The objective of this study is to assess the agro-economic impacts of invest-ments in gravity-fed irrigation schemes in the paddylands of Southeast Asia, and to determine whether and how the quality of operation and maintenance (O&M) services influences the sustainability of those impacts. The study team selected six gravity-fed irrigation schemes with reservoirs for water storage in Thailand, Myanmar, and Vietnam. Four were large schemes-at least 40,000 hectares-and the other two were small tanks of about 1,000 hectares. The six schemes, which were widely dispersed across the region, were chosen for their variety and not their representativeness. Nevertheless, the findings were similar at all sites, which suggests that the lessons learned have wider application. The study included an audit of a flood control and drainage project at three sites in Bangladesh to identify differences in O&:M organization and effectiveness between irrigation and flood control. Map 1 shows all nine scheme sites.
Field work was carried out in three phases in late 1994 and early 1995. An Operations Evaluation Department (OED) impact study team comprising Bank staff and international and local consultants visited farmers and offi-cials at the scheme sites and pertinent public irrigation authorities. The field work had a participatory orientation, as the team arranged interactive group and household interviews in all four countries. The field work was also carried out rapidly: on average the team spent one-and-a-half weeks at each site.
O&.M performance: the current model
With few exceptions, O&M performance by both agencies and irrigators on the large, government-operated, gravity-fed irrigation schemes in Southeast Asia is dismal. This conclusion confirms, but goes beyond, the frequent re-ports of degraded public infrastructure in developing countries, particularly irrigation structures. Measures to reverse the problem of "rusting, crumbling infrastructure" were central concerns of the World Bank's World Development Report 1994: Infrastructure for Development (World Bank 1994,4).
Poor O&M and lowered benefits from irrigation investments are causally linked. A recent report (Jones 1995) reviewed not only the Bank's record, but experience in similar non-Bank projects throughout the world. The review looked in particular at the problems of paddy irrigation O&:M in the humid tropics of Asia. It focused on alarming behavioral patterns that suggest that unusually intense irrigator resistance to O&:M design standards degenerates into anarchy and chaos.
Social scientists have paid increasing attention to the role of formal and infor-mal associations of beneficiaries of public assets. Nowhere is this more evident than in the literature on irrigation, where researchers have attempted to define the characteristics of association that promote improved O&M. Elinor Ostrom's seminal publication, Crafting Institutions for Self-Governing Irrigation Systems (Ostrom 1992), offers eight design principles. Researchers both within and outside the Bank have prepared other lists of conditions for success. In the vanguard is the International Irrigation Management Institute, established in Colombo, Sri Lanka, in 1984 with international funds and a mandate to study all aspects of irrigation organization and management. The institute has led the way in promoting the transfer of certain O&M responsibilities by public agencies to water user groups (WUGs). Whereas the literature conveys a sense of hope that O&M will improve with greater par-ticipation by farmers, it is less optimistic in its assessments of the prospects for public agency performance. Observers generally believe that deep-seated cultural features of irrigation bureaucracies inhibit effective O&M work.
In short, the dominant model is one of incompetent bureaucracies combined with nonexistent or weak irrigator associations struggling, largely without success, to impose a sophisticated O&M routine upon opportunistic farmers, with the result that production benefits attributable to the irrigation are far below their potential inevitably, this model guided the design of the impact study.
Agro-economic impacts of the six schemes
The predominant crop in all the schemes is paddy, and they have many other similarities. The most important distinguishing factor is the degree of water abundance or scarcity at the reservoir level compared with the area to be irri-gated. Four of the schemes have more water than they use. The other two, located in the central dry zone of Myanmar, have much less than planned.
At four of the six sites, including three of the large ones, the areas supplied by the irrigation systems are significantly less than planned. Overoptimism, engineering errors, lower than normal rainfall, and failure to extend the ter-tiary canals are the leading explanations, but vary in importance from site to site. Cropping intensities are also substantially lower than expected at three sites and are falling at a fourth. Without a high level of water recapture by small private pumps on fields beyond reach of the canals at Dau Tieng in Vietnam, that scheme would also show much lower average cropping inten-sities than projected. Only one scheme-a tank in Myanmar-has attained both its area and intensity targets. Paddy yields vary widely-between schemes and in comparison with expectations-but a weighted average for the wet and dry seasons at all the schemes is about 3.3 tons, or 85 percent of appraisal projections. However, even where soil conditions permit a shift to other field and specialty crops, farmers have not diversified out of paddy. Indeed, the concentration on paddy has increased.
Re-estimates of total scheme production of paddy and of a few other major crops at the two schemes where paddy is not completely dominant show that output is between 32 and 73 percent of appraisal estimates for five schemes (four of them are below 50 percent). The Myanmar tank is again the exception, but accounts for only 1,200 hectares out of a total of 207,000 hectares served by all six schemes. These production shortfalls undermine the economic rate of return for each scheme. The returns have also been driven down by the decline of the international price of rice between 1981 and 1986. Rice prices projected by the Bank in early were appraised. The small upturn in late 1995 had little effect on the out-come. Together, the production shortfalls and lower prices result in rates of return at or below 7 percent for all schemes and a negative rate for one scheme.
Smallholder irrigated paddy can no longer provide the basis for a growing, or even stable, household economy. Farmers' incremental and total financial incomes from average size holdings range from about $600 to $2,000 per year. For Vietnam and Thailand, actual incomes are only 10 to 30 percent of appraisal estimates. The gap is lower in Myanmar, but mostly because ap-praisal projections were less ambitious. The implications vary; for example, the accelerating rate of outmigration from the two Thailand schemes con-trasts with the stability of farm communities in Myanmar. However, as econ-omies expand, irrigated paddy will not be able to compete with the incomes to be had from other employment opportunities.
O&.M performance: in practice
The field surveys concentrated on assessing agencies' and irrigators' perfor-mance in operating and maintaining the schemes. Team members noted the condition of canals and control structures; agency activity in allocating, dis-tributing, and maintaining the flow of water; and the strengths and weak-nesses of farmer O&M, especially as managed through informal associations, WUGs, and the few higher level federations of WUGs encountered. Overall, agency and irrigator performance appears to be substantially better than the image presented in the depressing model described earlier, and runs counter to allegations about farmers' disinterest in maintaining the irrigation assets that serve them, about feuds over water supplies that tend toward anarchy, and about an insurmountable bias against O&M among agency engineers.
Even in the two schemes short of water and on the ends of the distributary canals in the other schemes where periodic water deficits appear despite an overall abundance of water, relationships between headenders and tailenders are more civil, accommodating, and fair than the model suggests. Advan-taged irrigators do use their advantages and other irrigators do complain, but nowhere is the level of agitation alarming. The absence of a significant yield differential between the heads and tails of the watercourses underlies this low level of conflict, but it also suggests that water is reaching the tailenders.
This civility and relatively fair sharing of water has been accomplished even though the WUGs-which are present on all schemes-are not functioning at expected standards. The one exception is the internationally assisted sections of the scheme at Lam Pao, Thailand, where both the WUGs and the federated groups of WUGs organized along some of the distributaries meet the criteria laid down in Ostrom (1992) and clearly illustrate the improvements in the system and on the farms that follow effective organization. However, that level of organizational performance is unique among the schemes studied. Elsewhere at Lam Pao the WUGs accomplish their basic purpose, which is to keep the tertiary canals and watercourses open and to assemble labor to help the agency keep the larger canals clear, but they do not bring about any larger group cohesion or participatory activity. In Myanmar the WUGs are subordinate to the village councils and do not seek or achieve any higher purpose. In Vietnam the WUGs are barely more than arms of the provincial irrigation authority. In short, strong WUGs are not a primary cause of the relatively successful O&M activity observed in the schemes studied. Farmers cooperate to achieve at least basic O&M objectives regardless of the level of maturity of the formal organization.
The contrast with the flood control schemes is instructive. In Bangladesh no attempt was made to form user groups associated with the flood control structures. The character of these structures and the benefits they provide are such that the beneficiaries do not even associate spontaneously to take care of them. The rate of degradation of the embankments, sluices, and other equipment is much more alarming than on the irrigation schemes. Moreover, with no official stimulus to promote farmer association to protect scheme as-sets, even the small irrigation inlets that were installed in the embankments at two of the three sites were, for the most part, neglected. Flood control O&M fits the model better than irrigation O&M.
O&.M influence on agro-economic impacts
Do the lapses and failures of operation and maintenance that the study team observed at all the sites contribute to lower production on the five poorly performing schemes? Again, the answers are unexpected. The study reveals no substantial negative constraints on irrigated production attributable to poor performance in O&M. Those O&M operations that are essential to keep sufficient supplies of water flowing to the great majority of the fields are ad-equately carried out. Yet it was the assumption that such a negative relation-ship did exist that prompted the study. In many other countries and on many other schemes the record is undoubtedly worse. However, the study suggests that a more discriminating analysis of O&M is warranted.
Clearly some components of O&M are under control and others are not. This report discusses four parts of an O&M matrix: agency operations, agency maintenance, irrigator operations, and irrigator maintenance. Com-mon weaknesses are agencies' inability to keep some of the larger distributaries clean of silt and weeds throughout the cropping season and farmers' indifference to and neglect and destruction of tertiary gates. These failures are not systemic, however, and budget constraints, scheduling problems, and farm-level disincentives that require tailored and well-crafted solutions can usually explain them.
The study covers a number of other issues, including the following:
- The dismantling of complex technological control systems imported in the 19808 by foreign consultants. The schemes in Thailand and Myanmar had adopted a computerized water allocation scheduling and monitor-ing program (WASAM), but it proved to be too demanding on agency staff, and the protocol and measuring devices have been abandoned.
- The ongoing attempt by a group of consultants in Thailand to modernize the control system by simplifying WASAM and substituting weirs and gates that require less human intervention. The modifications are of two types: fixed structures that have no adjustments and structures that adjust automatically to changes in water levels, each of which has its par-ticular advantages.
- The rejection by farmers of both rotations and gates. Rotations do occur, but they tend to break down under conditions of shortage, which is when they are most needed. The biggest problem is not the sharing of water within a tertiary system but among tertiaries.
- The ability of one country's agency-supported by the local administra-ti on-to bypass rotations altogether by simply cutting off tailend tertia-ries. The irrigation agency informs the farmers in advance which of them cannot be supplied. In other parts of the world, shortages are usually shared equally.
- The different degrees of success in mobilizing free irrigator labor to clean the larger canals. In Myanmar the authorities can gather large numbers of irrigating and nonirrigating farmers. Group work on dis-tributaries is rooted in Burmese irrigation traditions.
Findings and recommendations
The finding that dominates the study has little to do with O&M. Given that they offer poor economics and low incomes, these paddy irrigation schemes face an uncertain future. Improved O&M performance will not rescue them. In fact, the study finds that this causality is being reversed. As the lack of competitiveness of paddy farming drives younger family members off the farms and the older members who stay behind concentrate on basic subsis-tence crops, social capital will erode and O&M standards are likely to suffer.
Based on the six schemes studied, the team has come up with a dozen recom-mendations grouped into four sets of proposals:
- Sharpen the response to O&M failures by disaggregating O&M; identifying the poorly performing components; and dealing with disincentives spe-cific to each, such as the tertiary gates that farmers below consider un-friendly.
- Simplify the infrastructure and operations technology by converting to fixed and automatic controls that need less human intervention and by sup-porting authorities who plan with the farmers to abandon equitable rota-tions by rationing water during emergencies.
- Promote the transfer of management to farmers and their WUGs judiciously by recognizing that organizing user groups pays off, but also accepting that immature WUGs cannot handle some management responsibilities.
- Improve household earnings by diversifying cropping systems and sup~ porting research, extension, and marketing services keyed to specialty crops and integrated, high-value farming.
The relevance of these recommendations beyond the schemes studied is un-certain, because they depend on cultural and institutional parameters that may be country specific and on engineering and agronomic considerations that may be project specific. For example, readers who commented on a draft of this report were concerned that the findings were both too rough on and too forgiving of O&:M performance in the region. For the moment, these recommendations are better viewed as hypotheses. Additional empirical work is needed to validate the range of countries and projects within or outside the region for which these recommendations are appropriate. OED has pro-posed holding a regional workshop where validation would be one of the principal objectives.