PART I: AN HISTORICAL PERSPECTIVE

I. THE CAUSES OF THE POOR PERFORMANCE OF IRRIGATION PROJECTS: AN UNFINISHED DEBATE

"Irrigation systems in many parts of the world are known to be performing well below their potential." (ICID past president)

Most analytical reports on the irrigation and drainage sector start with a laudatory statement on the contribution of irrigation and drainage to world food security during the last three decades and an observation on the declining growth of irrigated lands worldwide.^[1] These are followed by a discussion on the projected contribution of irrigation to meeting the food and fibre needs of the world population by 2025. Next is the observation that the overall performance of irrigation and drainage investments has too often fallen short of the expectations of planners, governments and financing institutions alike (FAO). The consensus between irrigation analysts ends at this point. Most recent reports differ on the causes of the poor performance of irrigation projects. The focus may reflect the main interest or, in some cases, the bias or ideology of the individual author or of the agency. This report moves straightforward to an historical review of the causes of the poor performance of the irrigation systems that have been identified in many reports since the great expansion of irrigation in developing countries in the 1960s and 1970s.

This report first presents the reasons of the poor performance that are rooted in the perceived weaknesses of design and management concepts of irrigation projects. It then presents the administrative and behavioural reasons through a review of the critics made by some analysts against other groups, such as design engineers and financial institutions.

Perceived deficiencies in technical design and management

A forty-year-old report of the former Office National des Irrigations in Morocco noted that the question of how to make the most rational use of large investments in the construction of dams and large irrigation projects was a matter of great concern worldwide: "It became evident during the 1950s that even with the installation of an expensive irrigation infrastructure, water use was below the expected level. This is attributed to the need for a programme of land consolidation in conjunction with the irrigation investment." At that time, irrigation infrastructure built by the governments consisted only of primary and secondary canals, with a few farm outlets. Farmers were then expected to bring water to their own plots by building tertiary canals and ditches. This type of investment without on-farm development was the model for irrigation development in a number of countries in the 1960s. While this model promoted the rapid growth of irrigation, it was obviously inadequate in the countries with smallholders, who could not organize themselves for the financing and implementation of on-farm works and adopt modern irrigation water delivery. Farmers were obliged to continue with century-old methods of cultivation and irrigation fostering mediocre crop yields. This cause of poor performance was partly corrected by the systematic construction of the tertiary system by the irrigation agencies with, in some cases, the financial or labour participation of the farmers. This practice is now well accepted by national governments and donor agencies. A few countries, such as Morocco and Thailand, decided in the 1960s and 1970s to proceed with a consolidation programme of the irrigable lands before undertaking the construction of the tertiary system and on-farm works. Some form of land consolidation is essential for optimal use of water in projects where excessive land fragmentation prevails.

Extension of the construction of the irrigation systems down to the tertiary systems was not enough, however, to push the performance of irrigation systems to their expected level. Since the most apparent problems of water management are wastage of water below the farm outlets, the common response in the 1970s was to promote on-farm development, including introduction of modern water application methods and precise land levelling. Another response at that time was to promote the creation of water user groups at the level of tertiary canals. External assistance supported these approaches in countries such as Pakistan, Egypt and the Philippines. This response addressed only part of the problem, since no efforts were made to improve management at the higher level. A third technical response to the disappointing performance of schemes consisted in refining measuring techniques. Since it was widely accepted that water measurement is essential to effective water management, many donor-supported projects finance the installation of measuring devices at each branching point of irrigation systems. However, water measurement in irrigation systems should be adapted to the actual field conditions to be effective. The conditions in irrigation systems are very different from flow measurements in hydrology and hydraulics research. Repetitive use of conventional measuring devices requires training and dedication of operators and does not prevent malfunctioning of control structures. These and other developments in the technical hardware contributed little, however, to solving the problems encountered in irrigation schemes. In the 1970s certain leading professionals started to pay attention to what they described as the software of irrigation systems.

Conventional engineering solutions failed to solve the problem of irrigation performance and this progressively led to a new way of thinking in the 1980s which is still strongly entrenched among the irrigation community. Widespread wisdom has it that the poor performance of irrigation systems is due predominantly to management. A keynote speaker at an ICID congress in 1992 said: "There is now a wide recognition that deficiencies in management and related institutional problems, rather than the technology of irrigation, were the chief constraints of poor performance of irrigation systems."^[2] This statement was cited and hardened by the author of an article on the problems of irrigation in developing countries, which observed: "In the developing countries, the heads of the agencies concerned are usually engineers, but they often lack the knowledge of critically important non-technical factors such as the social structure of the farmers to be benefited, economic constraints at local and national levels, and environmental issues" (Kirpich). A discussant of that article went even further by stating that the technical solutions to the irrigation and drainage projects were trivial when compared to political, institutional and cultural problems. He therefore recommended that degrees for professional engineering include courses in anthropology, business development and economics. The support given to the creation of the International Irrigation Management Institute in 1984 was based on the emerging consensus among irrigation professionals that most solutions were to be found in the field of management. The focus of IIMI has continuously been on management, and irrigation technology received a very low level of attention since the elevation of IIMI activities to those of IWMI.

Admittedly, there are important management-related and institutional deficiencies in irrigation, such as conflicts between farmers and irrigation agencies, poor cost recovery of investments and recurrent costs, lack of coordination between agriculture and irrigation agencies, and lack of farmer participation in design and management. However, the advocates of the key role of management in irrigation performance have yet to fully explore the technical deficiencies in the design of irrigation projects. These technical deficiencies will be discussed in Part II of this document.

Administrative and behavioural reasons

Once the technical solutions to the poor performance of irrigation projects were apparently exhausted, some analysts started looking at the professional competence and capability of other experts to achieve high irrigation performance. Engineers were the first obvious target of these critics. However, irrigation agencies and donor organizations were also strongly criticized.

Criticism of engineers

Social scientists have generally been in the frontline of criticism against engineers. Diemer (1996) states that irrigation engineers know little about the actual principles of water distribution in schemes in developing countries. They often assume that, in their schemes, which are mostly gravity-irrigated, there is no better way of distributing water than according to the rules they had in mind when they designed and built the irrigation systems. The design procedure focuses on crop and construction issues with the aim of reducing expenditure on construction, management and maintenance. The engineers usually base their design solely on physical data. The only social components likely to be considered are demographic information (such as labour force and land tenure) and the potential economic and financial yield of the scheme and plots. Empirical data on the diversity in dynamics of farms, group of irrigators, organizational patterns and local political structures are rarely available.

The revised FAO guidelines on the preparation of irrigation projects support these views: "Building ownership and commitment through participation has often been difficult to achieve in the past. The conventional sequence of identification/preparation carried out against tight deadlines by external planning teams has seldom allowed time for genuine participation either by government staff or farmers. On implementation, government engineers, for their part, have usually seen irrigation only from an engineering rather than a farming or social perspective. They have been reluctant to adopt participatory approaches with farmers, mainly because of a misplaced belief that farmers are unable to understand or make any contribution to technical matters, or because of concern that participation might delay implementation or result in design changes that compromise the quality of the final product."

According to Diemer, the institutional contexts of scheme development and scheme management do not encourage irrigation engineers to acquire or disseminate knowledge on actual distribution practices either. Foreign engineers are usually contracted by donor agencies to produce designs for new schemes or to supervise construction. They are rarely involved in the management of their schemes and so cannot incorporate feedback on the distribution practices into their design methods and their assumptions about management. This lack of feedback has led to many schemes deteriorating quickly and needing rehabilitation after only a few years. In theory, such rehabilitation provides an opportunity to take into account the management patterns of operators and irrigators. In practice, rehabilitation simply re-establishes the physical configuration of the original system.

Engineers contracted to produce a feasibility report will hesitate to describe a proposal as unfeasible because they risk losing their contracts, either for the design and implementation of the proposal or for the assessment of new proposals. Helweg goes as far as accusing the consulting firms from developed countries of wasting millions of dollars because they lack cultural literacy.

The professional context explains why design irrigation engineers know little about actual distribution processes. It is sufficient and even more beneficial to them to accept current assumptions about the cultivator and his crops, to see farmers as a group and to accept the need for central management of the schemes, because these assumptions fit the goals of the donor agency and the recipient government. The designers' interest lies in maintaining the status quo.

Engineers from government agencies have also been the targets of the critics: because the funding of most irrigation agencies is dependent on budget allocations and not on their performance in water delivery, most national engineers have little incentives to wrangle with farmers, colleagues and politicians to improve water delivery. Operation usually deviates from the assumptions in design. Political patronage and corruption are endemic in many schemes because they form part of the national political landscape. Maintaining the status quo is also the interest of national engineers. (Diemer)

Criticism of financial institutions

For a deeper understanding of technical assistance in irrigation one needs to look beyond the engineers to the donor agencies that manage the public development funds and to the national departments of planning, agriculture or water that set objectives for agricultural development. Together, these institutions define the terms of reference that the irrigation engineers are contracted to implement. Almost invariably, these bureaucracies are the initiators of the large schemes. After conclusive feasibility studies, calculations of the possible internal rates of return of various design options, and negotiations on funding, the donor agency allocates the millions of dollars requested.

Some critics suggest that financial institutions should review their lending policies and priorities: "Most of the financial institutions tend to give priority to hardware development - probably because software development is more difficult to plan and implement. They argue that software development is essential and should receive higher priority." This view is supported by the statement, similar to the one from ICID quoted earlier, that most irrigation projects fail to realize their targets not because of engineering shortcomings but rather due to the local organization's shortcomings (Anukul-armphai)^[3].

Nijman has studied the links between donor agencies, national governments and irrigation agencies and their connections with their environments, with the aim of identifying the causes of the annual loss of million of dollars in the irrigation sector. Several points stand out:

i) The first is that development banks and other donor agencies have so much public capital at their disposal that is earmarked for investment in developing countries that their officers have difficulty finding sufficient outlets and are under constant pressure to maximize loans and grants. This pressure often adversely affects the quality of the investment decisions. Real-life feasibility and functionality of the investments, as opposed to the feasibility and functionality assumed in the design reports, are not assessed. Performance of the agency and of similar schemes is ignored.

ii) Investment appraisal techniques such as the economic rate of return, cost-benefit analysis and related sensitivity analysis did not render any of the cases (studied by Nijman) unfeasible, as these studies were done after the political decision had been taken to construct a scheme at a certain site. The studies were used to justify the subsidies for irrigation, not to improve economic decision-making.

iii) The fact that development banks are under pressure to transfer public funds to developing countries, plus the fact that much irrigation investment is politically motivated, have instilled an attitude of rent-seeking in the national irrigation agencies. (Diemer)

The performance of international research and donor organizations is criticized in even stronger terms: "A number of international organizations sponsored regional task forces and technical assistance committees...They, however, did not even try to address long-term social and planning issues, which could have provided comprehensive solutions to some of the [water management] problems. On balance, none of the international organizations made a decisive impact on water management in the water-short areas of the world" (Tibor). This rather pessimistic statement lacks fairness, as it does not recognize the contribution of these organizations to the promotion of user participation in irrigation management and of the service concept in the definition of modernization since the mid 1990s.

The slow recognition of design as a main reason of the poor performance of irrigation systems

Few writers have challenged the widespread view that managerial and institutional deficiencies, not the technology of irrigation, are the main causes of poor performance of irrigation projects. However, a few authors have consistently alerted the irrigation community to the importance of technology in the performance of irrigation projects:

i) The question of whether irrigation performance can be improved solely through improved management or whether physical facilities need to be upgraded as well was raised in a World Bank technical paper fifteen years ago (Plusquellec 1985). That paper was cautious enough to state that the different approaches to design of irrigation projects have different managerial and financial requirements and should not be used indiscriminately. It examined the design and management of water distribution at farm level and the relation of the main system to the tertiary system. Finally it concluded that management alone may not substantially improve irrigation performance, unless combined with physical improvements, some at a modest cost.

ii) An audio-visual training programme on how to improve the operation of canal irrigation systems was produced by the external training department of the World Bank in 1988. It stated that the planning, design and construction process must produce a system and conditions capable of accommodating effective management practices. Although significant results were achieved through improved system management in several pilot projects in the 1980s, the programme concluded that changes in physical infrastructure were often needed to push the performance of these systems to a higher level.

iii) Examples of unrealistic designs and operational procedures were discussed in detail in a World Bank technical paper (Plusquellec et al. 1994). Many designs are difficult to manage under real field conditions. Many failures and problems are caused by a design approach that pays insufficient attention to operational aspects.

iv) The ICID publication on automation of canal irrigation systems states that "one of the main factors contributing to poor performance is lack of effective water control" (Goussard). Without further elaboration on that issue, that publication moves to the discussion of the problems of canal operation and presents a state-of-the-art review of the concepts and technologies applicable to automatic operation of canal irrigation systems above farm level.

The author of an IWMI publication on the dilemmas of irrigation systems design (Horst 1998) raises rather unusual and provocative questions: "Is management really the crux of irrigation problems?... Do we need to apply cosmetic surgery by only trying to improve the management environment without considering the technology? Is it not time to examine the root of the problem: the design of irrigation systems?"

Horst recognizes the link between design and management by raising the question of whether it would be possible to design irrigation systems taking into account human and institutional aspects and, if so, what the repercussion on the type of technology would be. He raises some valid questions about complicated technologies and operational procedures, and advocates the simplification of design and operation. He further states that the underlying reasons for writing his book were the combination of denial of the importance of technology vis-à-vis management, increasing indifference to system design and lack of transparency of technology and operational procedures.

Despite the evidence of the negative effect of inappropriate design technology on irrigation performance, it is still puzzling that not much attention has been given to that aspect by policymakers and donors, and even by research institutions. The International Programme for Technological Research on Irrigation and Drainage was created at the initiative of the International Commission on Irrigation and Drainage to specifically address the technical aspects of irrigation research and development. Modernization was one of the themes identified as a major gap in irrigation research in developing countries. However, that theme did not really attract the interest of donors. That the international research institutes involved in irrigation research do not attach more importance to the issue of technology and design of irrigation projects is a matter of concern. The 1998 SWIM paper prepared jointly by IWMI and IRRI staff discusses five main strategies or options for increasing the effective use of irrigation water in rice irrigation systems. That otherwise excellent paper is weak in discussing the rehabilitation and modernization of irrigation projects. It refers to an evaluation of paddy irrigation systems in Asia (Rice) by the Operations Evaluation Department (OED) of the World Bank and cites its questionable conclusion that poor operation and management have a negligible impact on irrigated crop production. It fails, however, to report the main conclusion of the OED study, which was that faulty designs were the main causes of performance far below expected targets.

It is also a matter of concern that, until recently, the importance of appropriate and necessary technology was largely left out of the discussion in the intensive campaign for the transfer of irrigation management to user associations under the World Bank initiative called the International Network for Participatory Irrigation Management.

The dawn of a new approach to irrigation design and management

The historical background discussed above is rather pessimistic, and may leave not much hope that a new approach to irrigation management is going to be adopted worldwide soon. However, the efforts deployed during the last few years by highly motivated individuals from international organizations are encouraging:

i) In the 1990s, the Information Techniques for Irrigation Systems (ITIS) of IIMI together with FAO and national research institutions organized a number of international meetings in Sri Lanka, Pakistan, Malaysia, Morocco and India to exchange experiences on the application of information techniques in irrigation systems and on practical improvements for manual operation.

ii) The FAO Office for Asia and the Pacific organized an expert consultation on "Modernization of irrigation systems - past experiences and future options" in 1996 with the aims to examine the various aspects of modernization and to provide a framework for assessing the need and possibility for adopting the measures required for modernization. This consultation led to the adoption of a multi-disciplinary definition of modernization. There was also a consensus on a number of conclusions including: a) the lack of an appropriate knowledge base to provide adequate forecast of the impact of specific modernization steps; b) the need for essential institutional and policy changes, such as accountability of providers of water services, enabling legislation and enforcement capability; c) the need to expand training from policymakers to farmers; and d) the development of upgraded design and procedure manuals.

iii) In 1994, the World Bank published a technical paper entitled "Modern water control in irrigation: concepts, issues and applications" to stimulate debate among professionals and to increase awareness of the potential of modern technologies for water control and sustainable agriculture. The authors argued that modern design was a thought process which started with the definition of a proper operational plan based on the concept of service.

iv) In response to the need to document the impact of modernization on the performance of irrigation projects, in 1996 the World Bank financed a comparative study of 16 projects. A new method of rapid assessment based on a well-structured questionnaire was developed and used for that study, which evaluates both the input/output external indicators and internal indicators that reflect the mechanisms of operation and management. That study was seminal for the series of training courses on irrigation modernization currently carried out by FAO in Asian countries (Thailand, Iran, Viet Nam).

The above studies and international events remain the initiatives of a few individual experts or staff members (which may last only as long as the staff members remain in place) rather than the result of a policy shared at the highest decision level of their organizations. The need to improve the performance of irrigation projects through a re-visioning of management and design is not given a high priority on the agendas of international forums on water and supporting organizations, such as the World Water Council and the Global Water Partnership. The low profile of irrigation in the debates of the World Water Forum in The Hague in March 2000 contrasting with the fierce debates about water supply and privatization is disturbing. The background paper "A vision of water for food and rural development" presented at the Hague water forum is a comprehensive document dealing with the food demand and the growth in water supply for irrigation and rural development over the next 25 years. The proposed comprehensive strategy to realize the vision includes actions for the development of institutions and human resources and for private-sector development, investment in infrastructure and investment policies. However, the paper is very brief on the deficiencies of existing systems and it fails to present the magnitude of the investments required for rehabilitating and improving the existing irrigation infrastructure. The paper rightly states that the technology to supply crops with an optimal amount of water is already available in drip and sprinkler systems but concedes that there is little chance that all the gravity areas will be converted to more efficient pressure techniques.^[4] The proposed strategy to realize the vision of water for food suggests direct investments to increase water productivity in a number of areas, including the modernization of irrigation and drainage systems, particularly in water-scarce areas, and the dry-season irrigation schemes in Southeast Asia. This seems to ignore the deficient performance of irrigation projects in the humid tropics during the wet season. Overall, the recommendations of the specialized FAO agency at the Water and Sustainable Development International conference in Paris in March 1998 have not been well echoed in the Hague forum, a highly visible international event.

Unfortunately, few large-scale irrigation systems provide on-demand irrigation service to farmers, which is a precondition for efficient water use. There is an urgent need to modernize and upgrade the water control system in most large irrigation schemes by introducing modern management principles, such as volumetric water charges, in order to facilitate crop diversification. Application of new technology generally requires a conducive environment, including knowledge, finance and markets, and needs to be inserted in adequate policies that lift the constraints of agriculture (H. Wolter).

It is encouraging that the topic of modernization was given full recognition at the electronic conference organized by INPIM and FAO in September 2001^[5]. Recognizing the role of the users in the modernization process and the importance of modernization for the sustainability of water user associations is a major step forward.

Considerable efforts have been made by lending agencies to revise their strategy for the water sector and to encourage governments to reform legislation and the role of agencies. These efforts at the global water level are now followed by work to define a new irrigation strategy and to identify the actions needed to implement it. The India Irrigation Sector Review in 1998 is a major initiative of the World Bank in that direction. In a preface to this review, the Indian ministry of Water Resources emphasizes that what is needed is a total revolution in irrigated agriculture, with more focus on the improvement of the performance of existing facilities and provision of a client-focused irrigation service. It supports the recommendations of:

• launching planned programmes, linked with irrigation management transfer to water user associations and participation in decisions and investment costs by these associations, to rehabilitate and then progressively modernize the irrigation systems and

• forming water user associations at the minor and distribution levels and federating them to provide advice on water management at higher hydraulic levels.

These recommendations encompass the most important elements of the new strategy, which will be developed in this paper: the need to shift toward a service-oriented mode of operation and to involve the users in the modernization of the irrigation systems.

II. PERFORMANCE OF IRRIGATION SYSTEMS

Most research studies on performance of irrigation have aimed to monitor the performance over time, for example to determine the impact of a management change, or to analyse the performance of comparable projects. These evaluations mostly focus on an analysis of inputs and outputs of irrigation projects (water, land, labour, value of production, cost of operation and maintenance). These indicators are often referred to as external indicators. These indicators in general do not provide significant information when comparing projects. Obviously projects producing fruit and vegetables have a better productivity than single-crop rice projects. In this chapter, we refer to the performance of irrigation projects compared to the values expected at appraisal or feasibility stage. In Chapter 13, we will define and discuss the use of internal indicators as a tool for the diagnosis of irrigation projects.

The World Bank, as other donor agencies, evaluates the performance of all its operations at the end of the implementation phase, shortly after final disbursements. These evaluations undertaken by its independent Operations Evaluation Department compare outcome then with expectations at appraisal.

Out of the 430 irrigation projects that have been approved by the World Bank since 1950, 313 have been the subject of a post-evaluation. About two thirds of the evaluated projects have had a satisfactory outcome, which is better than the average for all Bank-supported agricultural projects, but slightly worse than the figure for all Bank projects. However, since the late 1980s, the Bank projects are rated for not only their outcome but also their sustainability and institutional development. About 35 percent of the irrigation projects were rated sustainable and about 35 percent would have a satisfactory institutional development impact. These results cast serious doubt about the long-term performance of irrigation projects.

Completion of a project is an opportune time to assess the extent to which operations achieve their stated objectives and to draw valuable lessons for the future. However, conclusions about the technical and economic efficiency of irrigation projects are still speculative at the time when the Bank makes its final disbursements. Impact evaluations are done for a small proportion of operations at full development about five years after completion to determine project impact and sustainability. Impact evaluations are particularly appropriate for irrigation projects whose benefits are long to mature. In 1987, an assessment study of the performance of large-scale gravity irrigation projects was carried out in six countries in different climatic and social environments. An important lesson of that study was the need for more realistic assumptions in the adoption of design standards, especially irrigation efficiency, which in turn affect the cropping intensity, the overall productivity of the project and its economic viability. The main cause for the lower-than-expected performance in economic terms was related to the frequent overoptimistic assumptions regarding efficiency, and the often overlooked impact of poor physical performance in terms of water distribution and concurrent poor construction standards on agricultural productivity (Plusquellec 1990).

The findings of the above study were confirmed by a formal review of 21 evaluations of irrigation projects carried out up to 1990. The review showed that the performance of irrigation projects in economic terms had been less than satisfactory at full development than at either appraisal or completion of their investment phase. For the 21 projects the average outcomes were 17.7 percent at appraisal, 14.8 percent at project completion and only 9.3 percent at impact evaluation. However, their social impact had been substantial and their contribution to food security and poverty alleviation was not in doubt.

Overall efficiency values used for Bank-supported projects in India during the peak lending period 1975-95 were systematically above 50 percent. Most of these projects were rated unsatisfactory at completion when OED started to attach more importance to the links between physical and economic performance of irrigation projects.

Another OED study carried out in 1997 examined the impact of investments in six gravity irrigation schemes in Southeast Asia (Rice). The estimates of economic rates of return not only fell short of appraisal projections by substantial margins, but were all below 8 percent. In one case, the economic rate of return was even negative because the project could not supply half of its design command area. The study stated that the dominant paradigm for government-operated gravity-fed irrigation schemes in the humid tropics was to ascribe the low economic return of irrigation projects to poor operation and maintenance and inadequate farmer organizations. Findings from this 1997 review contradicted this model. The reasons identified for the performance gaps were falling paddy prices, over-optimism about the crop area to be served and projects design faults, including the choice of unsuitable technology.

Contrary to well-entrenched ideas, actual low-price commodities were not the key factor driving the economic rates of return to low levels. For one of the projects, substituting the inflated rice prices at appraisal for the actual prices at completion lifted the re-estimated rate of return by only one point. Another fact which depressed the economic rate of return of these projects was that diversification out of paddy failed to occur at any scheme. Even if the 1980 projections of the price of rice had been realized, a combination of lower-than-expected production and lack of diversification would nevertheless have undermined the economic viability of the investments. That study, which was the more perspicacious of the OED studies on irrigation, was completed when lending for the irrigation sector fell to its lowest historical level. Its findings are still overlooked in the policy discussions on the water and irrigation sectors.

III. A REVIEW OF THE ROLE OF INTERNATIONAL ORGANIZATIONS

The objective of this chapter is to discuss the validity of the criticisms of the international organizations in their support of irrigation development. This chapter reviews the policies and procedures of lending agencies that may affect the performance of irrigation projects. Since the FAO Cooperative Programme (FAO/CP) assists the governments in the preparation of most agricultural projects financed by the World Bank, the Asian Development Bank and other donor agencies, the FAO/CP guidelines are also examined.

1. The World Bank

The lack of an irrigation policy paper

The World Bank has not issued any paper on irrigation policy during its thirty years of lending for that sector, although irrigation has accounted for about 10 percent of its total lending.^{[6] [7]}

During the period of expansion of irrigation, between 1970 and the early 1980s, the Central Department of the World Bank stressed mostly the importance of drainage and greater cost recovery. Cost recovery was particularly salient for the Bank. Policy discussions on cost recovery dominated the debate on irrigation during that period. In 1993, the World Bank published a policy paper on water resources management. The issue of irrigation water pricing generated consid-erable debate during the review process of the paper. The paper advocates a comprehensive approach to water resources, decentralization, stakeholder participation and environmental protection. However, a water policy paper is not a substitute for an irrigation paper, which should provide sector-specific guidance on diagnosing and improving the performance of irrigation projects. The only reference to modern irrigation systems in the water policy paper is made for the objective of pursuing pricing policies that encourage conservation and efficient use of water.

As advocated in this paper, there is a strong need to move from a broad water strategy to the specifics of an irrigation strategy and to adopt a long-term perspective for the improvement and sustainability of irrigated agriculture. Some countries such as India and Brazil, which have developed an irrigation policy with the collaboration of the World Bank, are showing the lead in the right direction.

Pressure to lend

The donor agencies are frequently criticized for their pressure to lend. That issue is frequently addressed within these organizations and strongly rejected by their high-level management. However, any task manager and member of a preparation or appraisal mission is aware of the consequences of a negative evaluation. The efforts of all the members of the mission naturally concur to make the project attractive.

The Bank evaluates the viability of its supported projects in terms of technical, economic, financial and environmental viability. The key parameter of these evaluations has consistently been the estimation of the project economic rate of return, which should exceed the opportunity cost. As mentioned in Chapter 2, a PhD thesis found that the investment appraisal techniques did not render any of the case studied unfeasible, as the studies were done after the political decision had been taken to construct a scheme at a certain site. This strong criticism deserves some comments. On the one hand, the projects examined by the Bank and not submitted to its board for approval, whatever the reasons, are not entered in Bank statistics. It is therefore impossible to determine the proportion of projects that have been rejected because they were not justified economically. On the other hand, senior Bank staff and consulting firms are familiar with the sensitivity of rate of return calculation. They have gained the "expertise" needed for exceeding the rate-of-return threshold value by "manipulating" the key estimated parameters which are used for its calculation, within reasonable limits. This has always been well known but very few designers or Bank staff members have ever pleaded guilty until recently^[8]. The Thai government rejected the Irrigation XIII project after its appraisal by the Bank because of overproduction of rice and sharply declining world prices in the early 1980s. If approved, this marginally justified project would have been rated unsatisfactory at completion. Most of the projects that are dropped during preparation are discarded because of political or other government considerations.

Use of overoptimistic assumptions during design and appraisal

The use of overoptimistic assumptions during project design and evaluation was noted in Chapter 1. This point has been well emphasized in some country irrigation studies. For example, the India irrigation sector review in 1998 stated that there was a tendency to overstate water availability through the analysis procedures used because of social pressures to maximize area coverage and because irrigation efficiency was systematically overestimated. The same India review stated that dependability of water was based on averages rather than on statistical analysis of demand, which would better show the peak demand in dry years, a point which will be examined further in this paper.

2. The Asian Development Bank

In 1998, the Asian Development Bank (ADB) published a working paper "The Bank Policy on Water" after intensive consultation with other policy stakeholders, including member countries, the private sector, NGOs and other external support agencies.^[9] These consultations result in the formulation of principles of effective water policy, which includes the delegation of water services to autonomous and accountable public, private or cooperative agencies providing measured water services to their customers for an appropriate fee. The ADB paper identifies the main challenges to meet these objectives as: i) how to increase investments in new water delivery systems that will effectively meet customer demand; ii) how to upgrade and manage existing systems to reduce demand and run more efficiently; and iii) how to reduce water pollution through recycling. The consultations pointed out the need for higher efficiency in irrigation requiring policies and legal provisions on water rights. The paper states that successful modernization of irrigation systems will generally require viable cycles of investment, operation and management by autonomous and accountable service agencies, with user participation. Explicitly, the ADB water policy recognizes:

• the need to shift toward a service-oriented mode of operation of irrigation systems;

• the importance of modernization of irrigation systems for the successful implementation of the global water policy; and

• the importance to involve the users in the modernization process.

3. The FAO Cooperative Programme guidelines

Identification and preparation of irrigation projects (1984)

Like other financing institutions, the World Bank requires that feasibility studies for conventional specific projects be completed before appraisal. National or international consulting firms commonly assist borrowers in the preparation of projects. Although this phase in the planning process is in principle the responsibility of the borrower, the FAO Cooperative Programme was created in 1967 to provide assistance to borrowers where national capacity was inadequate. In 1984, this FAO division prepared a paper on the "Identification and preparation of irrigation projects" to provide guidance to its staff and that of consulting firms.

The document provided general guidance on the different activities for the comparison of various options, such as: review of available database; assessment of topography, soils and land capability; estimates of irrigation water requirements; assessment of available water resources; and preliminary cost-benefit analysis

The document also provided guidance for the preparation of the engineering studies necessary for the planning of the preferred option before the appraisal of the project. The objective was to provide the necessary technical information to produce preliminary designs upon which estimates of quantities and cost estimates, and ultimately the economic analysis, could be based.

Remarkably, the document did not provide any specific guidance on the technical aspects of project preparation, with the exception of the estimates of irrigation water requirements for the proposed cropping patterns. The document suggested specifically to use an 80-percent probability of excess effective rainfall in the determination of project water requirements and to convert from net to gross water requirements on the basis of empirical local data for efficiency of the types of irrigation systems under consideration or from the FAO Irrigation and Drainage Paper No 24. The last revision (1992) of the FAO document on "Crop water requirements" provides values of conveyance, field canal distribution (Ed) and field application (Ea) efficiencies based on a 1974 survey of ICID and U.S. sources with the remark that these values are applicable to well-designed schemes in operation for some years.

The overall project efficiencies of projects with rotational supply and surface or sprinkler application methods, using ICID value ranges, do not exceed 38 percent for surface application methods and 43 percent for sprinkler.^[10]

Table 1 Overall efficiency rates based on the ICID survey

These overall efficiencies are of the same order as the values estimated at the impact evaluation of irrigation projects and confirm the over-optimism of the values adopted during the planning of irrigation projects. The FAO document, particularly its last revision, should have called the attention of the users of these guidelines to the importance of realistic estimates of the overall efficiency. As indicated in Chapter 1, the gap between appraisal estimates and actual efficiency rates can reach about 40 percent.

Why are planners and designers of irrigation projects from government agencies and financing institutions so optimistic about the hydraulic performance of irrigation projects during the planning process? A participant, from a consulting firm, to a World Bank irrigation seminar in the 1980s answered that question in blunt terms: "If we were realistic, all of us would be out of business." Low efficiency reduces the irrigable areas and/or the cropping patterns and affects negatively the economic viability of the project. It is very intriguing that the over-optimism of consulting firms was neither raised as an issue during the preparation phase of irrigation projects with FAO/CP assistance or during the appraisal by the financing agencies. It is acknowledged here that water lost in a surface irrigation project can be re-used beneficially through recirculation or further downstream. This question of project versus basin efficiency is further discussed in Chapter 6.

The efficiencies assumed during the planning process could be obtained if a number of conditions were met. Designs have to be not only technically sound but also realistic when taking into consideration social and institutional aspects and practical considerations such as access-road conditions, night-shift work and motivation of low-pay staff members, which affect the efficiency of water delivery.

Updating of the 1984 FAO irrigation guidelines (1996)

The 1996 document entitled "Guidelines for planning irrigation and drainage investment projects" prepared by the FAO Investment Centre focuses on new types of thinking and approaches to the planning process of irrigation as crystallized in the 1993 World Bank policy paper "Water resources management" and in the findings of a 1992 Portfolio Management Task Force of the World Bank. That task force concluded that the main problems that constrain the performance of investment projects in various sectors are inadequate consideration of institutional constraints and poor planning of implementation, and a lack of commitment to the success of the projects by the government and users. The 1996 FAO guidelines assume that water policy reviews indicate that irrigation is a justifiable option within the context of the overall water resource strategy of a country. The guidelines therefore discuss issues specific to the implementation ability of the irrigation sector:

• the participation of all stakeholders in the planning and implementation process, to create a sense of ownership and of commitment to the project;

• the creation of water user associations and the transfer of operation and maintenance responsibility;

• the possible role of NGOs in participatory development; and

• the issue of fiscal sustainability, including contribution to capital costs.

The technical deficiencies of irrigation projects and the alternative options to improve design and operation were discussed in sundry workshops, conferences and seminars in the 1980s and 1990s. However, these aspects were deliberately not covered when updating the 1984 FAO paper, because they were available in a number of FAO, World Bank and other papers, conference proceedings and miscellaneous publications. Most of the purely technical content of the 1984 documents reflecting the conventional approach to project planning was repeated where appropriate.

The only additions to the 1984 version on aspects of water management and system operation were limited to short discussions of the planning process and choice of technology and to the questions to be addressed in a typical project document. The complete text of the FAO guidelines on water management aspects is reproduced in annex.^[11]

The section on the choice of technology is a brief reference to a long and passionate debate within the World Bank on competing design visions which was oversimplified in the 1994 review of the Bank experience in irrigation by presenting a polarization of the bank engineers into two camps.

One subgroup sees the problem largely as one of the hydraulic instability of extensively gated, manually operated systems and sees the solution as being the modernization of these systems with automatic downstream control structures and feedback mechanisms to achieve hydraulic stability. The other subgroup of design engineers has accepted the reality of farmer damage in the wet season and gone to the cruder and more robust "structured design", giving up the possibility of just-on-time "on demand" delivery of water to crops in the hope of preserving the civil works.

The conclusion of the World Bank review was that there was inconclusive evidence to favour one camp or the other. The discussion on technology was closed by stating that it was beyond the review's scope to compare and assess the merits of the crop-based or water-based systems. The FAO document states that the discussion need not be continued since it is well covered elsewhere. Alluding to the issue does not provide much guidance to the users of the guidelines.

In summary, both the 1984 and 1996 FAO guidelines for preparation of irrigation projects as well as the World Bank instructions are still driven by the concern of banking institutions. The objective of the engineering studies is to obtain a cost estimate with plus or minus 15-percent accuracy. The changes in project design to improve implementation ability through the devolution of operation and maintenance responsibilities to the users was driven by the recognition that the challenges of operating and maintaining the irrigation infrastructure are often beyond the financial capacity of public-sector institutions.

Not much progress has been made in official thinking about the challenge of closing the gap between the expected and actual outcome of irrigation projects, about improving service to the users, increasing food production and preserving the environment within the constraints of decreasing water resource allocation for irrigation.

Lending for irrigation has progressively changed over time from project-specific investments to sector loans or national/regional projects supporting the objectives of participation and capacity building. These projects often are a mix of low-cost rehabilitation endeavours and management reforms with attention to improved operation and maintenance, and user participation. Low-cost rehabilitation of irrigation infrastructure, in some cases an investment to catch up with years of differed maintenance, cannot correct the deficiencies of the original design, if the causes of such deficiencies are not identified through an in-depth diagnosis of the current system.

IV. TECHNICAL VERSUS MANAGERIAL CHANGES

The idea that the performance of irrigation can be improved by managerial changes was and is still widely spread within the irrigation community. Indeed there are some examples of improved system performance achieved through operational changes supported by effective communication between the agency and the farmers. One of these is the Lower Talavera irrigation system in the Philippines. The research programme carried out by the Rice International Research Institute (IRRI) and the National Irrigation Agency (NIA) developed a rotational water supply schedule which produced dramatic results. Operation of the system was simply converted from continuous supply to a supply by turns between the upper and lower sections of the lateral canals. Water efficiency and productivity were enhanced because of reduced runoff from the head-end areas and increased yields of tail-end farms. This change in operation was indeed rather crude.

Much more complex was the change made in the Dantiwada project in Gujarat, India, where the operation of the main canal system was upgraded to near downstream control through frequent communications between gatekeepers of cross regulators^[12]. It is important to emphasize that this change was achieved through intensive training of the field staff: this unique case might not pass the test of time because of the intensive management and dedication required.

Malano correctly argues that one level of service can be provided with several types of flow control and, conversely, one type of flow control can be used to provide different levels of service. For example, the same water-control technology is used in Mexico to provide water on prearranged demand and in Thailand and the Philippines on the basis of centralized scheduling. The provision of a higher level of service with a given type of flow control requires additional staffing with greater skills and proficiency for planning and executing the system operations. However, the number of staff members needed to operate a system under, say, manual upstream control can be substantially reduced if the same level of service is provided by an automatic system. In most modern irrigation projects in Morocco, where operating on arranged demand would be feasible, water distribution is decided by the irrigation agency, including flow rate, duration and frequency. A local attempt by an innovative water-master in the Doukkala project confirms, if need be, the feasibility of operating these systems on arranged demand.

The trade-off between flow control technology and management has implications on the operational efficiency of the system. It is easy to supply water on demand with a manually operated upstream-controlled system by continuously operating the system largely in excess of the actual demand. Such an operation results in significant wastage of water when demand is low, for instance at night. Thus, the Grand Valley water user association in Colorado provides water on demand with upstream control. The district operates the canals at high flow rate. However, there is a large percentage of spillback to the river. The Seyhan project in Turkey, with abundant water resources, is operated with minimum adjustments for limited demand at night. This mode of operation is known in Pakistan as operation by refusal.

There are unfortunately many unsuccessful examples of schemes throughout Asia where operational changes attempted with the support of research programmes have failed (for example, the introduction of a rigid water distribution pattern in the Nong Wai project in Thailand). The question is whether the practices introduced through pilot projects will continue once these projects end. The more equitable new water-saving distribution strategy is often discontinued because big landowners at the head of the systems exercise their political power to restore their privileges or because the farmers downstream are not consulted.

Focus on managerial changes was the basis of the Bank-supported National Water Management project in India in the mid 1980s. The most important element was the preparation of an operational plan. On the basis of water availability, system characteristics and agricultural options, the plan was expected to define how the system would be operated with respect to the timing and quantity of water deliveries. Only low-cost improvements to the infrastructure needed to support the improved operational plan were supported by the project. The results fell short of appraisal estimates. The completion report concluded that "projects of this type which not only involve technical changes but also have significant social aspects require a high level of farmer participation in irrigation management to be successful". It also stated that introduction of such concepts will require in turn the use of more advanced technologies in irrigation management, which are currently available in India and abroad.

In the same way, examples abound where technological changes alone have not yielded the expected benefits because of a lack of training capabilities during design, construction and operation of the scheme. One such example is the Sidorejo project in Indonesia (Box 3), which was implemented without the full commitment of the irrigation agency.

Sustainable improved performance in irrigation is obtained when combining physical, managerial and institutional changes. The case of the State of Victoria in Australia where the reform process combined all these changes is described in Box 4.

The Office du Niger in Mali, which was often referred to as an example of a fiscally burdened organization, is now a success story for Africa and other regions (Box 5). In both cases, the restructuring of the agencies was combined with a modernization plan of the scheme to improve the flexibility and reliability of water delivery

There is a need to revise the strategy for irrigated agriculture and to address the poor management practices of the large canal irrigation systems through a rethinking of irrigation policy, formulation and technical design of projects, as illustrated by the two successful examples.

V. TECHNOLOGY VERSUS INSTITUTIONAL REFORMS: USER PARTICIPATION

From social to business associations

By the mid 1970s some irrigation researchers from non-governmental organizations, such as the Ford Foundation, and others argued that sustainable irrigation systems required the active participation of the users. By the 1980s several countries started to implement irrigation management transfer programmes whereby irrigators were encouraged to participate in operation and maintenance. Some of these programmes consisted in the creation of pilot user associations spread over a number of irrigation schemes to test the feasibility of involving the farmers in operation and maintenance activities (Maharashtra State in India). Other programmes consisted in a large-scale transfer of the lower level, such as in Pakistan, the Philippines or Indonesia. In Pakistan, over 17 000 associations have been created at the watercourse level, but very few are still active after completion of the lining programme^[13]. In 1987, the Indonesian government decided that all irrigation systems of less than 500 hectares would be transferred to water user associations by the year 2003, with priority given to the systems of less than 150 hectares. The objective of these countrywide programmes was to involve the farmers directly in maintenance activities of the lower level of the irrigation systems, and in some cases to assist the government agencies in collecting irrigation service fees. Management transfer involved only a partial devolution of responsibilities. The government retained some control over operation and maintenance plans and continued to contribute to the financing of operation and maintenance activities. The number of farmers and the areas covered by each association (from 200 to 500 hectares) were usually small.

The approach to irrigation management transfer took a different orientation in the late 1980s with the implementation of the transfer programme in Mexico. In the first phase, the user associations in that country took over the financial and managerial responsibilities for operating the systems below the main canals. In the second phase of the process, still ongoing, the responsibilities of operation and maintenance of the main systems have been handed over to limited-responsibility societies. The average size of the 406 associations created in Mexico by the end of 1999 was about 7 000 hectares, with some reaching 30 000 hectares. The same approach was later successfully adopted in Turkey, where the irrigation agency transferred the management of about 1.6 million hectares^[14]. This leap in scale and rate of transfer has stimulated some other countries to shift to accelerated programmes, for example the State of Andhra Pradesh in India, which created over 10 000 associations in July-August 1997.

A key difference exists between the approach adopted in the 1970s-1980s, mostly in Asian countries, and the one developed in Mexico and Turkey. In these two countries, the programmes aimed to involve farmers in representative governance and not to maximize direct user farmer participation in operation and maintenance. The associations created in these two countries have similar designs, responsibilities and functions to those of the existing associations in the United States, Spain and some Latin America countries such as Chile, Peru and Colombia. The Asian and Mexican models are sometimes referred to as "social" and "business" associations respectively (Facon). The creation of social water user associations has tended to follow a gradual approach experimenting with pilot associations.

The business-prone associations are responsible for water distribution, fee collection, maintenance, conflict resolution, and representation of farmers with discussions with public agencies. To fulfil these functions, these associations are legal entities that can enter into contracts. They have the power to enforce rules and regulations. The farmers are not directly involved in the management of their systems. These associations hire professional staff for the actual management. The members of these associations through their elected boards define the rules and regulations to be followed by the hired employees.

The social associations are seldom self-sustaining. The concept of business-type associations is rejected by the irrigation bureaucratic establishment or not applicable because of the perceived incapacity of farmers to manage large systems. A multi-tiered organization is now emerging as a possible solution for the management of large-scale irrigation schemes with a large number of small farmers. This model consisting in multi-level organizations is consistent with the definition of modern irrigation. A modern irrigation system consists of several subsystems or levels with clearly defined interfaces where water is controlled and measured. The strict application of that model with a formal independent organization at each level could result in an excessive number of layers of management and formal farmer organizations. For example, the proposed organizational structure of the Mahakali project in Western Nepal would include a central coordinating committee at the project level, eight coordinating committees at area level, forty user associations at the block level, 1405 tertiary committees and nearly 10 000 water user groups at the outlet level. It would be not only time consuming but even counterproductive to organize so many associations and groups at each level.

In practice, some levels of management can be combined within one organization, while maintaining the modern concept of service from the higher to the lower level. In addition, the responsibilities transferred to each user organization could be either governance or management functions, or a combination of the two. For example, the higher-level user organizations could be involved in major decision processes, such as rules and regulations, the annual maintenance work programme, revision of the water charge structure, the annual budget and the timing of the irrigation season, the day-to-day management of the main system being the responsibility of an irrigation agency or company. Operation and maintenance of the two middle levels of a system could be transferred to water user organizations. This model can be applied in a pragmatic way depending on the size of the scheme and configuration of the irrigation system, the capability of the farmers and the willingness of the irrigation agencies to accept fundamental changes in their roles.

Figure 1 A multi-tiered user organization/agency of a surface irrigation scheme

Note: Each user organization may have either governance and management functions or both

Impact of irrigation management transfer over performance of irrigation projects

A number of specialists thought that taking irrigation and drainage system management out of the direct governmental sphere would inevitably lead to improvements in the sustainability of irrigation and drainage systems and in agricultural production. The philosophy was that users were more likely to operate systems effectively and according to their requirements and also pay for the operation if they were also the owners. The dominant perception was that public irrigation management organizations lacked the incentives and responsiveness to enhance performance whereas water users had a direct interest in cost efficiency, profitability and proper physical condition of the irrigation facilities.

However, despite the widespread adoption of management transfer programmes, there is still little information about their impact on the agricultural performance of irrigation systems. A review of PIM impact studies in 1997 noted that the impact is typically not noticeable in terms of agricultural performance. In Sri Lanka there has been no detectable change in irrigated area, crop patterns, cropping intensity or yields. PIM has neither improved nor interfered with agricultural productivity (Vermillion).

Another review of experiences in irrigation management transfer in selected countries in Asia revealed that the main impact has been a gradual decline in government financing of the operation and maintenance of irrigation systems, whereas water user groups are making a modest contribution towards maintenance. The analysis also shows that there has been a modest improvement in the irrigation service following transfer. The review concludes: "the evidence of the impact of IMT on systems' operations is not conclusive but seems to suggest that it has not resulted in a deterioration of systems' operations nor in a decline in agricultural performance" (Samad).

The general impression of an international workshop in Cali, Colombia, in 1996 organized to examine the impact of irrigation management transfer (IMT) in selected countries was that after turnover, services were substantially improved in terms of timeliness, reliability and equity in four countries (Mexico, Turkey, Colombia and Taiwan). By contrast, the social water user associations that were developed for the purpose of providing cheap labour for maintenance and collecting water fees were consistently found weak or paper associations. The business-type water user associations that hired staff to distribute water and ran the distribution system similar to a business operation were often very strong.

Two papers on Mexico irrigation presented at the ICID Congress in Granada, Spain, in 1997 illustrate the inconclusive evidence of the impact of irrigation management transfer on the performance of irrigation projects. Johnson noted that the water user associations have proven capable of managing irrigation systems and in the process have reduced annual government subsidies for irrigation water by more than US$200 million. Maintenance activities by the user associations have stopped the deterioration of the infrastructure.

However, Johnson concluded on a pessimistic projection by stating that "In Mexico with relatively good irrigation under the irrigation agency, it is unlikely that IMT alone would result in dramatic increases in production". A second paper presented by specialists from the Mexican irrigation agency noted that the agricultural productivity of 38 irrigation districts transferred in 1994 had dramatically increased. The average crop yield increased by 39 percent from 7.9 to 11 ton/ha and the water productivity by 62 percent from 750 to 1220 kg/1000 m³ between 1989 and 1996. Wheat yield increased by 41 percent after the transfer. This paper, however, noted that this increase was the result of the transfer programme and two complementary programmes: a rehabilitation, modernization and on-farm improvement programme coupled with the modernization and improved techniques of on-farm irrigation.

Some projects claim a substantial increase in cropping intensity and crop yields, but these projects were performing at a very low level before the transfer. The transfer of management to user associations may have contributed to reducing the chaos in water distribution and the level of inequity between head- and tail-enders. The impact is less evident in projects that were previously managed by irrigation agencies according to well-established rules. These agencies have transferred their practices to the associations. For example, the user associations have adopted the mode of water delivery on pre-arranged demand used in most Latin American countries after transfer. Some improvement in the service provided, such as reducing the time lag between demand and actual delivery could be made through technological changes such as an improved communications system, computers and operational procedures.

In conclusion of this chapter on user participation, this paper argues that:

• Simply rearranging the "deck chairs" is not likely to achieve significant improvements in irrigated agricultural productivity and will not meet the broader objectives of integrated water resource management (Malano);

• Small and social associations or water groups responsible for operation and maintenance at the tertiary level have little potential for improvement;

• Most institutional improvements cannot be fully implemented without the right physical environment;

• Physical and institutional improvements in irrigation are not isolated actions but are self-supportive;

• Any strategy for improving performance of the irrigation sector should consider the relationship between the design of user associations and their functions and the strategy for a higher level of service.

The rapid devolution of management to business-minded associations adopted by some countries is likely to be a better strategy than the formation of social associations. However, it requires a high level of commitment by the political authorities and the government agencies and by the farmers, a massive mobilization effort by the government to convince the farmers to take over and to organize the associations and a massive training programme in a number of water management, accounting and operational issues. This does not imply that formation of small groups of users is a bad strategy. These small groups are needed to organize water distribution at the lowest level in irrigated areas where small farms dominate. However, if these water groups are nested in a multi-tiered organization up to the main system or project level, they can participate in important decisions on activities affecting their lives. This is the basis of the emerging new concept where new concepts of user participation and modernization of the systems are converging.

VI. IMPROVED IRRIGATION IN THE CONTEXT OF WATER RESOURCE MANAGEMENT

In recent years there has been a growing interest, particularly among IWMI researchers, in improving the understanding of the concepts of efficiency, water loss and agricultural water use. Although only a part of the water diverted to an irrigation system is effectively used by the crops, typically less than 50 percent, the remaining water either drains to the river system or seeps into the ground. In both cases it can be used downstream for another purpose or pumped for irrigation, very much increasing the overall efficiency at the project and at the basin levels. Drainage water that flows back into a stream or to subsurface areas is not lost or wasted in physical terms. The distinction between field efficiency, project efficiency and river basin efficiency is very important. Improving irrigation efficiency to increase cropping intensity or expand irrigable land might deprive another irrigation system or another use downstream, such as navigation, water supply or environmental flows to control marine intrusion in deltaic areas. These issues have led to an ongoing discussion on whether efficiency improvements can produce any water that can be reallocated to other users. For example, some California planners dismiss the potential for water use efficiency improvements in that state because, they argue, such improvements will not produce much "real" water.

This discussion should however not be an excuse for not improving irrigation systems.

The main cause of water-logging and salinization, requiring expensive drainage works, is excessive irrigation. Improving water application can substantially reduce the hazard of salinization as well as the cost of drainage. Increasing irrigation efficiency can have a significant effect in reducing the load of salts that must be removed from the soil annually (Hillel). Drainage water gets polluted by removing salts from the soil, as well as by fertilizers and pesticides.

Inefficiency at irrigation-project level increases pumping costs in projects that depend on lift from a river, such as many projects in Romania pumping from the Danube River or the very large Kashi project in Uzbekistan.

Inefficiency makes uneconomic a number of investments in irrigation because it reduces the areas that can be irrigated and the expected benefits. Inefficiency also depresses crop yields, particularly in the tail end of irrigation projects, because of changes in irrigation scheduling such as increased intervals of irrigation, to compensate for the rapid depletion of water resources.

Inefficiency increases non-beneficial losses by increasing evaporation and transpiration from soil and free water surfaces. This is the case, for example, in the Tarim basin in China where inefficient irrigation has caused large waterlogged and saline areas surrounding the lower areas of surface irrigation projects. It has been claimed that the overall efficiency of the Nile system in Egypt may be as high as 90 percent, counting part of the water flows released to the sea as requirements for environmental purposes. However, it is not known how much water is lost through unproductive evaporation from fallow and wet lands throughout the valley and the delta.

Efficiency improvements also produce other benefits, including improvements in human health, more reliable in-stream flows, ecosystem and habitat restoration, reductions in the cost of treating drinking water, less environmental contamination by agricultural chemicals, and reductions in the economic cost of multiple unnecessary withdrawals of water (Gleick).

It would be a mistake to leave the impression that improving the efficiency of irrigation systems is not an issue because the lost water can be re-circulated or used further downstream. It would also be incorrect to suggest that the highest efficiency should be obtained in all circumstances, especially where there is a high potential for groundwater reuse.