Pole 1: Summary of the lessons learnt from the programme, by Alain Morel à l'Huissier (Cergene), with contributions from Barthélémy Gbemade (Crepa) and Souleymane Bouaré (DNHE-Mali)

• Why carry out research on this subject?
• What can we learn from demand analysis?
• What can we learn from cost analysis?
• What can we learn from economic analysis of the private sector?
• The contribution made by the Ouagadougou workshop
• The limitations of the research carried out and possible avenues for further exploration
• Reports relating to this summary

Why carry out research on this subject?

The underlying assumption of our research and pilot action programme relates explicitly to an key economic problem. Small centres and peri-urban areas were selected as priority research areas because they have been identified as "intermediate" spaces in which the technological models of both urban and village hydraulic systems are stretched to the limits of their application: the former because of their cost (urban network type systems with individual connections are not profitable enough), and the latter because of the nature of the demand (expectations from the service outstrip what can be provided by improving traditional, shared water collection points in rural areas).

Technological solutions do not necessarily depend on technical innovation, since they are already well-known (standposts, small, simplified piping systems, autonomous water collection points with storage facilities and distribution pipes with taps, etc.) The difficulty and the innovative approaches that need to be found relate more to the way in which a system within which differentiated levels of service - meeting types of demand which themselves differ - existing alongside one another, can be planned and managed in a consistent manner.

Is economic analysis, i.e. comparing the costs and the advantages or benefits resulting from choices such as investment decisions or price-setting policies, actually necessary? To answer this question, we must explore certain side issues.

In reality, although millions of francs a year are spent on drinking water supply projects, these investments are very rarely subject to strict economic analysis. Most funders and national governments exempt them from the economic evaluations commonly required for investments made in other sectors such as transport or housing. Why?

Certainly not because they do not raise economic questions: financial resources are limited, and priorities need to be set and judgements made to reach a decision to invest in one sector rather than another, or to fund a drinking water supply project rather than some other. The question of what level of service to adopt in the context of a given project regularly arises, and whether the additional cost of choosing a higher level can be justified. Similarly, choices relating to charging are as critical in the water sector as in other public commercial services, since cost recovery is one of the manager's chief objectives, and water is a limited resource which has to be shared between different users who find themselves in competition with each other.

In fact, the reasons for the absence or superficiality of economic analyses are inextricably linked to the attitude of water sector professionals and decision-makers, and the way they present their arguments. Most of them in fact share the view – to varying degrees – that applying economic analysis to water supplies is a fruitless exercise. This opinion is based on two main arguments:

1. The first relates to an ethical conviction: many in the sector consider that access to drinking water is a fundamental right for all individuals and that supplying healthy water to all is an obligation that no economic analysis should need to justify. This position of principle underlies declarations which have been long and widely broadcast such as "water is the source of life", "one cannot put a price on water", etc.
2. The second, which is apparent more in a certain degree of scepticism than in a rigid position, is that of the professionals, who doubt that the economic benefits of an improved water supply can be sufficiently accurately assessed to justify choices made in practice. For these, the problem lies not in finding out if it would be useful to measure people's preferences for an improved service, but rather if this is feasible given project constraints.

Both these arguments, although they rest on different principles, basically point to the same real difficulty: how to measure non-perceived health benefits. By definition, the economic advantage an individual gains from the consumption of given goods or services comes down to the value he or she attributes to it. It is generally agreed, however, that the future beneficiaries of a drinking water supply are not aware, until they have experienced its advantages, of the impact of water on their health. And if they are not perceived, these benefits cannot be included in the individual economic benefits capable of being directly measured. Professionals in the sector using the first argument outlined above, and notably health specialists, take it for granted that health advantages heavily outweigh all others. Unfortunately, as previous research has shown and further research carried out within the programme has confirmed, the alternative, which consists in measuring this kind of impact indirectly, also poses serious methodological problems, so that this issue cannot be definitively resolved. Thus, given our inability to assess with any accuracy the individual or collective advantages of an improved water supply on health, we are unable to judge the relative importance of these compared to overall economic advantages.

Nevertheless, whether to convince themselves or to convince others that a given project or given water policy decisions are well-founded, most professionals in the sector clearly expect some answers on water economics.

In order to meet this operational need on the part of operators and decision-makers, the programme set itself the objective of promoting research and pilot activities capable of working towards developing tools to aid economic decision-making and management. Pursuing this objective implied directing investigations in two directions: filling a certain number of perceived gaps in our knowledge and designing operational tools integrating this improved knowledge and understanding in the form of models, methodologies and recommendations likely to help planning and management.

Gaps in our knowledge in the field of economics often result from a lack of horizontal studies using the comparative analyses of several case studies to move beyond merely amassing monographs and to systematically organise our knowledge in three areas considered to be priorities:

1. Understanding the key factors of demand for improved water supply services (a field relating to theme 1.2).
2. Understanding the mechanisms determining the costs of these services using various technologies or ways of providing them (a field relating to theme 1.2).
3. Understanding the significance of private, often informal, operators, who supply small centres and informal city areas with water (the macro-economic approach), as well as the ways in which their enterprises work (the micro-economic approach) (a field relating to theme 1.3.).

It was decided to concentrate on developing certain operational tools based on these three layers of understanding. These are respectively:

• tools for evaluating demand for improved water services (theme 1.1);
• predictive models and tools to assist the cost analysis of these services using various technologies or ways of providing services (theme 1.2);
• recommendations on what policy to adopt vis à vis private operators, the relevance of a strategy to integrate their activities and the possible ways of achieving this (theme 1.3).

The two main research activities selected within this pole of the programme (RA 3 and RA 9) pursued these two avenues.

The key feature of the water market in the particular context of "semi-urban" areas (popular areas of the cities, secondary centres) lies in the existence of competitive and complementary links between various forms of supply. Some of these are derived from "modern" supply services, often qualified as "improved" (household connections and shared water collection points); others are provided by traditional "free" water supplies, such as wells, springs, rainwater, rivers or seasonal ponds; yet others are supplied by operators (often informal) in the form of house-to-house delivery of water which they themselves have obtained in various ways.

The failure of the supply to meet the demand, however, has serious repercussions on the sustainability of built structures, on whether the population will accept responsibility for them, on the efficiency of a health policy, as well as on the financial security of water distribution companies or the delegated managers of collective water points. Operational tools, which are simple and cheap to use, would therefore be needed to assess people's willingness to pay for various levels of service, and to assess the consequences of this information on the choices of systems to be introduced, of the investments to be made and of the charges to be adopted.

The ability to attribute an economic value to environmental resources is a key issue in sustainable development in both industrial and developing countries. A lively and contradictory debate on the relative merits of two opposing approaches has been raging for the past twenty-five years (Cergrene, RA 3):

The indirect approach consists in collecting data on current observable behaviour (how much water is taken from different sources of supply, how long is taken to collect it, the costs borne) and then to infer from these, using models based on consumer demand theory, how much the latter would be prepared to pay for an improved service.

The direct approach, or "contingency evaluation method", consists in surveying a representative sample of potential users and asking them what they are prepared to pay for various types and levels of hypothetical services. In addition, the whole debate is complicated by the question of the transfer of economic benefits to the environment, which attempts to transpose by analogy one population group's evaluations in order to estimate how a second would evaluate the same resource.

To what extent are the direct and indirect methods for evaluating demand reliable tools? Which should preferably be used, and what are their respective advantages and disadvantages? What precautions should one take when using them and what credence can be given to the predictions based on their results?

These are the kind of questions the research steered by Cergrene (RA 3) addressed itself to by carrying out a "state of the art" review of these methods. This research showed that the indirect approach has so far hardly produced any models of which the field of application can be extended beyond the actual context of the site in which they were based, and that they tend to fail the test of time: behaviour patterns modelled on a town or small centre cannot be transposed to other sites, and are not even applicable in reliably predicting the choices which will prevail in the same town or the same centre a few years later.

The most promising demand models - and amongst the most robust - are those which were developed during Cergrene/Burgeap's research (RA 3) using a holistic approach. Using the results of surveys carried out in fourteen towns or secondary centres in four countries (Niger, Benin, Guinea, and Mali, i.e. over a thousand households surveyed in total), this research modelled the rates of utilisation of standposts by inhabitants not connected to a mains supply and their specific consumption at these water collection points. Using standposts for drinking water in the dry season is too widespread to allow a robust model, but the price of water being sold at standposts and the availability of wells (rare or frequent) explains to a great extent the fluctuation in the rate of utilisation of standposts for laundry water in the dry season. In the rainy season, the utilisation rate of standposts for drinking water correlates strongly to the same two variables, and to the distance to be covered: price and distance explain approximately two thirds of the fluctuation in consumption.

Amongst the direct methods, only "willingness to pay" surveys have been scientifically validated. Research conducted by the World Bank's Water Research Team showed, in one specific case, that predictions of household choice based on a willingness to pay survey could achieve great accuracy, provided, however, that certain methodological rules were rigorously followed. Until now, most willingness to pay surveys carried out have in fact been seriously flawed.

The research steered by Cergrene was based on a bibliographic review and on the recorded experience of analysts, sociologists and researchers involved in implementing these innovative techniques. It formulates detailed recommendations likely to help to achieve this good level of predictive performance, particularly by avoiding – or by knowing how to test and correct – the main sources of bias which often flaw the results of such surveys. Bias can result from the questionnaire, the survey staff, the information of those being surveyed, and it may be hypothetical, strategic and compliant. The main limitation of the indirect method rests in the fact that willingness to pay for improved services is highly sensitive to how much the potential users know about these beforehand (informational bias).

From a strictly financial point of view, the high costs of contingent evaluation surveys (between 150,000 and 200,000 FRF) could be avoided if efforts to develop behavioural models were pursued until they achieved a predictive power at least equal to direct methods. However, contingency evaluation surveys encourage a dialogue process, whereas the use of predetermined models on the contrary encourages the concentration of decisional powers by alienating the users and those locally responsible for the planning process.

Using the large amount of data collected from a wide range of case studies (essentially those produced in the context of research steered by HydroConseil: AR 9), Cergrene's research (RA 3) showed that modelling the costs of investment and operation of various water distribution systems for predictive purposes faces insurmountable hurdles.

The costs of producing and distributing water include too many parameters to meet the specifications of modelling which can be used for predictive purposes: our research shows that the various elements making up the cost of water vary widely depending on the technical choices made, that calculating depreciation or provision for renewal are closely dependent on national policy in this area, that certain parameters do not lend themselves easily to economic analysis, but at the same time they play a key role in ensuring the financial break-even of commercial agents.

Finally, our research shows that some of these parameters are difficult to evaluate accurately. Modelling using "main headings" is still, however, possible and is of major benefit for technical services or research consultants required to work in this sector, both from an economic point of view (to encourage commercial agents to improve their performance) and from an educational point of view (to raise awareness amongst contracting bodies of the consequences of their technical choices).

Based on a critical analysis of the accounting records of comparable centres or neighbourhoods in which a water distribution system has been in operation for several years, good estimates of the cost of water do exist and can provide a useful management tool for those commercially exploiting drinking water supply systems or local authorities. To this end, the research steered by Cergrene (RA 3) has developed a computerised system to assist cost analysis using a common data-base programme which can easily be adapted to any country.

Past research (notably by Morel à l'Huissier – 1990) had shown that activities related to "water redistribution" (people re-selling water in their neighbourhood, frequently found in peri-urban areas; water delivery-carriers, motorised or not; standpost managers, etc.) are not the by-product of the misfortunes befalling the growth in drinking water distribution systems. On the contrary, they reflect the form of production peculiar to the poor, urban economy system which is suited to the economic and social constraints which burden the agents concerned – both producers and consumers. They do not merely fill the gap left by the absence of a supply which should be provided by the distribution system (the "modern system"), but form a veritable dual system which – although it is closely linked in terms of structural dependence to the modern system – nevertheless has its own consistency, dynamics and rationale.

The research steered by HydroConseil (RA 9) within the programme is the first of this scale to focus systematically on studying the micro and macro-economic aspects of private enterprises active at different levels of the water supply chain, not only in the popular neighbourhoods of large cities, but also in small centres.

This research allowed the initial hypothesis according to which "some of the tasks of the commercial exploitation of drinking water supply systems is delegated to private operators under contract, but the vast majority of these operators are in the informal sector" to be both validated and criticised. It shows that pump maintenance, which is vital to the sustainability of the service, continues most often to be provided by State technical services, that there is generally a public monopoly in this area, as there is in the commercial exploitation of urban networks, but that privatisation is seen as one possible avenue here, given the increasing difficulties States face in subsidising the service. Private service operators therefore only occupy market "niches" which correspond to services that the State or municipal services cannot provide, through lack of material means, availability or staff (small repairs to pipes and engines in small centres, redistribution of water from mains supplies).

The research showed that the macro-economic weight of private operators in the urban activity of the "water" production and distribution chain is considerable. In the five centres studied, they accounted for between 21% and 84% of the total added value of the chain and, in line with the highly labour intensive nature of "informal" activities, the proportion of jobs created by these enterprises is even higher than in licensed enterprises (three to fifteen times more), even though such jobs are often insecure or temporary.

Finally, on the basis of this diagnostic study, the research formulates a number of recommendations, notably on the opportunity and the feasibility of projects tending to encourage the promotion and the involvement of private operators, as well as on ways of safeguarding or encouraging beneficial and complementary competition between various categories of operators.

Some thirty participants from various backgrounds (State hydraulic departments, consultants, research centres or universities, etc.) attended this workshop which was devoted to pole 1 of the programme. During the days of presentation and debate held at Ouagadougou from 26 to 28 May 1998, it allowed a debate to evolve around the central issues raised by the themes addressed in pole 1, in the light of the reports made by those involved in research or pilot activities.

A consensus emerged amongst the participants that the excessive size of installations is the cause of the failure of most drinking water supply projects, and that there are two main reasons for this:

• poor estimation of individual demand,
• poor estimation of the total number of inhabitants to be supplied.

The workshop recommended that everything possible should be done to make all the actors aware of this fact and for them to agree to devote the resources necessary to a careful and realistic assessment of demand (rather than of needs, a term which all too easily implies more or less external and unsuitable pre-suppositions or norms). Three methodological recommendations were formulated to help to achieve this objective:

1. carefully check and monitor demographic data and changes to them;
2. using the "state of the art" guidelines which have now been established, implement methods for measuring people's willingness to pay for an improved water supply;
3. use participatory approaches in order to ensure that demand will be taken into account at the various stages of project design, from identification to implementation.

Several case studies, notably those carried out in Kindia (Guinea, RA 1) and Yaoundé (Cameroon, PA 1), showed the pre-eminence of traditional supply systems (mainly wells and springs) in popular practices. They also showed that some of these sources do not necessarily supply water unfit for consumption, in terms of use for laundry or washing purposes, and that moreover households rate different sources of water according to their supposed quality. This rating comes fairly close to reality: thus, mains water is preferred to water from other sources, that of improved springs to that of non-improved springs and wells. The resulting hierarchy dictates the order of preference of sources for the uses to which the water extracted from these sources is put.

The quality attributed to drinking water, (water which "doesn't make you ill"), is a continuing preoccupation for the great majority of the inhabitants of urban areas. For this "higher" purpose, mains water is preferred (even if this means buying it sometimes at high prices at standposts or from neighbours with a mains connection, or going further afield to acquire it), but water from improved springs, or, in the rainy season, directly harvested rainwater, is also used for this purpose.

All the indications are that the "drinkable" quality of mains water is therefore now favourably perceived by inhabitants, who are increasingly aware of the health implications of water consumption. These results unquestionably come as a welcome surprise. The seminar participants consequently recommended that when conducting preliminary studies, there should be a systematic examination of the extent to which traditional water collection points could be integrated into the overall improved drinking water supply system, and how, (e.g. rehabilitation, boundary protection, quality control and public information), and the uses for which they might form an alternative, complementing the water distribution network, probably on a occasional basis.

Although it was decided to focus the call for submissions for research and pilot activities on fairly precise subjects, we believe the results of the programme are far from having answered all the questions raised. Indeed, it may even have raised more new questions than it has provided answers. At first sight this might seem somewhat frustrating, but it is on the contrary a hopeful sign. Several pilot activities (notably those in Yaoundé - PA 1- and in Kayes, Ségoua and Mopti – PA 5), and the research steered by ACT Consultants (RA 1) have provided case studies rich in material which has to a large extent not yet been fully exploited from the point of view of the subject areas addressed in this pole of the programme. The same is true of the many case studies resulting from RA 2 research activities (steered by Burgeap) and RA 9 (steered by HydroConseil).

Each of these case studies in itself is undoubtedly of interest, particularly thanks to the range of angles of approach and analysis chosen (technical, financial, institutional and management aspects), but including them in comparative and horizontal economic analyses would be of the greater benefit.

Research into evaluating demand shed little light on the relevance of a certain number of techniques available as alternatives to surveys, despite these being probably better suited to running small-scale projects in one (or more) urban area(s) or secondary centre(s). This is notably the case for techniques commonly used in village hydraulics projects at their information and awareness-raising stages. These should be considered for possible use within iterative methods, encouraging on the one hand less expensive and less constricting ways of expressing and measuring demand than willingness to pay surveys, and on the other taking account of this demand at various stages of the project development, from identification to implementation.

The most serious limitation, however, of the analyses carried out on demand in project mode relates not to the restricted range of techniques envisaged, but rather to the deliberately static perspective in which we have chosen to position ourselves. In a summary note prepared for the research action steered by Burgeap (RA 2), Henri Coing explains that, "... any over-anticipation of demand (the volume consumed, or by over-ambitious programmes for individual connections) leads to failure. Conversely, any static definition of demand blocks demographic and spatial dynamics...". He therefore recommends that the systems envisaged should be capable of change and that the ways in which these changes could occur should be anticipated from the outset.

This is in fact a well known principle of economic rationality. Let us remember that economic efficiency means notably supplying the service at the lowest possible cost thanks to optimal investment planning. Such economic optimisation, however, requires that the size of the system equipment should be calculated with a view to minimising the total, up-dated costs of the additional investments necessary to increase its capacity when demand increases. The demand estimation at the time the system is put into service will therefore always be too low. It is in fact changing demand over the whole life-time of the equipment installed which needs to be estimated in order to determine its size. Depending on each case (economies of scale vary according to the nature of the equipment), this might be done in the light of demand at the end of this life-time, or in several stages, by anticipating for each of these a short-term sizing policy. This principle, however, is very commonly set aside by projects, particularly when they involve or cover areas where socio-demographic and spatial change can be rapid and unpredictable, which is often the case for secondary centres and above all peri-urban areas.

The problem is far from simple. Models including dates and optimal expansion sizes have certainly been developed for the various kinds of equipment (pumping, treatment, storage, distribution) needed for drinking water supply systems. As they stand at present, however, these are unable to take into account the often major uncertainty surrounding changes in the very economic parameters which are fed into them (since specific consumption changes with living standards, and with demographic growth, but also with unit costs and frequency of up-dating).

"Social" policies, based on various financial incentive mechanisms, have failed to allow low-income population groups to benefit to any significant or sustainable extent from household service levels. For the past twenty years or so, so-called "social" water policies have relied on three instruments, massively adopted by distribution companies: an incremental tariff structure, an initial subsidised tariff band, and connections subsidised or paid for on loans. These have, however, proved ineffective in sub-Saharan Africa, or, more seriously, have had the opposite effect to the objective in mind.

A first question therefore arises and was debated in the Ouagadougou seminar workshop on pole 1 of the programme: How far can one go in lowering the initial cost of connection to the user?

Taking as their starting point the principle that the initial cost of a connection acts as a major deterrent (since it frequently represents several months' income for over half the population), many countries have chosen to pass only a fraction of this cost on to the user, or to allow the user to pay it off on credit. In all cases, the low savings capacity of households in informal or evolving housing areas prevents most of them from being able to benefit from such measures.

In a city such as Abidjan, for example, over 60% of householders state that they cannot save. Connection policies based on these principles are therefore more akin to a policy of faster catching up with demand – as yet unmet but capable of paying - than to a genuine social policy. Moreover, while a large proportion of homes remain unconnected, the equity of such policies is questionable, since the mechanism for subsidising connections by the State or the communes in question lead inevitably to their being partly financed by the very households which are unable to benefit from them. Finally, the example of Côte d'Ivoire, which since the mid-70s has successfully followed a policy of subsidised connections, clearly shows that resolving the problem of the initial connection cost is not enough, since when faced with paying the bill for the water consumed, these new customers find it so difficult to regularly scrape together the amounts they owe, that the service is frequently suspended or disconnected.

The second question the workshop considered was as follows: Are incremental rates and a "social" band rate equitable?

When tariffs fail to cover the operating costs of drinking water supply systems, subsidies have to be provided by the State or sometimes by municipalities to keep the service going. These subsidies are, however, proportional to the water consumed. As better-off families consume the largest quantities of water, such subsidies in fact benefit the richer sections of society at the expense of the poorer. The need for a certain amount of redistribution of income through tariff structures has not however escaped distribution companies since most of them have chosen to use an incremental tariff structure.

Whatever the variations within the tariff structure, the first band, generally known as the "social" band, is supposedly designed to allow the richest households to subsidise the consumption of the poorest. The principle rests on the generally observed correlation between levels of consumption and income. Unfortunately, here again, indirect effects lead to the reverse distribution of income from the poor to the rich, contrary to the objective in mind. Research undertaken has shown for example that in densely populated areas several households living around a single courtyard frequently share a common tap and split the bill between them. Incremental tariffs therefore mean that these households pay more for their water than a better-off family which has its own connection. Finally, several case studies documented during the programme (e.g. PA 5) have described situations in which several poor families buy water from a neighbour who has a connection, increasing the consumption of that connection and consequently the unit cost of the water: the seller passes these prices on to the buyers who are then subsidising the consumption of better-off customers who can benefit from the social band.

Given these observations, the workshop participants pointed out with justification that it is now a matter of urgency to question the logic of the principles of the tariff structures commonly in use in Africa, in order to re-establish an apparently forgotten compromise between efficiency and equity in drinking water tariffs.

Reports relating to this summary

ADELINE et al., 1998. Rapport final de l'AP 1.

COLLIGNON et al., 1998. Rapport final de l'AR 9.

ETIENNE J. et al., 1998. Rapport final de l’AR 2.

MOREL à L'HUISSIER A. et al., 1998. Rapport final de l'AR 3.

MOREL à L'HUISSIER A. et VERDEIL V., 1996. Gestion des bornes-fontaines : étude comparative et évaluation de projets réalisés ou en cours de réalisation (villes de Kayes, Ségou, Mopti). published by Cergrene. 200 p.

ROMANN D. et al., 1998. Rapport final de l'AR 1.