What we have learnt from post-1994 innovations in pro-poor service delivery in South Africa: a case study-based analysis

This paper is based partly on consultation work done under Servaas van der Berg on the effectiveness of alternative social delivery mechanisms – as commissioned by the World Bank for its World Development Report 2004.

Abstract

Service delivery is vital for alleviating poverty in South Africa. This paper contributes to the dialogue on how to maximise the impact of pro-poor service delivery by considering evidence from a wide selection of case studies to distinguish the successes and failures of post-1994 pro-poor service delivery. Case evidence brings to light four important points: that decentralisation and participation can reinforce historical distributions of privilege; that community ownership is neither a necessary nor a sufficient condition for effective service delivery to individuals in rural communities; that when managed well private outsourcing can benefit the poor; and that the abolition of user fees is often not the best way to ensure access to basic services. The paper cautions against overly ambitious and idealistic policy making. When a policy fails because of its lack of flexibility or its disregard for the constraints of the implementation context, this failure should be attributed to short-sighted policy making and not to implementation failure.

This paper is based partly on consultation work done under Servaas van der Berg on the effectiveness of alternative social delivery mechanisms – as commissioned by the World Bank for its World Development Report 2004.

1. Introduction

Service delivery in South Africa is of crucial importance particularly because of the central role it can play in poverty alleviation. In the short term, services can help relieve some of the most severe burdens of destitution, while over the long term the subsidisation of investments in health and education can help provide an exit out of persistent poverty. Poverty alleviation is clearly a high social priority in a country where it is estimated that approximately 37 per cent of households survive on less than R1000 per month (Woolard, 2002). Motivated by this urgent socio-economic imperative and evidence that service delivery is often biased against the poor, this paper focuses specifically on pro-poor service delivery.

The discussion here will distinguish between service outputs and service outcomes. The first term is used to refer merely to quantities and the second is more encompassing and asks how the service delivered has actually improved lives, thus incorporating quality dimensions. The terms ‘efficiency’ and ‘effectiveness’ are used here in relation to the inputs–outputs ratio and the inputs–outcomes ratio respectively.

The paper considers to what extent four recent trends in service delivery have succeeded in improving the efficiency and effectiveness of pro-poor service delivery since 1994. This focus is pertinent in the light of empirical evidence demonstrating that there is no necessary relationship between expenditure on services and service outcomes (e.g. Hanushek, 1989). Figure 1 shows the translation of fiscal resources into social outcomes. It demonstrates that fiscal shifts have little significance unless they can be converted to improved social outcomes.

Figure 1: A scheme for analysing the link between fiscal resource shifts and social outcomes. Source: Van der Berg & Burger (2002)

The paper opts for an economic perspective on service delivery efficiency, concentrating on the roles of monitoring, incentives and accountability in determining the behaviour of agents. It contributes to the dialogue about how to maximise the impact of pro-poor service delivery by confronting four traditional views with case study evidence. More than 100 case studies were tracked down and read in preparation for this paper – most of them from non-academic or unpublished sources that often are not accessible to the research community. Although the small sample size of case studies limits the extent to which we can generalise from the conclusions of the research, case studies remain important information sources for understanding service delivery because in most circumstances there is no other empirical feedback to improve our understanding of the process whereby allocated funds are translated into social outcomes.

This paper asks:

1. Does increased decentralisation and participation empower the poor?

2. Is community ownership required for successful rural service delivery?

3. Does outsourcing to private providers result in the underservicing and/or overcharging of the poor?

4. Do user fees exclude the poor?

2. Does increased decentralisation and participation empower the poor?

Arguments in favour of decentralisation usually claim that when the service for delivery is complex, characterised by heterogeneous demand, or when service goals are difficult to observe and measure, it is vital that service providers respond to client needs, thus making centralised service delivery inappropriate (World Bank, 2002). Because systems involving decentralisation and participation give more decision-making power to provinces, local authorities and communities, they are closely associated with many of the virtues of democracy such as allocative efficiency and the vertical separation of power.

Since 1994 there have been moves towards increasing the level of decentralisation in most service sectors – with the notable exception of police services (Pelser, 1999). In many cases the newly introduced policies have been slow to translate into the actual devolution of power, owing to ill-designed strategies and the central government's reluctance to hand over power.

To a large extent decentralisation was a symbolic and political move, and consequently decentralisation policies have often failed to take into account the characteristics and skills of specific service sectors, resulting in sub-optimal solutions. It is for instance not clear why school governing bodies had to earn their responsibility and decision-making powers while these were granted to other organisations without ensuring that they possessed the required skills and knowledge. It is also difficult to understand how community police forums can fulfil their envisaged role of functioning as an avenue for increased community input into police operations when they are added to the bottom of a highly centralised power structure.

The rest of this section considers case studies of the way community participation initiatives have affected service delivery to poor households, in particular the experiences of school governing bodies and community policing forums.

Following the political transition, the Department of Education introduced a two-tiered system for school governing bodies (SGBs), in terms of which those that are still building capacity receive less independence and fewer responsibilities. These SGBs have to earn the increased independence and additional responsibility accorded by section 21 status by demonstrating that they possess the required skills and capabilities. The Department has repeatedly stated that its goal is to help all schools progress to self-management status, so it has committed itself to supporting schools in their efforts to build the capacity required to qualify for section 21 status.

In the case of policing, Community Policing Forums (CPFs) were initially set up to help increase the legitimacy of police in black communities through informing communities about the activities of the police and providing an avenue for community input. By the late 1990s most police stations had CPFs (Shaw & Shearing, 1998).

The experiences described in the case studies caution against naïvely assuming that informally appointed or selected community representatives will legitimately represent all the concerns and interests held by the community and speak on behalf of the whole group. It has been shown that community participation initiatives are vulnerable to capture by elites or competing interest groups. Furthermore, it appears that participants who are illiterate or poor are often underrepresented in community committees. According to the case studies, poorer and less educated parents often lack the self-esteem required to play strong roles in the SGBs. Frequently participation is low in poor schools because parents do not feel they have anything to contribute. When poorer and less educated parents do become involved, they may not participate equally, either because they are hesitant to participate or because the more educated SGB members do not treat them as equals.

The Centre for Education Policy Development, Evaluation and Management (CEPD) Education 2000 Plus Project's report on case studies (2001) found persistent gender and race bias in some SGBs. SGBs are required to be representative of the school's learners, yet in many former Model C schools where most of the learners are now black, the SGBs are still dominated by white members. Some parents live far from the school and thus the timing of the meetings may partly explain the reluctance of black parents to serve on SGBs. Schools may also be to blame as it appears that they have made little effort to encourage participation from black parents. Besides this, SGBs are often still not representative of learners' gender. The report noted that in schools where respondents claimed that participation by SGB members remained unsatisfactory, the lack of participation was blamed on illiterate parents, ethnic divisions, a need for training and a lack of trust and understanding between SGB members.

In addition to these representation issues, case studies also indicate that the decentralisation of school governance has not been implemented with equal effectiveness in all communities. Despite government-initiated training schemes for participants (parents) in school governing bodies, many poor schools are having difficulties in implementing the system. The same has been observed in the case of CPFs. CPFs are generally more effective at improving crime prevention in more privileged areas and this may result in the migration of crime to less privileged areas, where individuals are generally less able to deal with its effects. In this way community participation could actually serve to entrench the social divisions that it was meant to overcome.

A case study report by McPherson (2000), comparing four schools serving different income groups, cites reasons for why a school's former ‘whites only’ status and the socio-economic background of learners attending the school are likely to coincide with competent management by the SGB. Using school fees as an indication of the average socio-economic background of the community served by the school, the author argues that the SGB of a formerly ‘whites only’ high-income Free State school (with school fees of R2000 per year) was outperforming the boards of three less privileged schools because of the superior fund raising and managerial capabilities of the parents on its board, the advantaged infrastructure and resource position of the school inherited from its former ‘whites only’ status, and its good track record which enabled the school to sustain high learner enrolments. In contrast, the report describes a dysfunctional SGB in a poor rural village school in KwaZulu-Natal (with school fees of R100 per year) where the principal was dominating decision making and there was little evidence of self-management. School policies had been formulated, but there were problems with implementation. The author blamed the SGB's lack of progress in achieving self- management on parents' hesitance and lack of knowledge. According to the findings the former may be a deeper problem than the latter. Kihato and Kabemba (2002) argue that even where SGBs claimed that they did not understand their mandate, this did not necessarily imply a malfunctioning SGB. In some cases parents were raising funds and getting things done despite not understanding their roles clearly. The same report also reasons that experience might be a more important asset than income or knowledge. The study found that SGB members in Gauteng generally had a better understanding of their roles. This was attributed partly to an intensive SGB training programme the province ran. This alone cannot, however, explain Gauteng SGB's superior organisational capacity and understanding of their roles. Gauteng SGB members who had not yet received training were also found to be more knowledgeable about their roles. The authors consider the political history of the region to be a possible explanation: in the 1980s there was a high level of Parent Teacher Student Associations in Gauteng because of the large-scale mobilisation around schooling issues in the apartheid era.

The CEPD Education 2000 Plus Project's report on case studies (2001) shows that formerly black schools (‘black schools’ here refers both to schools under the former Department of Education and Training and schools in the former homelands) are lagging behind other schools in their preparation for meeting the requirements for being granted section 21 status, and with it greater financial and managerial independence. Ten schools applied, seven former white and three former black, and of these all the former white ones were approved but only one former black. According to the report this school has been the benefactor of a range of private sector interventions and is thus not a typical formerly black school. Of the two other formerly black schools that applied, one had its application rejected and the other was still awaiting a response. Most formerly black schools had not applied for section 21 status because they were not yet ready to satisfy the requirements. In three cases schools did not apply because they had no knowledge of the section 21 provision or of the associated application process.

These observations are confirmed by a study done by Potterton and Christie (1997) which evaluated 32 schools that ‘operate well under difficult circumstances’. They found that almost all of the historically black schools had problems involving parents serving on school governing bodies.

Recent research by the National Secretariat for Safety and Security (as quoted in a report by Pelser, 1999) shows that CPFs experience problems similar to those of the SGBs. The report classified CPFs according to the focus of the forum in each of five stages of development: i) ensuring basic resources were available, ii) developing trust between police and community members, iii) improving participants' understanding of the policy, iv) raising additional resources, and v) forming partnerships with other role players against crime. The study found a high positive correlation between the CPF's stage of development and the level of privilege of the community. Most of the committees that had reached stage four (15 per cent of the total) and all of those that had reached stage five (6 per cent of the total) were situated in privileged areas. Some police employees working at stations struggling with basic resources described CPFs as a burden, and were angry about the additional demands community policing placed on them. Additionally, Altbeker and Rauch (1998) contend that:

black communities [are] typically more concerned with ameliorating socio-economic causes of crime and white communities [are] more concerned with keeping crime and criminals out of their areas. Because this pattern is also matched by very dramatic differences between levels of income, community participation in rich areas appears to focus on assisting the police in keeping crime out … it has been the consequence that the development of community-centred crime prevention programmes involving the police are much more developed in rich areas, than in poor, black areas.

The case studies indicate that decentralisation and community participation are not necessarily associated with better social outcomes for the poor. The effectiveness of service delivery is a function of both the skills and knowledge of the individuals involved. As community representatives and provincial staff in poorer areas are often less skilled and educated, service delivery in these areas is thus likely to be inferior in quality and quantity. This effect can be reduced by regulating the distribution of certain crucial responsibilities so these become conditional on a committee's ability to demonstrate the required skills and expertise.

3. Is community ownership required for successful rural service delivery?

Community participation and ownership have long been seen as crucial for the success of community development. After 1994, with the mass extension of service coverage to previously disadvantaged communities, the term ‘community ownership’ became part of the service delivery vernacular. When applied to an individual, ownership often presupposes a choice and results in a responsibility, and the way the term ‘community ownership’ is used in the literature reveals similar associations. It is described as the goal of community consultation and community-based decision-making processes, and is frequently used interchangeably with ‘community responsibility’.

In this section this concept is discussed in the context of the Department of Water Affairs and Forestry (DWAF) water services projects in rural areas. The DWAF makes a distinction between its approach to rural and urban service delivery. Service delivery to urban settlements is focused on individuals, while service delivery to rural settlements is centred on the community, involving community consultation processes and specifying community ownership as a major objective. Underlying this binary service delivery model are implicit assumptions about the cohesiveness of rural communities. Rural communities are assumed not only to be naturally more cohesive than urban settlements, but also sufficiently so to be treated as decision-making units. Specifically, it is assumed that a unit can make a binding choice and be held responsible for this choice.

In practice, implementation based on the rural service delivery model often means community representatives are used as intermediaries to manage the relationship between the service provider and the community. In these cases, the service provider transfers the responsibility for maximising community compliance (and by implication also individual compliance relating to payment for water services) to community representatives, who thus also become responsible for punishing individual non-compliance. The likelihood and expected severity of punishment will depend on the extent to which it is possible to observe individual non-compliance and the community representatives' ability and willingness to administer punishment. The individual's decision whether to comply or not will be determined by an evaluation of the expected cost and benefits of non-compliance versus the expected cost and benefits of compliance. The expected costs of non-compliance will be a function of the likelihood and expected severity of punishment for non-compliance. Note that punishment could include material penalties such as fines and the restriction of water use and non-material deprivation such as loss of reputation.

If the group's goal is sustainable, convenient access to safe water, then – within this framework – achieving this goal will mean aligning the incentives of the individual users and community representatives so that these coincide with what is optimal for the collective. Misalignment of incentives is not possible in this model if the community is a naturally cohesive unit, but this assumption displays little understanding of the complex collection of power relations and individual pursuits that exist within rural communities. Case studies show that this presumed community cohesiveness has often not been present and that project success has hinged on individual incentives and behaviour. Successful cost recovery has often been found to result from individuals associating payment with the receipt of a value-added service.

A comparison of 24 rural water project case studies by Dreyer (1998) leads the author to conclude that unwillingness to pay may be at least partly due to individuals' perceptions that the service is not adding value to their lives. The report found that in many cases the community's basic needs for water services were already met through traditional sources and that the community accordingly desired a higher level of service delivery than that offered by the national water supply programme. In contrast, villages where cost recovery was successful frequently had experienced serious problems accessing water prior to the initiation of the project. For instance, this was the case with the Vhutalu water project, described in the DWAF's account of 12 successful water projects (1998). Before the project, residents of the village had to fetch water from the rivers nearby. They were desperate for an alternative as it was impossible to walk up the steep muddy slopes from the river during the rainy season. A member of the community asked the water and sanitation NGO Mvula Trust for assistance, and consequently a project was launched through this organisation to give residents reliable water access in the form of a borehole pump. Mvula Trust asked the community for an upfront contribution of R19 000, and despite poverty – most community members were dependent on social pensions and crop sales – it managed to raise R7000 more than the required amount.

The DWAF's report concluded that the provision of a good (and appropriate) service was essential for successful cost recovery and mentioned that successful projects often had a business approach and a consumer focus. It also found that the availability of a selection of service levels promoted customer satisfaction. However, providing a value-adding service was found not to be sufficient for ensuring cost recovery. Users also needed to be made aware that the benefit and privilege of being provided with a service implied the obligation of payment in return. For instance, Dreyer (1998) mentions that in cases where taps were turned on before the community had completed the agreed-on payment for the water infrastructure investment, the community often saw no further need to pay.

In practice, the problems associated with excluding non-payers from accessing further water services have often made it difficult to forge a strong link between payment and receipt of the service. Some communities have, however, arrived at innovative solutions for addressing this issue. A Mvula Trust (2002) case study details the Nhlungwane community's system for restricting non-paying individuals' access to water. A warden is placed in control of each standpipe and keeps the keys for the standpipe. These wardens are responsible for overseeing the collection of households' water quotas at a specific time once a day, and keeping monthly records of payments by households for the provision of water services. When households need more than their allowed quota, they are required to pay an additional fee. The wardens are all women living close to the standpipes who provide the administration service voluntarily. This approach has achieved high compliance: approximately 90 per cent of households pay the monthly operation and maintenance fee of R7 regularly and early in 2002 the Village Water Committee had a positive bank balance of more than R11 000.

Successful water projects are also distinguished from others by their effective punishment mechanisms for non-paying individuals. The Vhutalu community decided to levy a flat monthly rate of R10 per household for water. When payments were not made on time, households had to pay a R30 fine. Initially there were a few late payers, but the water committee immediately fined them to show that the threat was not an idle one. The study reports that the water committee's bank account contained R26 000 the last time the balance was drawn.

The Motlhabe Ntswana-Le-Metsing water project is one featured in the DWAF report as a good example of successful cost recovery. This community decided to charge a flat rate of R15 per household per month for water. The implementation consultants proposed this tariff and then the water committee debated the proposed tariff with the community until the parties concerned arrived at an agreement. This village's payment system does not rely on the delivery of accounts. Water payments are collected from households and then marked off in a register listing all households in the village. The water committee visits any non-paying residents and judges whether a household is able to pay. If it is not, the case is investigated and the committee comes up with a proposal. Motlhabe has no unauthorised connections, possibly because of the tribal authorities' decision to fine any perpetrators R300. In the first four and a half months of operation the payment rate was 86 per cent and the report says it has increased since then. At the time the report was written, the community water account had a positive balance of R15 000.

The DWAF report also observed that successful projects gave sufficient attention to educating the customers about the need for payment, consumer rights and the consequences of non-payment. Case studies generally highlighted the pivotal role of the water committees that were elected by the community to manage water projects: water committees whose members had organisational skills and were perceived to be trustworthy were able to enforce punishment for individual non-compliance.

The analysis of these water service case studies shows that community ownership is not a sufficient condition for success. Evaluation of the reports indicates that the cost recovery success of rural water service projects is driven by the incentives and choices facing individuals. Furthermore, it appears that community ownership may frequently not be a necessary condition for success either. The case studies suggest that rural settlements very seldom possess the natural cohesion that is believed to differentiate them from urban ones. However, because the analysis in this section of the article was based solely on one service sector, i.e. water provision, it would be unwise to apply these findings to other service sectors.

4. Does outsourcing to private providers result in the underservicing and/or overcharging of the poor?

With legislation such as the Municipal Systems Act the government has acknowledged that it sees a role for private providers in service provision. Through public–private partnerships, private providers can supplement public sector capacity by offering the financial resources and management and technical skills that councils often lack (Stacey, 1997). However, efforts to partner with private providers have been met with fierce opposition, particularly in water services, where opponents of public–private partnerships have attempted to blame both water cut-offs and the recent KwaZulu-Natal cholera outbreak on the increased emphasis on cost recovery and privatisation (Bond, 2000).

Debates about the relative merits of public versus private service providers are often based on personal prejudices and abstract principles instead of on a careful comparison of the track records of these two approaches. The public versus private provider debate is frequently associated with the equality–freedom opposition, with public providers stereotypically seen as providing access for all, but suffering under a heavy burden of inefficiency; private providers, by contrast, are stereotypically linked with greater efficiency, but also with the exploitation of the poor (CASE & FAFO, 1999). Underlying the distrust of private providers is often an issue of control, a fear that private solutions will interfere with public priorities (CASE & FAFO, 1999). The government's ability to control the private provider is at the centre of all questions regarding the feasibility and desirability of private outsourcing.

In essence, this ability depends on the effectiveness of incentives and monitoring. For these to be effective, the desired outcome needs to be observable to the monitoring agent, and the monitoring agent must have the required leverage and skills to motivate the private provider to pursue this desired outcome. When either of these conditions does not hold, private outsourcing is unlikely to have the desired results.

It is useful to distinguish different types of monitoring on the basis of three possible monitoring agents: the service provider, the client and the funder. Self-monitoring is monitoring of the service provider by itself. It is presumed that if there is sufficient competition in the service sector, service providers will have an incentive to internally monitor the quality and pricing of their service. Clients can also act as monitors through signalling by their choice of service provider their satisfaction with the service received. Alternatively, if no choice is available, clients could be given a monitoring role via a service agreement assigning the service-evaluating function to clients, while the funder retains the enforcement function. Monitoring by the funder (for example the outsourcing agent) can occur both within and outside of the stipulations of a contract. The effectiveness of this kind of monitoring depends, among other things, on the agent's procurement and contract-setting skills and his or her knowledge of the service sector. Monitoring price increases is often particularly challenging.

Case studies show that where monitoring and incentives have been effective, private outsourcing has been associated with increased responsiveness and efficiency. This section focuses on private outsourcing in the primary health and water service sectors.

In the water services outsourcing case studies described below, the main monitoring agent is the outsourcing agent. In terms of the contract, the service provider is accountable to the local council rather than the client and has minimal interaction with the client. The council mediates between the provider and the client and this constrains provider responsiveness to client preferences and restricts mechanisms for monitoring by the client. Contracts are traditionally awarded for ten years or longer and there are few players in the industry, so there is little competition and hence not much incentive to self-monitor.

Despite these challenges, the case studies find little evidence to support the claim that outsourcing municipal services to a private provider is necessarily anti-poor and associated with price hikes, job losses and inferior service provision to less profitable customers. It appears that when the appropriate incentives are in place private providers will act in the interest of the poor, offering cost-effective services, expanding services to the benefit of the poor and investing in local communities. The key to ensuring this type of outcome is that the deliverables outlined in the service contract with the private provider should include pro-poor measurables such as service coverage of poor communities, community service projects and the use of local contractors and labour-intensive construction and maintenance methods.

According to a Palmer Development Group report by Timm (2000), the Queenstown Municipality's concern about its financial status prompted it to explore opportunities in public–private partnerships. In 1992, after a tendering process involving three potential providers, the municipality entered into a 25-year concession contract whereby the operations, maintenance and management of its water and sanitation systems were outsourced to Water and Sanitation South Africa (WSSA – then known as Aqua-Gold). The municipality hoped that the concession would lead to cost savings and increasing efficiency.

In 1995 the Queenstown Transitional Council was formed, amalgamating the old Queenstown municipal area and two neighbouring townships, Mlungisi and Ezibeleni. Under apartheid, services to the townships were administered by provincial and homeland authorities respectively and were of a lower standard than residents of the original Queenstown municipality were receiving. Infrastructure was deteriorating in these settlements, there were high levels of unaccounted for water, and response times to burst pipes were unacceptable. There were also capacity constraints at technical and managerial level. The council was concerned about the discrepancy between the quality of service delivery to the original municipal area of Queenstown and to Mlungisi and Ezibeleni, so it decided to extend the contract with WSSA to include the two townships. Following a public consultation process, a reformulated contract was signed with WSSA which included stipulations about the rehabilitation of infrastructure and the upgrading of service delivery in the townships. The contract provided for regular monitoring: WSSA is required to supply monthly reports to the municipality detailing the quality and quantity of water supplied.

Since the contract was signed, the cost to the council of providing water and sanitation services has dropped by 17 per cent (Moleke, 2000). Further, Timm reports that the quality of water supply has improved following the outsourcing of the operations, maintenance and management of Queenstown's water and sanitation systems. Unaccounted for water losses have decreased from 45 per cent to 21 per cent and the number of reported bursts has declined from 2 to 0.2 per year per kilometre of pipeline. Reported sewerage overflows have been reduced from 19 to 13 per year per kilometre of network. Sixty-five per cent of the townships' ageing water pipes have been replaced and water meters have been replaced and upgraded.

However, it seems that residents of Mlungisi and Ezibeleni have not noticed these improvements. Payment levels in the townships are low – according to January 2000 statistics only 56 per cent of Ezibeleni residents and 55 per cent of Mlungisi residents were paying for water services. Residents interviewed for the Timm study felt that the way the service was delivered had worsened, while the quality of water supplied remained the same. However, most of the residents' grievances were related to the council's poor customer management and tariff setting. As an example of the latter, the Timm report cited a Palmer Development Group study (1998) that found that 50 per cent of these households were spending more than 14 per cent of their household income on municipal services, excluding electricity.

Users in the townships complained of the council's high rates and perceived lack of consultation. (The municipality has remained responsible for setting tariffs because at the time of signing the contract South African legislation did not allow the outsourcing of the billing and collection function.) In 2000 Timm reported that the fixed component of township residents' utility bills (including water, sanitation, refuse removal and an infrastructure charge) added up to approximately R108. Households earning less than R1300 received a 40 per cent rebate (funded from the government's equitable share allocation) that reduced the fixed component of the bill to about R65. Before integration, Mlungisi and Ezibeleni residents paid a flat rate of R24 and R35 respectively for all services excluding electricity. According to a recent Development Bank of Southern Africa (DBSA) study (2000), the tariff adjustment was a one-off and there had not been any further increases in tariffs since the private provider took over in 1992.

In addition, the Timm report (2000) criticises the municipality's customer management and tariff setting policy and concludes that ‘with the exception of the 40 percent rebate, it could be argued that the council has not yet adopted pro-poor policies’. The report acknowledged that the council was in a difficult position since it was experiencing cash flow problems and had limited options available to it.

Apart from issues directly related to service delivery, there have been a number of reported positive spin-offs from the partnership. Firstly, the DBSA report (2000) notes that 27 permanent and 22 temporary jobs were created following the integration of Mlungisi and Ezibeleni into the original Queenstown municipality. WSSA's promotion policy favours employees from historically disadvantaged communities, and by 2000 (when this report was published) four staff members from historically disadvantaged communities had been promoted within the organisation. Secondly, by 2000 WSSA had donated R20 000 to the council to install fire hydrants at all major buildings, schools and churches in Ezibeleni, and had contributed R60 000 towards a playground for the township (DBSA, 2000). And, thirdly, the provider has a procurement policy that favours local suppliers. The Timm study mentions a long-standing relationship with a local small supplier for WSSA's pipe replacement programme. The small supplier uses labour-intensive methods and provides and manages labourers, while WSSA is responsible for providing overall supervision, equipment and materials.

The DBSA study also describes the experiences of the Stutterheim municipality after contracting out water and sanitation service provision to WSSA in 1993. In terms of the 10-year lease contract, water supply and sewerage services to the formerly predominantly white town of Stutterheim and bulk water provision and sewerage effluent treatment to the neighbouring township Mlungisi were outsourced to the firm.

According to the Plummer report (2002) the contract's lack of clarity about the distinction between maintenance and capital improvements resulted in disagreements between the council and WSSA, and capacity problems on the municipality's side prevented the municipality from acting as an effective decision-making partner and hampered its ability to use the private partner to pursue the new council's social goals. But despite these problems with the contract, the DBSA study reports that unaccounted-for water losses were reduced from 38 per cent to 24 per cent after the contract was signed and service disruptions decreased dramatically after improvements were made to the distribution network and new treatment works were introduced.

According to the DBSA report, all the municipal staff members were employed by the private provider. Four staff members from historically disadvantaged communities have received management and technical training and been promoted to senior positions. Furthermore, the private provider has a procurement policy favouring the use of local services and materials.

A case study report by Palmer et al. (2002) comparing the relative efficiency of private and public primary health clinics provides additional evidence to discredit the view that private providers will disadvantage the poor. In the case of private primary health provision there is scope for both self-monitoring and monitoring by the client: the competition between private providers and public clinics should encourage self-monitoring and the client's ability to discern quality should encourage monitoring by the client, though this is less reliable because discerning quality in health services is difficult, and particularly so for the less educated clients.

The Palmer et al. (2002) report shows that private clinics are more efficient but do not provide the full public clinic range of services. The study is an attempt to understand why – despite free provision of public primary health care – 30 per cent of individuals without health insurance choose to pay R50 to R100 per visit for private sector primary health services. The report compares the cost and quality of public and private providers. As private clinic cost estimates were based on data collected at two private clinics recommended by the chain's management as good examples of service delivery, the data may give a rosier picture of private clinics than is the case in reality. However, the study also included site visits to other private clinics, and on this basis the researchers argue that these two clinics are broadly representative of the operational model of the private clinic chain. Cost estimates for GP visits and public clinics came from two studies that were part of the same research project (see Sinanovic et al., 2001a and b).

The Palmer et al. study found that private and public clinics had comparable provider costs: for a private clinic the cost per visit was R35–44, which fell within the range of the public clinic's cost per visit of R27–68. The private clinics had high administrative costs and employed full-time doctors, yet their recurrent costs were only slightly higher than those of public clinics that did not employ private doctors (R33 vs R29). This low cost was made possible by savings in staff costs (enabled by the clinics' reliance on nurse practitioners as main service providers) and savings in drug costs (due to the clinics' strict computer-aided controls over the preparation, prescription and dispensing of drugs). Patients first saw a primary care worker (a lay health care worker), followed by a nurse, and only when necessary would the patient be referred to the doctor. Primary care workers and nurses were assisted by computers containing over 2000 treatment protocols based on the Cochrane Collaboration, which provides systematic up-to-date reviews of randomised controlled trials in all areas of health care.

According to focus group discussions conducted as part of a study by Schneider and Palmer (2002), users of private clinics were very satisfied with the service they were receiving and cited staff attitudes and waiting times as crucial differentiating factors. Users said they were treated with respect in private clinics and that the staff ‘made [them] feel important’. At private clinics the waiting times ranged between 10 and 40 minutes versus 50 minutes to 3 hours at public sector clinics. The report concludes that in rural areas private clinics are sometimes used because they are more accessible than public clinics while the greater use of private clinics in urban areas is attributed to the perceived higher quality of diagnosis, prescription and counselling (Usdin, 1993; Goldstein & Price, 1995), lower average waiting time (Usdin, 1993) and increased privacy (Beattie & Rispel, 1995; Rispel et al., 1995a, b, c).

The report also notes that private clinics did not offer a comprehensive primary care service and were concentrated in urban areas. The limited evidence available indicated that private clinics provided a superior quality of curative care, but that their collective chronic care record was weaker than that of public clinics. The technical quality of the curative care provided in private clinics appeared high judged by their Sexually Transmitted Infections (STI) treatment: 85 per cent of STI patients had been diagnosed using the syndromic approach (vs 68 per cent in the public clinic sample reviewed) and 97 per cent had received treatment recommended by the Department of Health (vs 80 per cent in the public clinic sample reviewed). Patients appeared to prefer public clinics for chronic treatment. In 30 per cent of interviewed cases, private clinic users had attended public clinics for chronic treatment in the past six months. Further, for 64 per cent of diabetic private clinic patient records reviewed and 48 per cent of the hypertension patient records reviewed, patients had visited the private clinic only once. Apart from providing relatively less chronic treatment than their public sector counterparts, private clinics also referred patients to public clinics or GPs for services relating to immunisation and TB treatment. Finally, private clinics offered no after-hours emergency services.

Palmer et al. (2002) concludes that their study of private clinics showed that low cost service delivery can be congruent with a satisfied customer base. The authors conclude that the study demonstrates the key role of management and efficiency in service delivery and argue that contracting out the management of public clinics might be an option to consider, provided that the public sector has the required capacity to manage the contracts. Rural areas with inadequate access to primary health services may benefit from such arrangements.

The case studies examined here indicate that there is little basis for claims that the profit orientation of private providers will inevitably lead these firms to underservice or overcharge the poor, or both. In cases where private providers' profit maximisation objectives can be aligned with pro-poor outcomes via incentives and effective monitoring, outsourcing service provision can be commensurate with both efficiency and affordable access to basic services for the poor.

5. Do user fees exclude the poor?

The government has recently introduced a set of ‘free basic’ programmes, including free primary health care, free basic water and free basic electricity. The implementation of plans to introduce free basic sanitation and a programme preventing the poorest schools from charging school fees were under way at the time of writing.

There are few case studies on the topic of user fees, but the information available indicates that eliminating fees increases access to services. However, there is also evidence that the abolition of user fees can result in a decrease in service quality, by stripping clients of their rights to demand a quality service in return for payment and thus removing the client–service provider feedback effect. If a service provider is dependent on the government for the bulk of its revenue, it will be tempted to concentrate its efforts on satisfying the government instead of the client (Jackson, 2002).

In this section the focus is on two case studies: the first evaluates the impact of the elimination of user fees in primary health, and the second examines the impact of school user fees. At the end of May 1994 pregnant women and children under six were exempted from paying user fees for primary health care, and two years later the programme was expanded to include the whole population.

A report by McCoy (1996) has evaluated the impact of the abolition of primary health care fees for pregnant women and children younger than six on the utilisation of such services. The report uses data on service utilisation collected from the records of hospitals and clinics at twelve sites in four provinces for the period January 1993 to July 1995. Aggregated service utilisation data was obtained from the provincial health authorities. A survey of users covered 252 individuals in four provinces. In addition, questionnaires were sent to district surgeons in rural areas of the Western Cape and Free State and to GPs who were affiliated to the South African Sentinel Practitioner Research Network.

According to the report, health service utilisation increased at most facilities following the introduction of the policy. There was a rise in the proportion of admissions of pediatric patients in all hospitals sampled, a trend which was confirmed by provincial aggregates from Mpumalanga and the North West Province. Survey responses by 40 district surgeons in the Western Cape and Free State indicated a strong increase in the district surgeon utilisation by pregnant women and children under six: in the Western Cape district surgeon utilisation by pregnant women and children under six increased by 659 per cent and 300 per cent respectively. In the Free State, visits by pregnant women and children under six rose by 51 per cent and 198 per cent respectively. Provincial averages for drug expenditure on part-time district surgeons were available for the Northern Cape, KwaZulu-Natal and Mpumalanga and showed increases of 68 per cent, 6 per cent and 18 per cent respectively. Average payments to part-time district surgeons for dispensing were 35 per cent higher in KwaZulu-Natal and rose by 4 per cent in Mpumalanga and 5 per cent in the Northern Province. Inappropriate use of clinics did not seem to be a problem. Since the introduction of the policy there had been a rise in the number and proportion of patients who required referral.

The report shows that the number of antenatal care visits has increased since the introduction of the policy. According to the National Household Survey, women have started attending antenatal care clinics earlier in their pregnancy. The most common gestational age for attending antenatal care clinics has fallen from five months to three months. Furthermore, the percentage of pregnant female respondents who reported frequent attendance at antenatal clinics increased from 79 per cent to 84 per cent.

It is difficult to gauge what impact the elimination of primary health care user fees has had on the quality of service provision. The report claims that users listed long queues, staff rudeness and supply shortages as barriers to the use of primary health care facilities. However, it is difficult to ascertain to what extent these quality concerns can be traced back to the introduction of the free basic primary health care programme. As discussed in the previous section, similar complaints were cited by studies based on surveys pre-dating the introduction of the programme. It is likely that the elimination of user fees could have decreased responsiveness to user dissatisfaction and complaints, but again it is difficult to verify the validity of this statement empirically.

However, abolishing user fees can have substantial fiscal implications and consequently there are concerns about the affordability and hence desirability of the ‘free basic’ service provision programmes (Jackson, 2002). The Department of Education has attempted to implement a hybrid solution for primary and secondary schooling whereby most users are obliged to contribute to financing the service (school governing bodies had the authority to set school fees), but with a provision for fee exemption when households can demonstrate that they cannot afford to pay the fees. The government tried to ensure universal access by stipulating that no child could be excluded from attending school on the grounds that his or her parents had failed to pay school fees. (The discussion and case studies cited here pre-date the announcement of plans to exempt a proportion of poor schools from user fees.)

It seems that the hybrid funding scheme has been reasonably successful in guaranteeing access for the poor to basic education services. Household surveys and Education Department data show that education up to age 13 has become virtually universal. According to Fiske and Ladd (2003), combining data from the Education Management Information System (EMIS) and national demographic data indicates that just below 100 per cent of children between 5 and 13 were attending school in 1995. However, according to anecdotal evidence, there have been cases where parents' inability to pay school fees has indeed resulted in exclusions. The CEPD's case studies report (2001) claims that not all parents are aware of the possibility of receiving fee exemptions. On the bright side, the report showed that there appeared to be growing awareness in this regard: 8 of the 27 schools in their 2000 sample reported that parents knew that they could apply for exemptions, and by 2001 17 of the 27 schools in the sample reported that parents were aware of this option.

Another objection to school fees is that they have contributed to the persistence of inequalities in education by creating a system where the parents' wealth and income determines the school's access to resources and therefore the quality of education provided. Fiske and Ladd (2003), for instance, argue that although the retention of school fees has helped to keep middle class families in the public school system (by allowing parents to self-fund additional resources aimed at guaranteeing the maintenance of educational standards), it has done little to improve the quality of education for disadvantaged students.

Subsidies or vouchers to the poor – as used in the housing sector – present an alternative way of giving the poor access to basic services. Subsidies or vouchers are a direct transfer of public resources to the poor that is arguably more empowering than (and on these grounds preferable to) in-kind transfers. They keep the provider–consumer relationship intact and allow the consumer to retain his or her rights and choices, thereby maintaining an avenue to influence service provision. In-kind transfers are inherently paternalistic because they assume that the individuals making the transfer have more knowledge and are thus in a better position to make choices for the recipients than they are themselves. Furthermore, receiving a voucher or subsidy instead of a service makes the recipient more aware of the cost of the service (CASE & FAFO, 1999). On the downside, vouchers and subsidies are administratively more cumbersome than in-kind transfers, which may increase the cost of the transfer.

The elimination of user fees is thus not necessarily the best way to ensure that poor households have access to basic services and it is also not a sufficient condition for ensuring access. The impact of abolishing user fees on utilisation of the service will depend on the proportion of the total usage cost that the user fee represents – with total usage cost including among other things transport costs and the user's time. Even if the service is provided to the client for free, factors such as the duration and cost of the trip to the site and waiting time at the site can still deter and restrict use.

6. Conclusion

Efficient and effective social service delivery is crucial for poverty alleviation. Fiscal redistribution has no real meaning if it does not translate into improved social outcomes.

This paper has studied the efficiency and effectiveness of four post-1994 trends in service delivery in South Africa. Based on the analysis of case studies describing these trends, the paper concludes that:

1. Decentralisation and increased participation can help to perpetuate historical distributions of privilege.

2. In water services, community ownership is not a sufficient condition for effective service delivery to individuals in rural communities. It is also not clear that it is a necessary condition in all cases.

3. Private outsourcing does not imply an anti-poor bias.

4. Abolition of user fees is not necessarily the most efficient way to ensure access to services for poor households.

There is a more general observation that can be added to these individual observations. The cross-sector analysis of service solutions highlights the extent to which the suitability of service delivery solutions is context-dependent, and the danger that lies in insisting on one-size-fits-all solutions in a country that exhibits such variation in skills, experience and income.

Capacity problems constrain the efficiency and responsiveness of service provision. Often they are collectively swept under the carpet as ‘implementation failure’, but the case studies appear to indicate that when a policy fails because of its inflexibility or its disregard for the implementation context, this failure should be attributed to short-sighted policy making and not to implementation failure. In neglecting to keep implementation constraints in mind, South African policy making often shows itself over-ambitious and too starry-eyed.

The author is grateful to Megan Louw, Stan du Plessis, Servaas van der Berg and Rulof Burger for their comments. The usual disclaimer applies.

Notes

This paper is based partly on consultation work done under Servaas van der Berg on the effectiveness of alternative social delivery mechanisms – as commissioned by the World Bank for its World Development Report 2004.