Cost and Tariff Model for Rural Water Supply Schemes

April 2003

Executive Summary

In 1981 the United Nations launched the International Drinking Water Supply and Sanitation Decade. The aim of the Decade was to bring about safe water and sanitation for all the people of the world. During the Decade, some 1600 million people were served with safe water. However by the end of the Decade it was estimated that there were still over one billion people without safe water (WHO, 1997). The continent with the highest percentage of unserved people was Africa, with 43%.

Now, twenty years after the decade began, it is estimated that, worldwide, 1.1 billion people are still without access to potable water (WHO, 2000). Some 38% of Africa's population are without access to safe water, but due to population growth the size of the backlog has actually increased. In addition a number of areas which are classified as served are known to receive water only intermittently.

The coverage figures quoted above do not tell the full story, for it is in rural areas that the percentage of unserved people is highest, and that the progress made is the lowest. In 1990 it was estimated that 56% of Africa's rural population were unserved. By 2000, despite major efforts, the figure had reduced by only 3% to 53% (WHO, 2000).

Over the last five years, a consensus has emerged on the principles to guide the provision of rural water supply. Internationally, policies call for treating water as an economic and social good managed at the lowest appropriate level. For the provision of water supply this requires that consumers be engaged in the process of selecting, financing and operating systems that meet their demands and willingness to pay.

Managing water as an economic good has strong implications for the establishment of proper financial arrangements for a project. Financial policies should send out correct signals linking service levels to actual cost, maximise cost recovery by capturing community willingness to pay, and make efficient and equitable use of subsidies.

In common with other parts of the developing world, the rural water sector in South Africa has been heavily subsidised for some time. Prior to the transition to universal franchise in 1994, government programmes in rural areas tended to focus on handpumps and windmills. These more simple water supply solutions (which due to institutional problems did not necessarily work) were thus associated with the "old" South Africa and regarded as third-class. Since 1994 the almost universal drive in the sector has been to get a piped water supply to within reach (classified as 200 metres) of every home. This is the standard practically every community aspires to, and delivery on this standard carries a very high political premium. However, it is being found that, again due to institutional problems, these reticulated schemes are, if anything, even more prone to failure than the old handpumps and windmills. The government's response to the problem of the poor sustainability of rural water supply has been to give an undertaking, on behalf of local government, that every family is entitled to 6 kl of water per month free of charge. In rural areas this undertaking is to be financed by local government primarily by using a portion of a grant (the Equitable Share) which has been made available to them for the purpose of making basic services affordable to the poor. It is expected that the more rural councils in South Africa may require at least three years to implement the free water policy, and it is realised that in some areas it may not be possible to meet the promise at the level of service implied by the policy (Department of Water Affairs and Forestry, 2001).

In this context, with international practice pushing towards increasing independence of government, down to the lowest level possible, and South African practice pushing towards increasing dependence on government, it is particularly relevant to ask the question: how much does rural water supply cost, anyway? How much does it cost national government, how much does it cost local government, and how much does it cost the body actually responsible for making the water supply work. The cost is highly dependent on many variables, such as the capital cost, the need to pump, or not, the need for ongoing support, or not, and the effectiveness of management. For example, all other things being equal, one scheme which has high transmission losses, low consumption levels and high levels of bad debt, could find that its water cost is ten or twenty times that of a scheme which has all those conditions optimised.

This report reviews a selection of the literature, local and international, discusses a number of water tariffing models which are used in the sector, and goes on to examine the legal and political framework in which water supply operates in South Africa. Thereafter a new model is described, a model which has been developed particularly for the evaluation of costs and tariffs for rural water supply schemes. The model is fairly simple to understand and use, and yet it can accommodate virtually any combination of subsidies, service levels and tariffing options. It can be used either at the planning stage to investigate the economic viability of a scheme, or as a tool during the operational phase to test different tariff scenarios.

Research Objectives

The objectives of this research were set out in the original proposal as follows:

Findings from field research
Over forty operational stand alone type rural water projects were visited and surveyed in order to obtain a picture of the kinds of operation and maintenance costs which are being experienced in the field. The data from these projects is summarised in tables and an appendix to this report.

The key observations from this field work were as follows:

i. Water Consumption

Although water consumption can in most cases only be estimated from the data gathered, the indications are that the median water consumption in rural areas is low, less than 4 kl per month. Schemes with metered yard taps recorded relatively low consumptions, but those with unmetered yard taps recorded higher consumptions. For example, the average consumption at Emayelisweni/Montebello was just 3.12 kl/hh/month, or 8.7 litres per person per day (but this is based on the assumption that each yard tap is shared by 12 people - which may be an overestimate). The only schemes which have consumption figures above 25 litres per person per day, or 6 kl/family/month, are the Fairview and Nomponjwana schemes in KwaZulu-Natal, both of which have unmetered yard connections. The estimated consumption figures for these two schemes are 16.07 and 8.14 kl/hh/month respectively.

ii. Cost per family per month

The budgeted costs per family per month are low. The mean figures in 1999 rands were R9.15, R5.49 and R3.87 for KwaZulu-Natal, the Northern Province and the Eastern Cape respectively. However the spread of figures is wide - the standard deviations of the mean in the provinces are R8.62, R6.81 and R2.80 respectively [The standard deviation gives an indication of the spread of the bulk of the data above and below the mean]. In other words the majority of projects are running on budgets of less than R15 per family per month, and some much less.

Caution needs to be exercised in drawing any conclusions from the data, as some projects were not able to declare the details of the operational subsidies which they are receiving. For example, in the Northern Province it is common for DWAF to pay the cost of pump servicing, fuel and operators wages. However, although it is common, it is also not standard on all projects, so it is hard to know how to adjust for this hidden subsidy.

iii. Cost per kilolitre

The mean budget cost per kilolitre is R4.96 in KwaZulu-Natal, with a standard deviation R3.58.

In the Northern Province the mean is R1.78, with a standard deviation R1.35. In the Eastern Cape the figure is R1.61, with a standard deviation R0.90. The higher costs in KwaZulu-Natal are indicative of the greater use of water meters or water vendors in that province.

iv. Water tariffs

Water tariffing practice varies from scheme to scheme. The flat-rate monthly payment system is the most popular cost-recovery option, employed by 31 of the 38 projects surveyed. Of the seven with water metering, six were in KwaZulu-Natal, and one was in the Eastern Cape.

The mean flat rate was R9.47/hh/month, with a standard deviation of R5.98. In other words, the typical range of monthly charges on community water projects with flat rate charges is between R4 and R16. In KwaZulu-Natal the mean flat rate charge was R14.63, with a standard deviation of R8.57, whereas in the other provinces it was lower (Northern Province mean R8.19 with std. dev. R2.32, Eastern Cape mean R5.30, std. dev. R2.68).

On the nine projects where water was being charged for at a metered rate, the mean rate was R6.33/kl, with a standard deviation of R2.36/kl.

v. Bad debt levels

The levels of bad debt are in general representative of projects which are operating. The typical levels of payment are 70%, i.e. some 30% of the people served are defaulting.

Some anomalies are evident in the tables. For example there are four schemes (KwaNyuswa, Mission and Mvunyane in KZN and Claremont in the Northern Province) which have reported receipts over the six month survey period in excess of budgeted income. There are two possible explanations: either the period included receipts from people who were paying off several months at once; or the receipts recorded do not distinguish between payments for water and connection fees. The latter is the more probable explanation.

vi. Absence of savings

The most notable thing about saving for asset replacement, is that it is not happening. This is to be expected in poor rural communities, where life is sustained on a hand to mouth basis. The expenditure figures obtained thus do not adequately reflect the long term operation and maintenance costs, and as such do not represent the real costs of operating these schemes.

However, there is concern that when a major repair cost does present, e.g .if the diesel engine needs to be reconditioned, or the electric motor is damaged by lightning, then there are no savings available to meet the cost.

vii. Schemes operating in deficit

Several of the schemes were operating in the red (the more notable examples being Dicks, Emayelisweni, Ngolokodo and Ezinqeni in KZN; and Claremont, Seokodibeng, Mankotsana and Mars in the Northern Province). These projects are still operating either because they are not paying their accounts or their staff and getting away with it (e.g. Ezinqeni), or because they receive operational assistance from DWAF.

viii. Cost effectiveness of gravity schemes

The 38 projects surveyed included three gravity schemes where no pumping was required, two in KwaZulu-Natal (Vukanathi and Zamimpilo) and one in the Eastern Cape (Tsita). A distinctive feature of these projects are their very low operating budgets. The budgets for the three projects are: Tsita R1.37/hh/month; Vukanathi R0.63/hh/month; and Zamimpilo R1.09/hh/month. The corresponding tariffs are R3/hh/month, R7 and R10. At Vukanathi it appears that no-one is paying the tariff, but as this is a gravity scheme it is at this stage still functioning.

ix. Cost effectiveness of community management

It is notable that the typical wages paid to staff are very modest. Wages of R200 to R500 per month are typical. Although the work is part-time, there is clearly a spirit of community service at work in these projects. These low staff costs are also a reflection of the cost effectiveness of community level management, compared with management from the nearest urban centre.

Also notable is the very negligible amounts being spent on transport. Most committees are not needing to spend more than R200 per month on transport, although the median expenditure is much less than that. This is again a reflection of the cost-efficiency of community level management.

Development of WRC Cost and Tariff Model for Rural Water Supply

Four existing water supply financial models are reviewed in this report. They are:

  1. the DWAF/Ninham Shand Water Supply Service Levels model;
  2. the Mvula Trust's Help Manual on Rural Water Credit;
  3. the Palmer Development Group's Water Supply Services Model; and
  4. the Raftelis model.

These models have each been developed for a specific purpose. The Mvula Trust's Help Manual for Rural Water Credit is structured to assess the advisability of using loan finance to upgrade a community water supply. It combines a needs analysis, income and affordability data and a cost model to assess whether a project will generate sufficient income to pay back a loan.

The Palmer model, on the other hand, has been developed primarily for the managers of urban water supply systems. The model enables planners to assess the viability of new infrastructure investments in the light of the needs of their changing and growing cities. This model is widely used in South Africa.

It was concluded that a cost and tariff model tailored specifically for rural water supply in South Africa should have the following features:

  1. It must be possible to run the model either at a simple level, or at a more complex level - i.e it must be possible to use the model even if accuracy is compromised, even if only the bare details regarding a scheme are known, or if the user does not have the time to fill in all the information which is known. However, the user that has the detailed information and the time to run the model at a more sophisticated level should be able to do so.
  2. The model must separate costs into a logical framework, i.e. capital costs, asset replacement costs, overhead costs, production costs, repair and maintenance costs, support and mentorship costs.
  3. No costs must be hidden. Any subsidies applicable must only be taken into account once all real costs have been determined. (One of the uses of the model is for the comparison of the economics of different options - this can obviously not be done if any costs are hidden).
  4. All data must be entered only once in the model, to avoid situations where a change in the value of a key variable at one point is not reflected at another point.
  5. It must be possible to model any simultaneous combination of Levels of Service. While a scheme may primarily supply water via community standpipes, for example, it may also include a large number of yard connections.
  6. Where possible the model layout, structure and terminology should reflect the Department of Water Affairs and Forestry's Operation and Maintenance reporting system for rural water schemes. During the last two years a fairly comprehensive O&M reporting system has been developed by DWAF and Umgeni Water specifically for rural water schemes. This reporting system uses certain terminology and cost codes, all of which have been reflected in the relevant pages of this model.
  7. The model must be able to process simultaneously various water demand scenarios. Cost per kilolitre is generally very sensitive to demand, mainly because the fixed costs (capital costs, asset replacement, rental charges and salaries) tend to be the dominant factor in the pricing of water, particularly rural water.
  8. The levels of water loss and bad debt must be explicit. Water losses and bad debt can jointly affect the cost of water by an order of magnitude. It is thus critical that these aspects of pricing are clearly dealt with at the relevant points in the model.
  9. The model must distinguish between total population and population served. It is misleading to base tariff calculations on the total population in a project area if not all these people are served by the project.
  10. Loan and grant finance must be separately specified.
  11. The model must reflect the effect of inflation on asset replacement costs.
  12. The model must allow for cost sharing between Water Services Authorities, Water Services Providers, National Government and other stakeholders.
  13. In setting tariffs, users must be able to use a combination of fixed charges and volumetric charges (with rising blocks, if required) for any level of service. The user must be able to specify different tariffs and charges for each different level of service.
  14. The model must reflect costs in terms of both volume of water supplied (R/kl) and in terms of households served (R/hh). The volume of water supplied here must exclude losses. The cost per kilolitre is important indicator of the effectiveness of a water supply scheme. The monthly cost per household served is an important indicator of affordability.

A model meeting the above criteria has been developed, and is described in Chapter 6 of this report. Provisionally this model is called the WRC Rural Water Supply Cost and Tariff Model.

Five case studies are presented in Chapter 7 - a pumped water scheme, a handpump programme, a shallow well programme, a gravity water scheme and a rainwater harvesting programme. The model shows how widely real costs vary, both within a case study according the level of consumption, and between studies. The major factors influencing external costs are the costs of asset replacement, and support and mentorship.

The model can be used to test how the Free Basic Water Policy can be implemented. It is concluded that the policy means that local government must carry practically all costs on rural water schemes (which means they will in some cases require additional assistance from national funds). The policy might therefore cause local government to rethink the suitability of more modest technologies such as handpumps and rainwater harvesting.


It is recommended that the WRC Rural Water Supply Cost and Tariff Model be demonstrated and freely distributed to practitioners in the field. If the model is found to be appropriate and useful, then it is further recommended that it is used as a standard for the calculation of costs and tariffs for rural water supply projects in South Africa, and is included with project planning reports for this purpose.

It is further recommended that the Department of Water Affairs and Forestry require all Water Services Authorities to keep up to date records of all operation and maintenance data on all rural water schemes under their jurisdiction, and that for this purpose a standard format is used to enable easy comparison and transfer of data. The standard format should provide sufficient information for completion of the DWAF O&M report for Rural Water Supply Schemes.

Finally, it is recommended that the WRC Rural Water Supply Cost and Tariff Model is updated according to the evolving needs of the model's users, and according to the observations made from the ongoing monitoring of rural water projects. Ideally, the up to date model should be available to the public from a website.

Finally, it is recommended that the WRC Rural Water Supply Cost and Tariff Model is updated according to the evolving needs of the model's users, and according to the observations made from the ongoing monitoring of rural water projects. Ideally, the up to date model should be available to the public from a website.