooperative venture between the 100 'most needy' stations and McKinsey Incorporated, an international consultancy. According to one of the project leaders, the effort was highly successful in the stations where it was piloted, but faltered in the roll-out phase after the consultancy withdrew.2

This early drive for police accountability has been derailed by a recent shift in police policy orientation, away from a focus on transforming the 'force' into a 'service' and towards effective crime combating. This shift corresponded with the change in executive management with the 1999 elections. The former Minister (Sidney Mufumadi) and his National [End Page 57] Commissioner (George Fivaz) spent a great deal of their energy setting policy and trying to integrate 11 separate police agencies into a single national unit. This necessarily diverted attention from the mechanics of operational policing, during a period of time when crime rates were reported to be rising dramatically.

In response to public outcry about a perceived rise in crime, a new Minister (Steve Tshwete) and Commissioner (Jackie Selebi) were appointed in the 1999 cabinet reshuffle. These men were marketed as far more aggressive in their approach to tackling crime than their predecessors, and this is reflected in their management style and operational focus. As Minister Tshwete himself said after taking office:

Most of the time last year [1998] was wasted on trying to formulate an appropriate crime strategy in the country. Research and whatnot. All these things have been done. Now, implementation is the name of the game.

Under the new administration, the civilian Secretariat, which, as noted above, was responsible for monitoring police functioning, was downscaled and sidelined, while attention was diverted to high-density operations using both police and military personnel. Unfortunately, the new emphasis on combating crime has meant that recorded crime levels have become the most important indicator of police performance, and the problems with this approach are discussed below.

South Africa's failure to come to grips with the problem of police performance measurement is not entirely the fault of the South African Police Service, as this topic is a contentious area internationally. This paper will first survey some of the key issues in this debate before looking at the possibilities for innovation in South Africa. It will conclude with suggestions for a set of indicators that should be available from existing systems, as well as some that could be monitored with the inclusion of additional paperwork.

The nature of police performance

The structures and duties of the police vary greatly between countries. South Africa is notable for having a highly centralised national service, which is only now being supplemented by local municipal agencies. The role of this service has varied over time, but is presently very broadly defined. The aim of the Department of Safety and Security is 'to prevent, combat, and investigate crime, maintain public order, protect and secure [End Page 58] the inhabitants of South Africa and their property, and uphold and enforce the law' (Department of Finance 2002:533). In practice, this has meant the SAPS absorbing many of the responsibilities generally assigned to more specialised units in other countries, such as:

  • • maintaining order in court rooms;

  • • guarding public buildings;

  • • transporting prisoners to and from court appearances and securing them throughout;

  • • protecting dignitaries and public officials;

  • • acting as commissioners of oaths in certifying documents;

  • • processing requests for licenses to own firearms, operate drinking establishments, and provide public transportation;

  • • providing border control;

  • • serving subpoenas and providing witness liaison during trial;

  • • serving court orders in other matters; and recently,

  • • providing a range of services to victims of domestic abuse.

Measuring what the police do anywhere is notoriously difficult. As Marx points out, 'the factors on which officers are rated may have little to do with what police actually do on patrol' (1976:56). Of all members of the civil service, their duties are perhaps the most varied and difficult to track. They also perform these duties in a range of settings, and are not confined to office premises like many other public service employees. In particular, the function of the uniformed 'patrol' member is so diverse that no single indicator could possibly summarise all its outputs.

The diversity in the challenges faced by individual stations, due to assignment of duties, police priorities, and differences in the natures of the communities served, led the RAND Corporation to conclude, 'it is a mistake to compare police departments or to seek ways of explaining real differences in police productivity' (Petersilia et al 1987:115).

One response to this complexity is to reduce all policing functioning to a single bottom line: the police are responsible for fighting crime, and thus the best single indicator of their performance is concluded to be the crime rate itself. This line of thinking is in keeping with an international trend in police philosophy, which emphasises proactive (crime prevention) policing rather than reactive (law enforcement) policing. Of course, it is very difficult to measure the crimes that never were, so the effectiveness of crime prevention programmes tends to be measured by fluctuations in crime rates. [End Page 59]

Using recorded crime as a performance indicator has some very powerful proponents, perhaps most notably those involved in the 'miracle' of crime reduction in New York City. New York's 'Compstat' (from 'compare statistics') system holds station commanders personally responsible for all crime in their area. Complex computer analysis of recorded crime data allows station commanders to identify patterns of criminal activity, and the Compstat system requires that plans be made to address trends. Failure to produce results has direct consequences down the chain of command (Silverman 1999). This system, along with a 'zero tolerance' approach to so called 'lifestyle crimes' (such as vandalism, vagrancy, and littering) is credited with a nearly 70 per cent reduction in murder in the city in five years.

While South Africa could learn much from the culture of informed accountability implicit in Compstat, using recorded crime as a performance indicator would be highly problematic for a number of reasons.

Firstly, it is wrong to measure the outputs of an agency by benchmarks that are outside the direct control of that agency (Carter et al 1992, also Steering Committee 2000, Mathie 1989). While the police can control their own organisation and efficient use of resources, they cannot control the many variables that go into an increase or decrease in the crime rate. Using the crime rate to judge police performance in South Africa would divert attention from those departments that are assigned to deal with the underlying social correlates of crime, and, in the final analysis, the extent to which the state can control crime at all is debatable. As Bayley points out:

The police do not prevent crime....repeated analysis has consistently failed to find any connection between the number of police officers and crime rates. ...the primary strategies adopted by modern police have been shown to have little or no effect on crime....when the number of police is suddenly reduced in communities...crime rates are... unaffected.

Even the causes of New York's crime reduction have been hotly contested. The massive crime reduction in New York occurred at the time of an unprecedented growth streak in the US economy and during the decline of a drug for which up to two thirds of all arrestees had tested positive during the previous decade - crack cocaine (Golub and Johnson 1997).

Secondly, the bulk of police functioning internationally is reactive and administrative (see, for example, Dadds and Scheide 2000; Bayley 1994, and the Criminal Justice Commission 1996), so measuring reductions in [End Page 60] reported crime misses out on most of what the police actually do. This is particularly true in South Africa, where the police are assigned to such a wide range of functions that, as will be discussed below, only a fraction of all SAPS personnel hours are spent on crime-fighting activities. Focusing on recorded crime may prompt the police to give short shrift to other areas of service that the public expects from the police, and may lead to such aggressive policing that human rights are threatened. This was the experience in New York, and a new report on the London Metropolitan Police found that a focus on performance measures that included recorded crime reduction had had a detrimental effect on public perceptions of the police (FitzGerald and Hough 2002).

Finally, and most importantly, good police performance could conceivably impact the crime rate in a variety of ways, and could actually lead to an increase in the number of crimes recorded, the basis on which crime rates are drawn. This is obvious in areas like drug offences, where the number of crimes recorded is almost exclusively reliant on affirmative police work, but it is also true for a range of other types of crime that may be picked up on patrol and which would otherwise go unreported.

The way that reporting rates can affect crime rates is illustrated in the following counterintuitive international examples from the United Nations Development Programme (1999) statistics:

  • • Canada has the second highest rate of recorded rape in the world (267 per 100,000), second only to Estonia in the UNDP statistics;

  • • the rate of drug crimes in Switzerland (574 per 100,000) is more than 10 times that in Columbia (40 per 100,000);

  • • the rate of total crimes in Denmark (10,508 per 100,000) is more than five times that in the Russian Federation (1779 per 100,000) and more than 100 times that of Indonesia (80 per 100,000).

Use of recorded crime as a performance measure in the current South African context would be especially problematic. Research in this country has shown that upwards of 50 per cent of crime in many important categories goes unreported (Louw 1998), as Figure 1 demonstrates. Historic distrust between the police and the public has led many communities to fail to report crime, and as this situation improves (partly due to successful police outreach and performance), this will affect the crime rate. This is particularly true for interpersonal crimes (such as domestic violence and rape), where growing consciousness of human rights teamed with a more victim-friendly legal and procedural framework should enhance reporting. [End Page 61]

Figure 1. Reporting rates of crime according to victimisation surveys
Click for larger view
View full resolution
Figure 1.

Reporting rates of crime according to victimisation surveys

In fact, a good deal of the increase of crime post-1994 may be due to the progressive enfranchisement of the majority of the population, including increased access to commodities that are known to increase reporting, such as vehicles, telephones and property insurance. In South Africa after 1994, rates for crimes that both internationally and locally show the highest levels of reporting (murder, business burglary, and auto theft) have actually been in decline, while commonly underreported crimes have been seen to increase (Figure 2). It is unusual for murder rates to decline while other forms of violent crime are increasing, and this trend is particularly striking because the percentage of murders committed with a firearm has increased (Hennop 1999). Thus, the number of non-firearm assaults that resulted in death has decreased dramatically, at the same time that non-lethal assaults have allegedly increased. A similar paradox is seen with the divergent trends in residential (up by a third since 1994) and business (stable) burglary. This incongruity suggests that much of South Africa's post-democracy crime wave is simply the recording of crimes that had been hidden in the past. [End Page 62]

Figure 2. Percentage change in total recorded crime, 1994-2001 Source: CIAC
Click for larger view
View full resolution
Figure 2.

Percentage change in total recorded crime, 1994-2001 Source: CIAC

Good police work in South Africa means finding this hidden crime, and, as police succeed in fulfilling this function, crime rates will necessarily go up. The operational policing strategy adopted under Jackie Selebi is largely based on high density operations in high crime areas (Department of Safety and Security 2001), including roadblocks and raids, which we can expect will turn up much unreported crime, such as possession of contraband, firearms offences, and violations of immigration laws. This effect is enhanced when legislation (such as the recent Domestic Violence Act) requires the recording of incidents as crime that might have otherwise been left undocumented.

This last example calls to mind the ways that using recorded crime as the primary performance indicator create a perverse incentive for the police to hide the real crime situation. If the police are told to show a reduction in the crime rate, it is easy to simply avoid recording a range of crimes they encounter, and domestic violence provides a good example of the way that using the crime rate as a performance measure can backfire. In many [End Page 63] domestic violence situations, the complainant can easily be discouraged from filing a complaint, and international studies have shown that most domestic violence calls are resolved without a case being opened (Bayley 1994:20). This is particularly true in South Africa (in contrast to New York), where consciousness of civil rights and gender equality are only now developing. The easiest way for the police to reduce the crime rate is simply to do nothing but record only those crimes where a case number is absolutely mandatory, such as cases involving deaths or insured property.

For these reasons, use of crime rates to evaluate police performance is inappropriate in South Africa, but the above discussion gives some insight into the criteria to be applied in coming up with valid indicators. To gauge how well the police are doing, indicators should be selected that:

  • • measure what the police actually do;

  • • are in the control of the police; and

  • • create incentives for the kind of police service desired by the country.

As will be discussed in the following sections, this is easier said than done.

What do the police do?

Police forces are traditionally divided into two core areas of operational activity: uniformed street work and investigations. The functions of these two roles are so different that independent sets of indicators are required for each. In addition, the members assigned to these functions are managed by supervisors and supported by administrators and other specialists. While these members must have their performance individually evaluated as well, their collective output is to support the core functions, and ideally their numbers should be kept as few as possible, as they represent resources that could have been devoted to the outputs expected of the police. Consequently, this article does not discuss the appropriate measures to be applied to supervisory or administrative functions. The SAPS also has units assigned full time to guarding and public order duties, which do not involve traditional street policing. If these activities are indeed to be maintained, they also need performance measures and the SAPS needs to be credited for their outputs, but this article will focus on the two areas that are indisputably the province of the police.

As was mentioned above, the uniformed patrol function is multifaceted. Street cops are required to respond to calls for assistance, whether initiated from a central dispatch system or from direct public requests. These contacts often require a range of skills, and are more often resolved through [End Page 64] mediation than through arrest. The uniformed member is required to make the determination of whether a crime has occurred and whether an arrest is possible and appropriate. They are responsible for generating accurate public records and may be called upon to testify in court.

Uniformed members also patrol the streets both to detect crime problems and to provide a visible presence that is believed to have a crime deterrent effect. They are often called upon to provide support to other state actors, such as medical emergency services and fire fighters. They also deal with traffic problems and accidents, whether or not criminal activity is involved, and maintain order at public gatherings, both formal and impromptu.

Measuring all this activity provides something of a challenge, and assigning relative value to individual tasks is nearly impossible. Does providing life saving CPR count for more or less than arresting a murder suspect? Does this evaluation change based on the verdict of the suspect's trial? How does one quantify the crime prevention effect of visible patrol? How much time should a 'good' police member spend in court? These questions can be confounding, and will be discussed in more detail below.

In contrast, the investigative function is less problematic. Investigative members are generally referred to as detectives. They follow up on cases once a complaint is entered. They can also initiate arrests as their investigations develop. Their primary task is to prepare cases for trial, but this involves a number of responsibilities. They must locate and liaise with victims and witnesses, enlisting their cooperation in the investigation and the trial. They must secure and document physical evidence. They may be called up to effect arrests once they have built a case against a suspect. They must brief the prosecution and provide testimony in court. In South Africa, detectives are also responsible for subpoenaing and producing witnesses even after the case has gone to court, as well as for their own testimony. In the end, however, each complaint assigned to a detective must be filed in one of four distinct categories:

  • • referred to court (accepted by the prosecutor);

  • • unfounded (affirmatively determined to be unfit for court);

  • • undetected (not presented to prosecutor due to the need for further investigation);

  • • withdrawn (not presented to the prosecutor because of affirmative impediments to building a prosecutable case, such as refusal of the complainant to cooperate).

This is stipulated in the SAPS (1995) Standing Order 392. [End Page 65] The percentage of cases referred to court combined with the percentage of cases affirmatively determined to be unfounded is termed the 'clearance rate' and represents fairly comprehensively the level of success of the detectives. Once a case is accepted by the prosecution as presenting a reasonable prospect of conviction and the accused is formally charged, the police lose their discretion to withdraw the case, essentially moving responsibility to the prosecution. In practice, this transfer of responsibility is not so clear cut, as detectives are still required for the purposes of testimony and witness liaison, but the willingness of the prosecution to accept the case indicates that the police have effectively completed their core mission - identifying and building a case against a suspect.

Thus, the clearance rate satisfies the criteria laid out above for a valid performance indicator, and it is used as such in police jurisdictions across the world. While it summarises a range of activities, it focuses on what the detectives actually do. The prosecution will not accept cases that have not been adequately investigated because their failure to secure a conviction after acceptance is seen as a negative performance indicator on their part. The prosecution then becomes the final judge of whether the police have performed their duties adequately, and a set of checks and balances is established.

Use of clearance rates has been criticised by Burrows (1989) because variations in rates between jurisdictions are more attributable to 'secondary clearances' - those clearances that come as a result of asking offenders about past cases - than real detective performance. Burrows explains that some areas emphasise the practice of questioning arrestees about past activity, while some don't. He suggests that a clear distinction be kept between 'primary' and 'secondary' clearances for accounting purposes (Burrows 1989).

But can the police really have more control over their clearance rate than they have over the crime rate? What do the police really control?

What do the police control?

While the police cannot be said to control the amount of crime reported, they can control their own internal management, including the distribution and use of resources. If resources are being spent on activities for which the police are not credited, either these resources need to be reallocated to areas the public deems more important or performance measures need to be expanded to account for all that is actually being produced. If these [End Page 66] resources are limited or the demand on these resources is great, performance expectations will have to be amended accordingly and flexibility may be limited. While all public servants complain about a lack of resources, how well endowed is the SAPS?

In South Africa, objective measures suggest that resources should be sufficient for the task at hand. SAPS total personnel number in excess of 125,000, or about 312 members for every 100,000 people in the population.3 Since the role of the police varies between countries, doing international comparisons is difficult, but, according to Newman (1999), this is more than are found in the US (300), Australia (275), or Canada (249). South Africa has one police member for every 320 people. A ratio of one per every 400 is considered good, while one per every 600 is considered bad (Bayley 1994).

This does not account for the high crime levels experienced in South Africa, however. Approximately 2.5 million crimes were reported in this country in 2000 (Crime Information Analysis Centre 2001). The uniformed (not detective or civilian) police force is about 80,000 members. This means that each uniformed member was responsible for handling an average of 32 crimes for the year, or about one every eight working days.

This does not sound like much to the layperson, and comparing workloads across time reinforces the impression that the police are not presently challenged for numbers. In 1978, a total force of 35,019 members all told handled just over a million offences and an equal number of 'infringements of the law' (mainly violations of apartheid segregation laws). According to the Commissioner at the time and subsequent observers, a disproportionate amount of police time was spent on enforcing the race laws, including a good deal of public order work and even counter-terrorist enforcement abroad. Discounting this fact and focusing only on the enforcement of comparable laws to those in force today, contemporary police members handle at least one third less crime per member than did their counterparts in the 1970s (Prinsloo 1997).

Furthermore, the police do not seem unduly burdened by international standards. There were approximately 4.6 violent crimes (defined for comparison as murder, rape, robbery, and aggravated assault) per member in 2000. This is less than Canada (4.7) but more than the US (2.8)(according to Bayley 1994:37). On its face, current crime-to-cop ratios do not seem to constitute an untenable situation, if all of these members were assigned to tasks directly involving crime. [End Page 67]

That this is not the case is demonstrated by the much less favourable ratio of detectives to crimes. Since most detectives are involved in handing recorded crime, their workload will more fairly represent the burden placed on frontline police staff. About 18 per cent of total police personnel are detectives, which is more than in Britain or the United States (15 per cent), but less than Japan (20 per cent) (Bayley 1994:25). These 22,000 detectives are responsible for processing this same input of 2.5 million crimes, or about one new crime every other working day. The ratio of crimes to detective varies quite a bit between crime types, with detectives specialising in murders generally having a much lighter caseload than those dealing with petty thefts, for example. Once factors such as time spent on service of summons and court time are added in, this caseload could represent quite a challenging situation, even given the fact that a good share of incoming cases are given cursory treatment when evidence is limited. Only about 17 per cent of sworn police members are detectives, however, and this share is within the control of the SAPS. The fact that these frontline staff may be challenged does not mean that total available human resources could not be better deployed to deal with handling crime. Indeed, the need for more detectives has been recognised by the SAPS, with Detective Services being the fastest growing portion of the SAPS budget (Department of Finance 2002).

The SAPS also controls about 27,000 vehicles, or more than one vehicle for every four operational staff (uniformed and detective).{4} However, these members are never all on duty at the same time. With regard to street staff, most work four-day blocks of 12-hour shifts alternating with four days off, plus holidays. Thus, less than half the street staff will be working on any given day, and only a quarter during any given shift, allowing each member (in theory) to have his own vehicle.

Thus, there should be enough members and vehicles to deal with the crime in South Africa, if these resources were focused on dealing with crime. Unfortunately, as was discussed above, most personnel-hours are spent on other sorts of activities. If combating crime (rather than, say, guarding public buildings) is seen as the primary purpose of the police, there is clearly a need for a reallocation of resources.

The fact that the raw numbers of members do not correspond to real street strength can be attributed in part to 'rank inflation'. In an effort to bolster police numbers and re-deploy members of the apartheid state, former homeland, and liberation forces, the SAPS has been subject to [End Page 68] waves of integration of large bodies of members with little or no real experience and limited training in police work. A desire for parity, a quota-based affirmative action policy, and union pressures have caused these members to be regularly promoted, mainly because pay increments are tied to rank. As a result, there are presently more than five times as many Inspectors (the senior non-commissioned rank) than Constables (the entry level rank) (Figure 3). Constables, intended to be the primary street level operatives, comprise less than seven per cent of total SAPS personnel.

The situation will be aggravated in the near future if the military proceeds with plans to transfer to the police over 7000 senior members who are no longer 'combat ready' due to age. These members will surely expect rank in their new designation, adding to an excess of management staff (Koopman 2002).

Figure 3. Total SAPS staff by rank, December 2001 [End Page 69]
Click for larger view
View full resolution
Figure 3.

Total SAPS staff by rank, December 2001 [End Page 69]

This development, teamed with a systematic de-militarisation of the police since the coming of democracy, has resulted in an effective collapse of the rank structure. The rank of Inspector is supposed to represent the senior field supervisory rank, the equivalent of 'Lieutenant' in many American departments (Figure 4). At present, however, Inspectors are being placed in charge of whole teams of other Inspectors, some of which have greater seniority in terms of years of service than their commanders. Without a sense of a clear chain of command, field accountability has been lost, highlighting the need for individual performance indicators. Once a member becomes a 'commissioned officer' with the rank of Captain, he is effectively removed from field duty, and the SAPS has almost as many Captains as it does Constables.

Figure 4. Organogram of hierarchy in the SAPS
Click for larger view
View full resolution
Figure 4.

Organogram of hierarchy in the SAPS

Despite periodic large-scale infusions of personnel, total SAPS staffing has been in decline in recent years. Natural attrition is about 5000 members per year.5 While there are plans to counteract this trend with a massive recruitment drive (16,000 members in three years) (Department of Finance 2002), even this effort will only result in 1997 staffing levels being regained in 2005 (Figure 5). [End Page 70]

Figure 5. Total SAPS staffing levels
Click for larger view
View full resolution
Figure 5.

Total SAPS staffing levels

Thus, the SAPS faces some serious challenges in redeploying resources to optimise performance. The Service is burdened with a range of responsibilities extraneous to its core functions and not tallied in the expectations of the South African public. There is an excess of management-level staff, many of whom cannot be deployed to street duties. Total personnel levels are in decline, and present training capabilities make countering this trend difficult.

When it comes to setting benchmarks for the South African police, it is therefore essential that the workload they are facing be taken into consideration. For example, the level of workload detectives face can substantially impact their ability to perform. If a detective has only half the time to devote to each case as his counterpart, he cannot be expected to achieve the same clearance rate.

On a national level, the SAPS detectives send to court about one out of every four cases (24 per cent) that come across their desks. This is better than the US (22 per cent), but worse than Britain (35 per cent) (Bayley 1994:27), although recent figures suggest that Britain's rate is comparable to South Africa's (The Economist, December 22, 2001:40). But referral rates vary quite a bit across crime categories (Figure 6) and between provinces (Figure 7). Some crime types demand significantly more detective attention per case than do others, while some provinces and stations must [End Page 71] process many more case per detective than do those situated in better-resourced areas. These factors need to be taken into consideration before clearance rates can be compared.

Figure 6. Percent of cases recorded referred to court by crime Source: CIAC
Click for larger view
View full resolution
Figure 6.

Percent of cases recorded referred to court by crime Source: CIAC

This is easily done by calculating the number of cases referred to court per detective, although this performance measure is far from perfect. It posits some sort of linear relationship between workload and referral rate, while surely there exists some point at which workload becomes overwhelming, and no point exists at which referrals will be 100 per cent. But its practical application to the South African situation produces some interesting results. [End Page 72]

Figure 7. Percent of cases recorded referred to court by province Source: CIAC
Click for larger view
View full resolution
Figure 7.

Percent of cases recorded referred to court by province Source: CIAC

Figure 8 plots crimes referred per detective for each of the nine provinces, as well as the national average.

Figure 8. Annual number of cases referred per detective (2000) Source: CIAC [End Page 73]
Click for larger view
View full resolution
Figure 8.

Annual number of cases referred per detective (2000) Source: CIAC [End Page 73]

The standard deviation is seven, all but three of the provinces are within one standard deviation, and only one lies outside two standard deviations - the Western Cape, which advances twice as many cases to court as the national average.6 Removing this outlier results in a standard deviation of five, with two provinces remaining outside one standard deviation. There thus appears to be enough consistency between provinces to suggest a reasonable performance standard could be set, with enough variation to suggest that some provinces are better applying their resources than others.

The number of cases referred to court per detective seems to satisfy our criteria for a valid performance indicator, and should be tracked on any level where sufficient discretion as to use of resources exists to hold decision makers accountable, including at individual (making allowance for crime category variations), station, sub-national, and national levels.

Coming up with an index for uniformed police is far more complicated, especially given than many uniformed members in South Africa are assigned to highly specialised duties rather than general patrol.

Uniformed performance measures

Unlike detectives, whose duties are fairly uniform between jurisdictions, probing the performance indicators for uniformed police quickly leads to a discussion of policing philosophy. Because their duties are so varied, choosing what to credit uniformed police for is tantamount to stating what their function is in society. That is partly why in countries like South Africa, where the purpose of the police is still being contested and refined, setting of standards is so difficult to do.

One of the best places to start is to look at what the police are already doing, evaluate whether this matches societal priorities, and setting standards to encourage changes in emphasis. Several countries produce detailed time use studies of the police (through periodic surveys) and some require daily accounting by all members of what was done during their working hours (though daily activity logs). Several states in Australia do periodic studies (Dadds and Scheide 2000), while many departments in the United States and New Zealand require daily reports. These reports facilitate both value-for-money accounting and assist internal decision making (Steering Committee 2000:13).

The Home Office of the United Kingdom conducted one such study recently, the results of which are summarised in Figure 9. This study found that members spent 17 per cent of their time on patrol, 23 per cent of their [End Page 74] time attending incidents, and nine per cent of their time on other crime related duties outside the station. Depending on the shift, between about a third and half their time was spent inside the station house, doing paperwork (about 41 per cent of in-station time), processing prisoners, and performing related duties (PA Consulting Group 2001).

Figure 9. Uniformed time use in UK
Click for larger view
View full resolution
Figure 9.

Uniformed time use in UK

Bayley (1994) cites British studies that found a different profile: 22 per cent of time spent patrolling and 12 per cent responding to incidents, with half the time in the station house and half that time on administration. He contrasts this to Australia (40 per cent incidents, 30 per cent patrol, 30 per cent administration).

Preliminary studies in South Africa have indicated a starkly different time use profile. This is largely due to the fact that only a fraction of station level staff are engaged in comparable street duties. These are generally members assigned to one of two divisions: 'crime prevention' (CP) and the 'client service centre' (CSC). In fact, generally only one unit of the CSC is assigned to street policing, the so-called 'outside' or responding unit. CP [End Page 75] is, as the name suggests, dedicated to crime prevention activities, with the responding CSC unit reacts to calls for assistance. This bifurcation of the patrol function seems odd to outsiders, since in most other countries uniformed members engage in both types of activities.

In one case study of a major metropolitan station, approximately 40 per cent of station staff was assigned to CP. This comprises by far the largest component of uniformed patrol, and represents a share of total staff somewhat less than is found in the US (65 per cent), Canada (64 per cent), Britain (56 per cent), and Australia (54 per cent), but about the same as Japan (40 per cent) (Bayley 1994:16). These members spent their time engaged in a series of group 'operations' targeting areas identified as crime 'hotspots' based on analysis of crime statistics. These operations generally involved roadblocks, 'cordon and search' activities, raids, and 'visible patrol'. While these members were required by regulations to respond to priority calls for assistance, in the case study few attended to radio calls. This was left to the CSC staff, who spent most of their time going from call to call, with brief intervals of random patrol. The station also had units dedicated to 'sector policing', focusing on community liaison with individual members dedicated to specified geographic areas within the station boundaries. Nearly a fifth of the staff was assigned to full time guard duty, as the station had responsibility for security at the local court houses as well as its own extensive holding cells. The remainder of the CSC staff was dedicated to acting as commissioners of oaths and otherwise maintaining the charge desk, while dedicated detectives took all criminal complaints.

This extreme specialisation of function means that time use must be examined at the station level, as individual profiles are likely to vary radically. The station level outputs observed in the case study, including detectives and support staff, are summarised in Figure 10. [End Page 76]

Figure 10. Case study station outputs
Click for larger view
View full resolution
Figure 10.

Case study station outputs

What this table makes clear is that only a small percentage of total station personnel hours are dedicated to the prevention of crime and field enforcement. Aside from administration, guarding is the single function that absorbed most of the time resources of the station, despite the fact that the SAPS is rarely credited with the number of hours of custodial and premise safety it provides. Assuming that, as the proclamations of the political leadership suggest, the combating of crime is the primary focus of the Service, then a radical rework of member deployment appears to be necessary.

Since internationally many police departments are required to account to the taxpayers for the services they render, looking at annual reports and other delivery-related research is one way to glean the kind of performance measures valued in other jurisdictions. In the United Kingdom, the Inspectorate of Constabulary has conducted detailed national studies of value for money in the police (Her Majesty's Inspectorate of Constabulary 1998), and the numerous local authorities produce strategic plans with quantified benchmarks (Avon and Somerset Police Authority 2000). In Australia, similar research has been done on state level (Carrington et al [End Page 77] 1996). In the United States, thousands of municipal police departments and county sheriffs set targets and track indicators (Bobb 2000).

The benchmarks set in many of these cases do involve recorded crime, but are tempered with a range of other performance, service, and organisational indicators, including:

  • • detection rates (similar to referral rates);

  • • response time, including emergency response;

  • • arrests, especially in discretionary areas like drug offences;

  • • seizures of contraband;

  • • public satisfaction (based on surveys);

  • • use of force indicators; and

  • • substantiated complaints (negative indicator).

The remainder of this article discusses a range of indicators that could be adopted in the South African context.

Response time

Internationally, the police are called upon whenever the public needs immediate response from the state, whenever there is a need for a quick restoration of order. One UK study found that 75 per cent of the calls received were for non-criminal matters (Her Majesty's Inspectorate of Constabulary 1998). Whether criminal in nature or not, one of the primary things that the public expects from the police is personal attention in a reasonable period of time in the case of an emergency. As such, response time comprises one of the most important indicators of police performance.

This figure is generally derived automatically from computer-aided dispatch systems, and capacity exists in South Africa for this information to be derived from 10111 call centres. Of course, unlike many other jurisdictions, not all South African calls for assistance are logged at a central dispatch system, so this information would be incomplete in most areas, and unavailable in station areas that do not have access to the 10111 system. Nonetheless, changes in protocol could make this information more complete in the future, and even a limited sample of cases could provide the basis for benchmarking in the future. Most areas that do set targets for response time make allowances for jurisdiction type and resources available. Most important is setting reasonable goals for improvement and working on strategies to enhance timely response to citizen requests. [End Page 78]

Arrests and seizures

Arrests and seizures, like many other crime related indicators, are problematic because they presume a set level of criminal activity against which targets can be set. Since the police should be working to reduce crime levels in addition to providing effective law enforcement, some conflict of motives is inherent. Seizing large amounts of imported drugs or guns may indicate good detective work, but it also indicates poor border control and a prevailing climate in which these commodities are in demand.

Further, unless the types of arrests are specified by crime category, 'padding' of figures with arrests for minor infractions is inevitable. If arrests are not tracked through the criminal justice system to determine whether they are sound enough to provide a basis for a conviction, an incentive is created to open a criminal case in matters where other forms of action may be more advisable. As Bayley (1994) and many others point out, very few of the 'criminal' situations the police encounter merit an arrest, and most are resolved through other forms of mediation, to the benefit of all concerned. Marx states, 'The question "How many arrests or tickets" is asked rather than "Was it wise to write a ticket, or make and arrest in this case?"' (1976:77).

The current criminal records system of the SAPS does not generate arrest figures, but this should be rectified once the integrated justice system IT network is established. At present, both arrests and seizures have been tracked manually as positive indicators in large-scale, high-density operations, such as 'Operation Crackdown'. As might be expected, the arrests made in Crackdown have been criticised by human rights groups, especially for their targeting of undocumented migrants, and allegations of abuses (including the destruction of legitimate visa papers) are common. While arrest figures itemised by crime category can be used as part of an overall evaluation strategy, particularly in areas requiring affirmative police action (like drug and weapons offences), they should not be allowed to become the primary measure of police performance.

With regard to investigations, the ratio of cases in which detectives make arrests to those in which a suspect was identified by the police can be used as a supplementary measure to referral rates. Detectives should be given extra credit for those cases in which they identify and arrest a suspect. [End Page 79]

Public satisfaction

The ultimate gauge of public service performance in a democracy is public satisfaction. This is especially true in the present South African context, in which the police are ostensibly trying to transform from an agency of repressive social control to a 'community oriented' service. Gathering this information requires the hiring of independent consultants to conduct a survey, but the returns on this investment are a bottom-line statement on successful service delivery. This information is particularly useful if collected longitudinally, so that innovations in policy can be tested.

One problem with general surveys is that popular prejudice against the police, which may be informed by media reports about activities outside the subject jurisdiction, can make progress in this area difficult. Davis (2000) found that while people's feelings about the police are affected by direct experience, they depend even more on culturally transmitted norms and beliefs. This is particularly true in a country like South Africa, where victimisation surveys have shown stark contrasts in perceptions between ethnic groups and communities.

One way to avoid this problem is to conduct exit polls or follow-up calls of those who have just received service, so that opinions are informed by direct experience. Both general opinion and that of service recipients were measured in a survey conducted by the Institute for Security Studies (ISS), and it was found that those who had had recent experience with the police had far better opinions of them than the general public (Pelser and Louw 2001). But in the end, the police will have to contend with the fact that their diverse constituencies have contradictory expectations, needs, and responses to interventions (Carter et al 1992).

It is also useful to use these polls to test matters of fact, rather than just opinion, by checking, for instance, that people having filed a complaint have a case number and detective contact. This allows an independent evaluation of real service provision not tainted by the vagaries of public perception.

The importance of ongoing public opinion polls as a complement to police generated numerical targets was underscored in a new report on the London Metropolitan Police Service (MPS). This report, described as 'the most in-depth analysis of policing in London for 20 years' (BBC News Service April 11, 2002) concluded, 'the MPS needs to be able to develop ways of managing performance that place greater emphasis of achieving professional standards and less emphasis on hitting numerical targets' [End Page 80] (FitzGerald and Hough 2002). As was the case in New York, public opinion appears to be cyclical, with demands for crime reduction being followed by a reaction against aggressive policing. While South Africa is surely on the pro-enforcement end of the spectrum at the present time, the SAPS must keep in mind that public opinion is dynamic, that public reception to policies must be continually revaluated, and regular polls must be conducted to measure the real impact of other indicators.

Use of force and complaints

Any system that emphasises indicators of crime combating exclusively will soon run afoul of the constitution, New York City being a case in point. It is therefore necessary that indicators that track police misconduct be included so that zeal for the law is leavened with respect for human rights. Rather than relying on the conscience of the members, the tension between optimal performance and respectful restraint should be enshrined in clear quantitative terms. Marx (1976) suggests (among other use of force indicators) keeping track of the ratio of total arrests to the number or arrests for resisting or disorderly conduct brought by an individual or a station as a whole. He points out that the number of police killings per member varies greatly between US cities, independently of the crime rate or perceived level of violent crime, suggesting that use of deadly force is more a matter of local policy than actual risk (Marx 1976).

There is always a problem with relying on an agency to collect its own negative performance indicators, so it is imperative that another agency be charged with this duty. In fact, the system probably works best when one agency's negative performance indicator is another's productivity measure. In South Africa, the Independent Complaints Directorate is ideally situated to play this role, but resource constraints mean that only a limited portion of complaints can be handled, particularly those involving death in police custody. Further endowing this agency with jurisdiction and commensurate resources could greatly enhance the performance measurement process.

Internal and other measures

In addition to service delivery measures, the station should be held accountable for internal management. One of the most potent indicators in this respect is absenteeism, which is rife in the SAPS. While failure to manage internal processes properly is likely to result in poor external performance, these indicators are helpful in getting to the root of the problem. [End Page 81]

While none of these indicators is perfect, it is important that all efforts be made to ensure that poor performance by one division does not impact negatively on other divisions. This may require the inclusion of additional procedures and paperwork to ensure that credit is given where due. For example, while the prosecution is ultimately responsible for the case once it is accepted to court, poor detective performance, either in witness liaison or testimony, can sink an otherwise sound case. In order to ensure that detectives continue to have an interest after credit has been given for their primary performance, it may become necessary for a complaints procedure be put in place for prosecutors to protect their own statistics. Similarly, detectives should be empowered to refer back to uniformed members any case that does not have sufficient or accurate baseline information on which a case can be built, and that rejected cases not be tallied against their referral rate.

Finally, if South Africa is serious about implementing community or sector policing after decades of an adversarial relationship to the public, some measure needs to be taken of the extent of public contact. One way of keeping track of field contacts is the 'field interview' card - an index card which lists the identity details of members of the uniformed member's primary beat (Adams 2001:332). In addition to encouraging public contact, these cards can become a source of intelligence about members of the community, both law abiding and not. The total number of such contacts documented can be used by both the station and individual members as a performance indicator.


Measuring performance is about more than apportioning credit and blame. The goals towards which an organisation strives essentially dictate its priorities. It is therefore vital that these objectives be formulated to encourage the behaviour desired from the organisation.

The police are a multipurpose agency whose behaviour is difficult to track and control. Sound performance measures are all the more important when a branch of government is given the kind of discretion and authority given to the police. Indicators must be selected that both allow accountability and promote socially beneficial activity.

In the South African context, the recent drive towards crime combating poses a substantial risk to individual liberties, and fails to address the underlying structural weaknesses present in the Service. Emphasis on [End Page 82] reducing the levels of crime recorded is both disempowering to a police service struggling with its operational priorities and produces perverse incentives, given that much crime remains unreported. While past efforts to track performance collapsed beneath a byzantine morass of paperwork, a few simple changes to present reporting procedures could greatly enhance both internal and public understanding of police performance.

Ted Leggett

Ted Leggett is Associate Research Fellow at the School of Development Studies, University of Natal, Durban, and Senior Researcher at the Institute for Security Studies, Pretoria.


1. A case in point in the Amendment to Section 49 of the Criminal Procedure Act, a piece of legislation that was passed in 1998 and has yet to be promulgated in the second quarter of 2002.

2. Interview with Noah Maltz, consultant for McKinsey Incorporated, March 25, 2002.

3. Personal fax, February 11, 2002, Divisional Commissioner: Logistics.

4. Personal fax, February 11, 2002, Divisional Commissioner: Logistics.

5. Personal communication between Martin Schönteich of the Institute for Security Studies and Senior Superintendent Johann Nelson, SAPS finance division, February 8, 2002.

6. This may be due in part to variations in crime profile. Violent crimes generally have higher referral rates than do property crimes, and the Western Cape has a high rate of violent crime.


Adams, T (2001) Police Field Operations. Upper Saddle River: Prentice Hall.
Alvazzi del Frate, A (1998) Victims of Crime in the Developing World. Rome: United Nations Interregional Crime and Justice Research Institute.
Avon and Somerset Police Authority (2000) Best Value Performance Plan.
Barolsky, V (1999) 'Victims and the Police: The National Victims of Crime Survey', Crime and Conflict 16.
Bayley, D (1994) Police for the Future. New York: Oxford.
BBC News Service (2002) 'Crime targets "hampering" police'. April 11, 2002.
Bobb, M (2000) Los Angeles County Sherriff's Department: 13th Semiannual Report.
Burrows, J (1989) 'Investigating burglary: the measurement of police peformance', in D Kenney (ed) Police and Policing: Contemporary Issues. New York: Praeger.
Camerer, L, A Louw, M Shaw, L Artz, W Scharf (1998) Crime in Cape Town: results of a city victim survey. ISS Monograph Series No 23. Pretoria: Institute for Security Studies. [End Page 83]
Carrington, R, N Puthucheary, D Rose and S Yaisawarng (1996) Performance Measurement in Government Service Provision: the case of police services in New South Wales. Report of the NSW Treasury.
Carter, N, R Klein, and P Day (1992) How organisations measure success: The use of performance indicators in government. London: Routledge.
Crime Information Analysis Centre (2001) Crime in the RSA per Police Area, 1994-2001. Pretoria: Crime Information Analysis Centre.
Criminal Justice Commission (1996) The Nature of General Police Work. Research Paper Series 3(2). Brisbane: Kingswood Press.
Cullinan, K (1999) 'Action Man', Siyaya Issue 5.
Dadds, V and T Scheide (2000) Police Performance and Activity Measurement. Australian Institute of Criminology Trends and Issues Series, Paper 180.
Davis, R (2000) The Use of Citizen Surveys as a Tool for Police Reform. New York: Vera Institute of Justice.
Department of Public Service and Administration (1997) White Paper on Transforming Public Service Delivery.
Department of Safety and Security (1998) White Paper on Safety and Security.
Department of Safety and Security (2001) Annual Report. Pretoria: Government Printer.
Department of Finance (2002) Estimates of National Expenditure: Vote 24 - Safety and Security.
FitzGerald, M and M Hough (2002)Policing for London: Key findings. http://www.policingforlondon.org/reports.shtml
Golub, A and B Johnson (1997) Crack's Decline: some surprise findings from US cities. National Institute of Justice Research Brief. Washington DC: Department of Justice.
Hennop, E (1999) 'Firearm related crime: lead up to new legislation', Nedcor/ISS Crime Index 3(3).
Her Majesty's Inspectorate of Constabulary (1998) What Price Policing? A Study of Efficiency and Value for Money in the Police Service.
Koopman, A (2002) 'Huge soldier boost for the police force', The Star, Tuesday, March 5.
Louw, A (1998) Crime in Pretoria: results of a victim survey. Pretoria: Institute for Security Studies and Institute for Democracy in South Africa.
——— (1999) 'Comparing crime in South Africa's cities', African Security Review 8(1).
Marx, G (1976) 'Alternative measures of police performance', in E Viano (ed) Criminal Justice Research. Lanham, Maryland: Lexington Books. [End Page 84]
Mathie, R (1989) 'Examining performance: some suggestions on measuring performances for evaluating police work', Policing 5(5).
Newman, G (ed) (1999) United Nations Office for Drug Control and Crime Prevention: global report on crime and justice. Oxford: Oxford University Press.
PA Consulting Group (2001) Diary of a Police Officer. Police Research Series Paper 149. London: Home Office.
Parliament of South Africa (1995) The South African Police Services Act.
Parliament of South Africa (1999) Public Finance Management Act. Section 40(3)(a).
Pelser, E and A Louw (2001) 'SAPS Service Delivery Survey Project Report'. Unpublished report, Institute for Security Studies.
Petersilia, J, A Abrahamse and J Wilson (1987) Police performance and case attrition. New York: The Rand Corporation.
Prinsloo, G (1977) Annual Report of the Commissioner of the South African Police for the period 1 July 1976 to 30 June 1977. Pretoria: Government Printers.
Public Service Commission of South Africa (2000) Survey of Compliance with Batho Pele Policy:148.
Reiner, R and S Spencer (eds) (1993) Accountable Policing: effectiveness, empowerment, and equity. Newark: Institute for Public Policy Research.
Robertshaw, R, A Louw, M Shaw, M Mashiyane and S Brettell (2001) Reducing Crime in Durban: a victim survey and a safer city strategy. ISS Monograph Series No 58. Pretoria: Institute for Security Studies.
SAPS (1995) Standing Order 392.
Shaw, M (2000) 'That #@$%^ Little Blue Book: priorities and police performance', Crime and Conflict 19.
Silverman, E (1999) NYPD Battles Crime: innovative strategies in policing. Boston: Northeastern University Press.
Steering Committee for the Review of Commonwealth/State Service Provision (2000) Linking Inputs and Outputs: activity measurement by the Police Services.
The Economist (2001) 'Bruising for a fight'. December 22.
United Nations Development Programme (1999) Human Development Report.
Weatheritt, M (1993) 'Measuring police performance: Accounting or accountability?', in R Reiner and S Spencer (eds). [End Page 85]

Additional Information

Print ISSN
Launched on MUSE
Open Access
Back To Top

This website uses cookies to ensure you get the best experience on our website. Without cookies your experience may not be seamless.