Report No. 15824 Irrigation O&M and System Performance in Southeast Asia: An OED Impact Study June 27, 1996 Operations EvalLuation Department Document of the World Bank Abbreviations and Acronyms BWDB Bangladesh Water Development Board ERR Economlic Rate of Return FCD Flood Control and Drainage ID Irrigation Department (Myanmar) IIMI Intemnational Irrigation Management Institute NEWMASIP North-East Water Management and System Improvement Project (Thailand) NIA National Irrigation Authority (Philippines) OED Operations Evaluation Department O&M Operation and Maintenance PAR Performance Audit Report PCR Project Completion Report PRA Participatory Rural Appraisal RID Royal Irrigation Department (Thailand) RRA Rapid Rural Appraisal SLORC State Law and Order Restoration Council (Myanmar) SAR Staff Appraisal Report WASAM Water Allocation Scheduling and Monitoring program WDR World Development Report WUGs Water User Groups The World Bank Washington, D.C. 20433 U.S.A. Office of the Director-General Operations Evaluabon June 27, 1996 MEMORANDUM TO THE EXECUTIVE DIRECTORS AND THE PRESIDENT SUBJECT: Irrigation O&M and System Performance in Southeast Asia: An OED Impact Study Attached is an impact study of irrigation investments and operation and maintenance (O&M) in the humid tropics of Southeast Asia, prepared by the Operations Evaluation Department. The study covers six gravity-fed irrigation schemes, with reservoirs, in Thailand, Myanmar and Vietnam. Four IDA Credits and two Bank Loans supported these schemes, which were the sole or dominant components of the projects. All were approved in FY74-FY83 and closed by FY9 1. Project Completion Reports, Performance Audit Reports and an earlier Impact Evaluation Report provide a substantial baseline for this review. To provide perspective on the irrigation portfolio, the field studies included an IDA-supported flood control and drainage project covering three schemes in Bangladesh. The study challenges common precepts about water management in the humid tropics, according to which major threats to the sustainablity of irrigation investments arise from mismanagement by official agencies and distributive anarchy due to opportunistic behavior by farmers. Based on field surveys, the impact evaluation concludes that paddy-based irrigation projects are yielding returns of 7 percent or less, well below appraisal estimates. But in most schemes surveyed, the gap between appraisal expectations and actual results cannot be attributed to decaying infrastructure or distributive waste. Rather, it is due to the combined result of adverse price developments for paddy, excessively optimistic estimates of crop areas served, project design faults, and construction inadequacies. The flood control schemes present a different profile. There, the production impacts have come closer to appraisal expectations, but maintenance standards are deplorable and the sustainability of the structures and benefits is in doubt. With that exception, field observations disclose broadly satisfactory public performance in O&M. In all the countries surveyed, vandalism and neglect affect mostly structures which are ill-suited to community needs, such as tertiary gates that inhibit the flexible operating protocols favored by farmers, inlets which induce excess flooding in the lower reaches, and embankments which prevent drainage. When siltation and weed infestation threaten irrigation, labor is readily mobilized to clean up watercourses, confirming that collective action will occur if substantial benefits broadly available to the community of irrigators are at stake. Water diversion by headenders is also not a major problem: the "free riding" chaos described by institutional economists is not noticeably present in the irrigation perimeters surveyed in Thailand, Myamnar and Vietnam. Within the constraints of engineering designs, tertiary systems distribute water efficiently (if not always fairly in terms of timeliness or reliability) through customary arrangements which tolerate significant advantages to headenders while allowing adequate water deliveries to tailenders. 2 This finding would be less surprising if all rather than the majority of the schemes could be classified as 'water abundant." However, the same behavioral profile is presented in the two schemes where persistent water deficits preclude a large minority of farmers from reliable supplies. Thus, while formal systems of rotations are rarely practiced, available "social capital" facilitates cooperation at the locai level. Water distribution arrangements at the tertiary turnout are more problematic, since, in the schemes under review, responsibility for allocation lies with unfederated tertiary water user groups and official agencies. The operation of tertiary gates is frequently shared among agency field staff and farmers and is almost as frequently mishandled. Highly sophisticated measurement and allocation programs promoted by consultants have been abandoned, both in Thailand and Myanmar. They are premature for paddy cultivation, given the loose institutional arrangements and high water tolerance of rice. Field studies identified a wide variety of organizational procedures, including some reasonably well administered systems developed locally (Myanmar and Vietnam), or with targeted technical assistance from donors (Thailand), involving mixtures of hierarchical authority and user participation shaped by country traditions. In all countries, cost recovery plays virtually no role in water management or O&M financing despite Bank conditionality. In the absence of financially autonomous irrigation associations, this result is not surprising. Given the lack of reliable and controlled deliveries, water can hardly be considered a toll good, especially when it is relatively abundant. Further, the "consumer surpluses" associated with declining paddy prices, and already transferred from irrigators to rice buyers, make further taxation of paddy cultivation politically unattractive. The turning of the terms of trade against paddy growers in the early 1980s-after these projects were approved-casts a shadow over all plans for improved O&M. Unless higher-value cropping systems are put in place, the sustainability of these paddy farms is in doubt. Present standards of farmer O&M are also at risk. The modernization of Thailand's economy pulls farmers to the factories, even as low paddy incomes push them out. Vietnam and Myanmar (in that order) can be expected to follow the same route. The study proposes specific adjustments in Bank irrigation lending practices for similarly situated countries. First, rehabilitation programs aimed at improved O&M must be more discriminating, focusing on the weakest activities and providing incentives to bring them to appropriate standards. Second, sophisticated water distribution and monitoring technologies should be put aside in favor of less demanding, more automatic control structures, at least until intensive, diversified cropping patterns are in place. Third, Bank projects should emphasize capacity building for effective water distribution associations, with priority accorded to federating user groups beyond the tertiary water course level. Hybrid organizational arrangements that take careful account of existing social networks and that combine community labor and official agency support should be piloted to improve maintenance of canals and gates. Fourth, the engineering of projects should take adequate account of hydrological, topographical, and social environments: participatory project design is important everywhere but should be mandatory in flood-prone, drainage-poor, densely populated areas. Fifth, government policy should favor crop diversification and intensification, supported by enhanced extension and marketing services; cost recovery exhortations should be toned down until water systems are reliable, remunerative crops are introduced, and volumetric water delivery becomes practical. Most of all, the popular myths surrounding irrigation development in the humid tropics should be set aside, replaced by pragmatic concepts which bring together public irrigation agencies, local authorities, and civil societies to remove specific constraints and strengthen appropriate incentives. Attachment Contents Preface ...............3 Evaluation Summary ...............5 1. Introduction and Background .11 2. Operation and Maintenance: Summary of the Literature A. O&M Performance .13 B. Institutions .15 C. Contributing Factors .17 D. Influence of O&M Performance on Production .18 3. Scheme Sites and Infrastructure A. Thailand .20 B. Myanmar .20 C. Vietnam .21 D. Bangladesh .21 4. Operation and Maintenance: Systems and Performance A. Agency Level .23 B. Irrigator Level .25 C. Water User Groups .26 D. Cost Recovery .28 E. Comparisons with Flood Control .29 5. Agro-Economic Impacts A. Agricultural Impacts .30 B. Farmer Financial Benefits .32 C. Economic Rates of Retur .34 6. Influence of O&M Performance on Agro-Economic Impacts .38 7. Findings and Recommendations A. Findings .41 (1) Operation and Maintenance .41 (2) Agro-Economic Impacts .43 (3) Relationship Between O&M and Impacts. 45 B. Recommendations .46 This report was prepared by Edward B. Rice (Task Manager), with support from Roger Slade (Bank), Robert Yoder, Jayantha Perera, Annemarie Brolsma and Douglas Vermillion (consultants). Afi Zornelo and Megan Kimball provided administrative support. Also participating in the field studies as members of the OED team were Vinh Le-Si, S.A.M. Rafiquzzamen (Bank), Azharal Haq (IIMI), U Tun Naing (UNDP), Tassanee Ounvichit (RID), Sinee Chuangcham, U Hla Myint, and Tran Kim Thanh (consultants). Appendix 1: Incremental and Total Paddy Incomes: per Farm and per Hectare ........... ........ S 1 Tables in Text 5.1 Irrigat.on Area, Intensity, Yields and Production ................................. .......................... 31 5.2 Annual Total Irrigated Paddy Incomes ........................................................... 33 5.3 Output-Input Price Ratios ........................................................... 34 5.4 Economic Rates of Return ........................................................... 35 Figure in Text 5.1 International Rice Prices-Actuals and Bank Forecasts ........................................................... 36 Map Study Area: South/Southeast Asia (IBRD 27561) ................................. 10 Annexes A. Thailand: Lamn Pao and Maeklong Right Bank Irrigation Schemes (Full Impact Report) B. Myanmar: Kinda and Tank Project Irrigation Schemes (Full Impact Report) C. Vietnam: Dau Tieng Irrigation Scheme (Full Impact Report) D. Bangladesh: Drainage and Flood Control II Project (Principle Findings of the Performance Audit) Maps at End of Volume Overview 1. Study Area: South/Southeast Asia (IBRD 27561) Annex A: Thailand 2. Lam Pao Irrigation SubProject (IBRD 27562) 3. Maeklong Right Bank Irrigation SubProject (IBRD 27563) 4. Chao Phraya and Maeklong Basins (IBRD 22765) Annex B: Myanmar 5. Kinda (Nyaunggyat) Darn Multipurpose Project (IBRD 27564) 6. Tank Irrigation Project, Kinmundaung Subproject (IBRD 27565) 7. Tank Irrigation Project, Azin Subproject (IBRD 27566) Annex C: Vietnam 8. Project Envisioned at Appraisal (IBRD 28133) 9. Assumed Development through 1988 (IBRD 23308) 10. Actual Development through 1995 (IBRD 27560) 11. Actual Development through 1995, Canal Detail (IBRD 27559) Annex D: Bangladesh 12. Bangladesh: Subproject Sites (IBRD 27713) 3 Preface This is an Impact Evaluation Report on a cluster of six gravity irrigation schemes supported by the Bank in three countries of Southeast Asia (Thailand, Myanmar and Vietnam). The main objective was to assess agro-economic impacts at least five years after completion of the investment operations, and the influence of operation and maintenance (O&M) performance on the sustainability of those impacts. To gain additional perspective on O&M, the study also assessed O&M performance at three schemes covered by a flood control project in Bangladesh. Prior to the study, Project Completion Reports (PCRs) had been prepared for the six projects that supported the nine schemes. Performance Audit Reports (PARs) had been prepared on two of those projects; two others were audited in tandem with this impact evaluation. Another one of the schemes in Thailand has been reviewed by OED in an earlier irrigation impact study, providing a stronger foundation for follow-up. OED wishes to express its appreciation to the four national irrigation agencies that assisted with the evaluations: the Royal Irrigation Department in Thailand, the Irrigation Department in Myanmar, the Ministry of Water Resources in Vietnam, and the Bangladesh Water Development Board. 5 Evaluation Summary Introduction 1. The objective of the study is to assess the agro-economic impacts of investments in gravity irrigation schemes in the paddy lands of Southeast Asia, and determine whether and how the quality of operation and maintenance services (O&M) influences the sustainability of those impacts. Six gravity irrigation schemes with reservoirs for water storage were selected in Thailand, Myanmar and Vietnam. Four were large schemes-at least 40,000 ha-and the other two were small tanks of about 1,000 ha. The six schemes, widely-dispersed in the region, were chosen for variety and not representativeness. Nevertheless, the findings are similar at all sites suggesting that the lessons have wider application. An audit of a flood control and drainage project at three sites in Bangladesh was included in the study so as to identify differences in O&M organization and effectiveness between irrigation and flood control. Map I (page 10) shows all nine scheme sites. 2. Field work was carried out in three phases in late 1994 and early 1995. An OED impact study team comprising Bank staff and intemational and local consultants visited farmers and officials at the scheme sites and responsible public irrigation authorities. The field work had a "participatory" orientation, as interactive group and household interviews were arranged in all four countries. The field work was also carried out rapidly: on average one and one-half weeks at each site. O&M Performance: the Current Paradigm 3. O&M performance by both agencies and irrigators on the large, govemment-operated, gravity-fed irrigation schemes in Southeast Asia is, with few exceptions, dismal. This conclusion confirms but goes beyond the frequent reports of degraded public infrastructure in developing countries, particularly irrigation structures. Measures to reverse the problem of "rusting, crumbling infrastructure" were central concems of the Bank's World Development Report 1994-Infrastructurefor Development. 4. Poor O&M and lowered benefits from irrigation investments are causally linked. OED's 1994 report A Review of World Bank Experience in Irrigation examined 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 pattems, that suggest irrigator resistance of unusual intensity to O&M design standards-degenerating into anarchy and chaos. 5. Social scientists have also paid increasing attention to the role of formal and informal 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 1. Thailand: Lam Pao scheme in the northeast, serving 50,000 ha; Maeklong right bank scheme, serving 40,000 ha. Myanmar: Kinda scheme in the central dry zone, serving 71,000 ha; Kinmundaung tank in the central dry zone, serving 1,000 ha; Azin tank in the south, serving 1,200 ha Vietnam: Dau Tieng scheme northwest of Ho Chi Minh City, serving 45,000 ha. 6 promote improved O&M. Elinor Ostrom's seminal publication Crafting Institutions for Self- Governing Irrigation offers eight "design principles." Other lists of "conditions" for success have been prepared by researchers inside and outside the Bank. In the vanguard is the International Irrigation Management Institute (IIMI), established in Colombo, Sri Lanka in 1984 with international funds and a mandate to study all aspects of irrigation organization and management. IIMI 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 for improved O&M from greater farmer participation, it is less optimistic in its assessments of public agency performance prospects. Deep seated cultural features of irrigation bureaucracies are widely held to inhibit effective O&M work. 6. In short, the dominant paradigm is one of incompetent bureaucracies combined with non-existent 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 potential. Inevitably, this paradigm guided the design of the impact study. Agro-economic Impacts of the Six Schemes 7. The predominant crop is paddy in all schemes, and there are many other similarities between them. The most important distinguishing factor is the degree of water abundance or scarcity at the level of the reservoir, compared with the area to be irrigated. Four of the schemes have more water than they use; the other two, in the central dry zone of Myanmar, have much less than planned. 8. At four of the six sites, including three of the large ones, the areas supplied by the irrigation systems are significantly less than planned. Over-optimism, engineering errors, lower than normal rainfall and failure to extend the tertiary canals are the leading explanations. They 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 too would show much lower average intensities than projected. Only one scheme-a successful tank in Myanmar-has reached 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. In fact, the concentration on paddy has increased. 9. Reestimates of total scheme production of paddy, and of a few other major crops at the two schemes where paddy is not completely dominant, show output to be between 32 percent 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 ha out of a total of 207,000 ha 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 1995 for the mid- and late-1990s were only 30 percent of the prices projected when these projects were appraised. The small upturn in late 1995 has little effect on the outcome. Together the production shortfalls and lower prices result in rates of return at or below 7 percent for all schemes and negative for one. 7 Welfare Impacts 10. Smallholder irrigated paddy can no longer provide the basis for a growing or even stable household economy. Incremental and total farmer financial incomes from average size holdings range from about US$600 to US$2,000. For Vietnam and Thailand actual incomes are only irrigated 10-30 percent of appraisal estimates. The gap is lower in Myanmar, but mostly because appraisal projections were less ambitious. The implications vary. For example the accelerating rate of outmigration from the two Thailand schemes contrasts with the stability of farm communities in Myanmar. As economies expand, irrigated paddy will not be able to compete with the incomes to be had from other employment opportunities. O&M Performance: in Practice 11. The OED field surveys concentrated on assessing the performance by agencies and irrigators in operating and maintaining the schemes. Observations were made of the condition of the canals and control structures, agency activity in allocating, distributing and maintaining the flow of water, and the strengths and weaknesses of farmer O&M, especially as managed through informal associations, their WUGs, and the few higher level federations of WUGs encountered. Overall, agency and irrigator performance appears substantially better than the image presented in the depressing paradigm described above and runs counter to allegations about farner disinterest in maintaining the irrigation assets that serve them, about feuds over water supplies tending to anarchy, and about an insurmountable bias among agency engineers against O&M. 12. Even in the two water-short schemes, and on the ends of the distributary canals in the other schemes where periodic water deficits appear in spite of overall scheme abundance, relations between headenders and tailenders are more civil, accommodating and fair than the paradigm suggests. Advantaged irrigators do use their advantages, and other irrigators do complain. But the level of agitation seems nowhere to be 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. 13. This civility, and relatively fair sharing of water, has been accomplished despite the fact the WUGs-which are present on all schemes-are not functioning up to expected standards. The one exception is the internationally-assisted sections of the scheme at Lam Pao, Thailand. There, both the WUGs and the federated groups of WUGs organized along some of the distributaries meet the criteria laid down in the Crafting book cited above 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 studies. 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 otherwise stimulate no larger group cohesion or participatory activity. In Myanmar the WUGs are subordinate to the village councils and also 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. 14. The contrast with the flood control schemes is instructive. In Bangladesh there was no attempt to form user groups associated with the flood control structures. The character of these 8 structures, and the benefits they provide, is 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 assets, 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 better fits the paradigm than irrigation O&M. O&M Influence on Agro-Economic Impacts 15. Do the lapses and failures of operation and maintenance that were everywhere observed contribute to lowering production on the five poorly perforning 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 which are essential to keep sufficient supplies of water flowing to the great majority of the fields are adequately carried out. Yet, it was the assumption that such a negative relationship did exist that prompted the study. In many other countries, and on many other schemes, the record is undoubtedly worse. But the study suggests that a more discriminating analysis of the O&M breakdown is warranted. 16. It is clear that some components of O&M are under control and others are not. Four parts of an O&M matrix are discussed: agency operations and agency maintenance, and irrigator operations and irrigator maintenance. Common weaknesses are agency inability to keep some of the larger distributaries clean of silt and weeds throughout the cropping season, and farmer indifference, neglect and destruction of tertiary gates. But these failures are not systemic and can usually be explained by budget constraints, scheduling problems and farm-level disincentives that require tailored and equally well "crafted" solutions. Other Issues 17. Among other issues covered by the study are: (1) the dismantling of complex technological control systems imported in the 1980s by foreign consultants. In Thailand and Myanmar a computerized water allocation and monitoring program (WASAM) was adopted, but it proved to be too demanding on agency staff and the protocol and measuring devices have been abandoned; (2) the ongoing attempt by a follow-up group of consultants in Thailand to "modernize" the control system by simplifying WASAM and substituting weirs and gates that require less human intervention. There are two types of modifications-fixed structures that have no adjustments, and structures that adjust automatically to changes in water levels-each with its own advantages; (3) farmer rejection of rotations as well as gates. Rotations do occur, but they tend to break down under conditions of shortage when they are needed most. The biggest problem is not the sharing of water within a tertiary system, but between tertiaries; (4) the ability of the agency in one country-supported by the local administration-to bypass rotations altogether by simply cutting off tailend tertiaries. The irrigation agency informs the farmers in advance which of them cannot be supplied. In other parts of the world, shortages usually are shared equally; and 9 (5) the different degrees of success in mobilizing free irrigator labor to clean the larger canals. In Myanmar the authorities are able to gather large numbers of irrigating and non- irrigating farmers. Group work on distributaries is rooted in Burmese irrigation traditions. Findings and Recommendations 18. The finding that dominates the study has little to do with O&M. Offering 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 uncompetitiveness of paddy farming drives the younger members off the farms, and the older members who stay behind concentrate on the basic subsistence crops, social capital will erode and O&M standards are likely to suffer. 19. Based on the six schemes studies, a dozen recommendations are made grouped into four sets: * Sharpen the response to O&Mfailures: by disaggregating O&M, identifying the poorly performing components, and dealing with disincentives specific to each, such as the tertiauy gates that farmers below consider unfriendly; * Simplify the technology of infrastructure and operations: by converting to fixed and automatic controls that need less human intervention and by supporting authorities who plan with the farmers to abandon equitable rotations by rationing water in emergencies; - Promote the transfer of management tofarmers and their WUGs, judiciously: recognizing that organizing user groups pays off, but accepting as well that there are some management responsibilities which immature WUGs cannot handle; and - Improve household earnings: through diversifying cropping systems and supporting research, extension and marketing services keyed to specialty crops and integrated, high value farming. 20. The relevance of these recommendations beyond the selected schemes is uncertain, since they depend on cultural and institutional parameters which may be country-specific, and on engineering and agronomic considerations which may be project-specific. For example, comments on a draft of this report showed concern that the findings were at once both too rough and too forgiving on 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 inside or outside the region for which these recommendations are appropriate. OED has proposed a regional workshop where validation would be one of the principal objectives. 10 Map l IBRD 27S61 SOUTH/SOUTHEAST ASIA OED IMPACT STUDY: IRRIGATION PROJECTS AND IRRIGATION 0 & M KIGncda COMPLETED PROJECTS UNDER SrUDY O NATIONAL CAPITALS _x:_ MAJOR RIVERS INTERNATIONAL BOUNDARIES 0 5W 1000 KILOMETERS The boundaries, colors, deunominaions and any oeJer inlannauian shown an this map dof not inmIy, an thA pat of The Word Bank Group, any judgment an dh. IP gat uu Of any twritowy, or any enorsemnt or acceptance of such bouncaris. - 'Chal--?S l Iyf '/ ffi .g*HrI Hoor C c H I A INDIA lNMAR' ( -~ | C~ EK ( 4p Annex B IRRIGATION O&M AND SYSTEM PERFORMANCE IN SOUTHEAST ASIA: AN OED IMPACT STUDY REVIEW OF THE KINDA AND TANK IRRIGATION PROJECT SCHEMES MYANMAR June 27, 1996 Operations Evaluation Department Abbreviations and Acronyms ASC Agriculture Supervisory Committee Bank World Bank DY Distributary ERR Economic Rate of Return GOM Government of Myanmar ID Irrigation Departmnent LMC Left Main Canal LORC Law and Order Restoration Council MAS Myanma Agricultural Service O&M Operation and Maintenance OED Operations Evaluation Department PAR Performance Audit Report PCR Project Completion Report RMC Right Main Canal SAR Staff Appraisal Report SLORC State Law and Order Restoration Council VTLORC Village Tract Law and Order Restoration Council WASAM Water Allocation Scheduling and Monitoring Program WUG Water User Group Contents 1. Introduction A. Background ............................3 B. Preparation ............................3 C. Implementation ............................4 D. Characteristics of the Command Areas ............................5 2. Kinda Scheme: Structure and Efficiency A. Structure of the Irrigation Scheme ........................7 B. Water Availability and Efficiency .........................8 3. Kinda Scheme: Operations and Maintenance A. Institutional Arrangements for O&M . . ........................ 10 1. Agency Level .......................... 10 2. Irrigator Level .......................... 13 B. Operational Performance ... ....................... 15 C. Maintenance Performance . . .8....................... I 1. Maintenance by the Agency ..................... . 18 2. Maintenance by the Irrigators ..................... 19 4. Kinda Scheme: Agro-Economic Impacts A. Agricultural Impact ....................... 21 B. Financial Impact, Farmers' Level ....................... 25 C. Economic Impact ....................... 26 5. Tanks Schemes A. Structure of the Irrigation Schemes . . .28 B. Operation and Maintenance . . . 28 1. Kinmundaung .28 2. Azin .30 C. Agro-Economic Impacts . . . 31 1. Kinmundaung .31 2. Azin .32 6. Conclusions: Influence of O&M Performance on Agro-Economic Impacts. 34 7. Other Conclusions .37 This report was prepared by Edward B. Rice (Task Manager), with support from Robert Yoder, Jayantha Perera, Annemarie Brolsma, Sinee Chuangcham, U Hia Myint (Consultants) and U Tun Naing (UNDP), who visited these projects in November 1994 and April and June 1995. Afi Zormelo and Megan Kimball provided administrative support. Tables 4.1. Kinda: Seasonal Cropping Pattern and Intensity .................................... 23 4.2. Kinda: Incremental Crop Production Estimates .................................... 25 Annexes I . Kinda Irrigation System Seasonal and Annual Efficiency .41 2. Kinda: Crop Area, Yield and Production Estimates .42 3. Kinda: Farm Income Calculations .44 4. Kinda: Economic Rate of Return Calculations .45 Maps 5. Kinda (Nyaunggyat) Dam Multipurpose Project (IBRD 27564) .end 6. Tank Irrigation Project, Kinmundaung Subproject (IBRD 27565) .end 7. Tank Irrigation Project, Azin Subproject (IBRD 27566) .end B 3 1. Introduction A. Background 1.1 The Kinda and Tank irrigation projects in Myanmar were selected as part of OED's impact study of gravity-fed irrigation projects in the humid tropics of Southeast Asia. Other gravity-fed schemes, also with storage, were selected in Thailand and Vietnam. A flood control and drainage project in Bangladesh was added to the regional study. to widen the thematic perspective on problems of operation and maintenance (O&M). 1.2 Credit 1031 -BA for the KindaI Dam Multipurpose Project was approved for US$90 million in May 1980, made effective in November 1980, and closed in March 1991 with a cancellation of US$3.04 million. A Project Completion Report (PCR) was issued in June 1992,2 covering both the power and irrigation components. OED has not audited the project, and this impact analysis refers only to the irrigation component. Credit 1315-BA for the Tank Irrigation Project-supporting the Kinrnundaung and Azin schemes-was approved for US$19 million equivalent in December 1982, made effective in April 1983 and closed in June 1990 with cancellation of US$3.00 million equivalent. A PCR was issued in October 1991.3 A Project Performance Audit Report (PAR) has been prepared in parallel with this impact study. 1.3 The primary subject of the impact study was the Kinda irrigation scheme. The report focuses first on O&M performance, second on agro-economic impacts, and third on the influence of O&M on those impacts. The discussion of O&M and impacts of the two tank subprojects, at Kinrundaung and Azin, are shorter and concerned mostly with similarities and differences compared to Kinda. B. Preparation 1.4 In the mid-I 970s the Government of Myanmar (GOM) initiated a program for extending irrigation through storage of wet season stream flows. Its objectives were twofold: to protect paddy production from the vagaries of monsoon rainfall in the central dry zone, and to develop a second paddy crop and promote nonpaddy crops there and in the southern high-rainfall coastal region. In the same period, and under the aegis of the Second and Third Four Year Plans (1975- 78, 1978-82), a series of agricultural reforms were implemented which included increased investment in agriculture, an intensified agricultural extension program, improved availability of inputs, and increased rice procurement prices. 1.5 The Kinda and Tank Projects were the Bank's fourth and fifth irrigation operations in Myanmar. The first three had concentrated on flood control, drainage and small-scale pumping 1. The project was appraised as the Nyaunggyat Darn Irrigation Project. Government requested the name be changed in 1981, a year after the credit was approved. 2. Agriculture Operations Division, Country Department 1, East Asia & Pacific Regional Office. Report No. 10698, June 8, 1992. 3. Agriculture Operations Division, Country Departnent 11, Asia Regional Office. Report No. 9995, October 11, 1991. B4 in the paddyfields and uplands of the delta of "Lower Burma."4 In the mid- 1970s Bank and FAO/CP missions began to assess the potential for: 1) rehabilitation of existing run-of-the-river diversion schemes, 2) large-scale storage and infrastructure improvements on other older schemes, and 3) medium-scale storage ("tanks") at both old and new sites-all three options in the central dry zone of "Upper Burma." The Kinda project was identified in 1974, and preparation was financed by funds from the Bank's first irrigation project. Preparation of the Tank project started four years later at several proposed sites in the central dry zone. Another cluster of sites on the southern coast of "Lower Burma" were incorporated in the analysis in 1980. The Tank Project as finally approved included one tank each from the dry zone (Kinmundaung) and coast (Azin). It was intended to be the first in a series of multiple-tank projects supported by the Bank in a time-slice progression. Tank II was appraised in 1988. But changes in political conditions in Myanmar later that year brought the processing of all further Bank projects to a halt, and no other irrigation projects have been or are currently being considered. GOM, meanwhile, has in the last several years accelerated the construction of new tanks using domestic funds, and is currently completing another major dam on the Zawgyi River adjacent to the Kinda scheme. End-Maps 5-7 cover the three individual schemes. C. Implementation 1.6 The Kinda Project was implemented by the Irrigation Department (ID) of the Ministry of Agriculture and Forest. ID was specifically responsible for construction of the dam, rehabilitation of the old "Panlaung" run-of-the-river diversion system serving 88,000 acres (36,000 ha),5 and construction of a new irrigation scheme serving 113,500 ac (46,000 ha) adjacent to the existing system on higher ground. The Myanma Electric Power Enterprise was responsible for construction of the powerhouse at the dam and a 50 km transmission line to the national grid. The Myanma Agriculture Service (MAS) was responsible for strengthening the extension service, constructing training facilities for irrigators, and establishing a cotton research farmn. 1.7 Works were started at Kinda in 1980. Excellent progress was made in the implementation of the dams, appurtenances and power generation facilities. The first release of water from storage was made in August 1985, and power reached the national grid on schedule in January 1986. The irrigation component of the project was delayed and completed only in 1991, over four years behind schedule. The principal factors explaining the delay were shortages of diesel fuel, reinforced steel and cement. Although irrigation water reached the first pilot activities in 1985, full supply up to the reservoir's capacity for the right and left main canals was not available until 1992. The PCR, issued that year, echoes the concern of the latest supervision reports that MAS' extension services had also fallen behind schedule, and the irrigators and their associations were not yet prepared to manage water efficiently. 1.8 The Tank Project was also implemented by ID. A small water supply component in the Azin subproject for a local town, Mudon, was implemented by the Construction Corporation. Works started simultaneously at Kinmundaung and Azin in 1984. As the shortage of diesel and 4. The first of the three, the Irrigation Project (Cr 483-BA), financed about 13,000 pumps, of which 2,760 were installed in the central dry zone and the rest in Lower Burma 5. Acres, rather than hectares, are used in this text when referring to scheme size and scheme-wise cropping pattems. This allows direct comparisons with GOM and Bank estimates of these key variables in the analysis. However, most references to farm size are metric, and all other measures, e.g. distances, are also metric. B 5 construction material worsened, ID concentrated on Kinmundaung and was able to make the first release, from an unfinished dam, on schedule in 1986. The first release from the Azin dam was made in 1989, three years later than planned. In this case, MAS was better prepared to assist the farmers. 1.9 For more complete discussions of implementation experience, the reader is referred to the two PCRs and the new PAR on the Tank Project. D. Characteristics of the Command Areas 1.10 Kinda. The old Panlaung scheme and its eastern neighbor, the Zawgyi River scheme, both in Mandalay Division, together constitute the ancient Kyauske system, the largest of the run-of-the-river irrigation-based granaries of the kingdoms of Upper Burma for over a thousand years. Both rivers originate in the Shan hills to the east, and provide a perennial source of water to farms along the rivers and diversion canals before they empty into the Irrawaddy River near Mandalay. There were five old weirs along the Panlaung River, diverting water mostly to low lying paddylands on either side. About 20 percent of the irrigable 88,000 ac in the Panlaung scheme is higher land ("ya-land"), better drained and suitable for a mixture of crops. 1.11 The British rehabilitated these and other systems in Upper Burma, but generally left the canal alignments and basic structures intact in recognition of their high technical and engineering standards. The Bank-supported Kinda project also did not modify in any significant way the basic conveyance and distribution structures of the existing system. 1.12 In olden times the irrigated farmlands belonged to the kings, and the farmers were serfs who followed cropping systems under instruction of the kings' agents. In more recent times the farmers were freeholders, until, in 1964, a new socialist regime assumed ownership of all agricultural land. The farmers were converted to tenants. But they consider their rights inalienable and an active, informal, and officially illegal land market exists. 1.13 The new system established a parallel canal alignment to the west that would command a strip of 107,000 ha of mostly ya-land. Only 15 percent of the land under the new command consisted of the heavy soils of fine texture, with imperfect drainage, that is suited to paddy but not other crops. None of this ya-land-light soils of coarse texture-had previously been irrigated. Except for about 4,000 ac of bush, the new area was cultivated with paddy and other rainfed crops during the monsoon season. The northern half of the strip was the driest and much of it abandoned in years of poor-to-modest rainfall. 1.14 The project area is about 80 km from north to south and 30 km east to west. It has a tropical monsoon climate with three distinct seasons: a rainy season from mid-May to mid- October, a cool dry season from mid-October to mid-February, and a hot dry season from mid- February to mid-May. There is practically no rain from December through March. The rainfall gradient is sharp, with much more rain in the hills behind the dam site than in the command area out on the plain. This is the edge of the central dry zone. Average rainfall in the command area is only about 800 mm; at the dam site it is just over 1,000 mm. Farms without access to the canals can only harvest a monsoon crop with reasonable, but far from absolute, security. Non- 6. At appraisal, the project's total net command area on both banks was estimated to comprise 85,400 ac of paddyland and 1 16,100 ac of ya-land. About 7,000 ac were unavailable for irrigation. B 6 paddy off-season rainfed cropping-based on early pre-monsoon rains and on the post-monsoon residual moisture-is a precarious practice. The old Panlaung diversion system helped secure the monsoon crop, and substantially improved the safety of a second crop in about 30 percent of the existing canal network. Pre-project annual cropping intensity in the old system was estimated at appraisal at 125 percent. 1.15 Kinmundaung. This site of 5,000 ac was selected for the first Tank project from a cluster of four sites in near proximity in the Yin basin of Magwe Division. The site is in the middle of the central dry zone, where, as in Kinda, low and unreliable rainfall confines unirrigated agriculture primarily to one monsoon crop, plus precarious efforts to bring crops to harvest during the pre- and post-monsoon periods. Without supplementary irrigation a paddy crop during the monsoon is even more risky than at Kinda, because of frequent failure of the late rains. Sesame, groundnuts, and other oil-seeds (Magwe Division is the center of oilseed production in the country), and other crops with lower moisture requirements than paddy, perform well in the monsoon season, and reasonably well in the pre-monsoon and post-monsoon seasons in about half the years. Without irrigation, cropping is impossible in the four, dry months December through March. 1.16 The area of the Yin basin is favored with good soils and a topography that provides occasional sites for storage on rivers-of relatively limited catchnrient area-running westward to the Yin River and on to the Irrawaddy. However in the mid-1970s there was no storage capacity in the Yin Basin and few run-of-the-river schemes of any significance. Kinmundaung was one of them. British engineers had rebuilt the earlier irrigation works on that river in 1911. But in the intervening years the canal had silted up, the infrastructure had deteriorated, and by 1980 only about 1,000 ac were still receiving supplementary wet season irrigation with any regularity.7 Thus, a large majority of the farmers to be incorporated in the 5,000 ac scheme had no previous experience with gravity flow irrigation, apart from watering their gardens from shallow dug wells. 1.17 Azin. The Azin site of 2,850 ac was one of a cluster of ten potential tank sites strung out in a 20 km line north and south of the town of Mudon in Mon State. The area comprises a corridor of land bounded by the Taungnyo hills to the east and the Salween River estuary to the west. The area is divided into two principal land forms: the piedmont fan, and the esturine flood plain. The area grows mostly rainfed paddy on the flood plain, while the piedmont fan is cultivated with rubber (mostly government estates), and private orchards in small gardens irrigated from shallow dug wells. Stored water can be used to support either or both the perennial tree crops or the paddy on the plains. This was the first tank built by ID on the southern coast. 1.18 The rainfall regime is radically different from the central dry zone. Monsoon rainfall was adequate to sustain paddy without supplementary irrigation. Yet the Azin River was ephemeral like the rest of these streams-dry in the dry season. The storage was intended to support a second paddy crop in this "summer" season. Initial plans did not include irrigation for orchards in the piedmont. Later, 850 ac of orchards were added to 2,000 ac of paddy to enhance the economic returns. 7. Despite rehabilitation by ID in 1961. B 7 2. Kinda Scheme: Structure and Efficiency A. Structure of the Irrigation Scheme 2.1 The Kinda Reservoir is formed by a 236 ft high rockfill dam on the Panlaung River and three smaller rockfill dams straddling saddles in hills ringing the reservoir.8 It is fed from a catchment of about 2,240 km2. Water is released from the reservoir through the hydroelectric plant. From the dam, water travels about 12 km down the Panlaung River bed to the old Kinda Weir site. The Kinda Weir was rebuilt as a regulating dam with 1,800 ac-ft of storage. The regulating dam releases water directly to the old right bank Kinda Canal, now called Right Bank Main Canal (RMC) and the new Left Bank Main Canal (LMC). It also controls releases down the riverbed to be picked up by the other 3 functioning weirs on the river. Map 5 identifies the main structures. 2.2 The canal network terminology used in the rest of this report conforms to local usage. It starts with the "main" canals.9 The distributaries off the mains are called just that, "distributaries" (DY), and these either10 feed a "tertiary" distribution line that feeds "watercourses," or, as in much of the old system, feed the watercourses directly. The watercourses feed the "ditches" that lead to individual fields. On the Kinda LMC, as at the other new sites in this study (Kinmundaung and Azin), the ditches were supposed to reach every field. But they were the responsibility of the farmers, and have not been completed on many tailend watercourses. The farms on the fringe depend on field-to-field flooding. In Kinda RMC field- to-field flooding had been more common and the project authorities did not intervene except in a pilot area (para 3.1 1). 2.3 The 125 km long LMC canal terminates in the Pyugan Tank.1' This tank is reported to have been built about the same time as the original Kinda Weir in the eleventh century. The tank serves 200 to 500 ac of paddy land reaching to the banks of the Irrawaddy River. The LMC follows the contour from the regulating weir to the tank. To reach it, which appears to have been one of the objectives of the design engineers, ID had to accept a lower gradient than it would otherwise have chosen. The low gradient made the conveyance capacity of the LMC sensitive to excess water use and weed growth in its upper reaches, among the factors that explain the shortage of water for distributaries close to the tank. Including the Pyugan Tank, the LMC feeds 35 distributary canals. The tank feeds another three distributaries. 2.4 The distributary canals run down the slope approximately perpendicular to the LMC. Many tertiary canal offtakes are incorporated into head regulator/drop structures. The tertiary canals run on the contour perpendicular to the distributary canals. They are equipped with 8. Storage at the maximum operating level is 786,000 ac-ft. The active storage area is 619,000 ac-ft. The surface area of the reservoir is 36 km2 at maximum and about 16 km2 at minimum operating elevations. The main spillway has four gates which, together with an emei-ger.- rjvmn e'nit spillway 180 m long on the left rim of the reservoir, are designed for safe passage of the inaximtiui, problaiie iv Joo Id 1 ing under full reservoir conditions. 9. And. in the case of the RlCv n n 'or2nor- n 10. In Kinmundaung two old distributaries are called "minors." In Azin there are "minors" in between the distributaries and the tertiaries. I1. About 80 km of the LMC is constructed of compacted earth and the remainder is lined with bricks to reduce seepage. B 8 concrete structures designed for proportional delivery of water to watercourses, which run down the slope perpendicular to the tertiaries. The grid is adjusted to irregularities in ground level. It is more reticulated than in the old system, partly because of the higher elevation of the main canal. The turnouts from the mains to the distributaries and from them to the tertiaries are all fitted with undershot gates, at least on the LMC. The turnouts to the watercourses are made from concrete with grooves for stop logs, though most of these are missing. On the Panlaung system, there are many instances where the older openings are ungated. 2.5 Few changes in main canal infrastructure were made by the project in the Panlaung system, including the RMC. None of the canals in this system are lined. The RMC has a steeper gradient than the LMC. Continuous operation for hundreds of years has proven the sustainability of these canals. The Kinda weir remains the upper-most diversion. Until the Kinda Project it only diverted water into the right bank Kinda Canal. But now, about eight km from the weir the RMC bifurcates into the Pyaungbya and Ngalaingzin (ex-Kinda) canals. (Map 5 shows details.) Drainage from the last three distributaries from the Ngalaingzin Canal as well as water from the Zawgyi River irrigation system to the east is picked up by a common drain. The Panlaung River portion of the Kyaukse system was fully incorporated into the new Kinda Project. There are three more weirs. The last, the Kyime Weir, is a major structure and diverts water to the left bank of the Panlaung River into the Sama Canal. The Sama Canal irrigates a strip of land about 38 km long. 2.6 The Panlaung system irrigates 88,000 ac. The LMC was planned to supply 107,000 ac. The combined irrigable area as designed was 195,000 ac. As explained below, the inflow to the reservoir has been 80 percent of the projected supply in the nine years since the dam was completed. The last sections of the LMC are rarely supplied. Thus the actual command is closer to 175,000 ac. At appraisal there were about 14,400 farms in the RMC/Panlaung system, and 9,000 farms in the LMC system (as designed). B. Water Availability and Efficiency 2.7 During the period of project preparation, water availability was estimated using all available hydrologic and meteorologic records. It was concluded that the average annual inflow to the reservoir would be about 1.1 million acre-feet (ac-ft). From 1986 through 1994 the average inflow was about 0.9 million ac-ft, or 79 percent of the expected average. In this nine year period the expected average annual inflow was reached only in 1992. The average release for the nine year period was about 850,000 ac-ft, roughly 78 percent of the planned average irrigation. It is still too early to conclude to what degree the expected inflow was overestimated, since rainfall was below average in the reservoir watershed during this period. However, ID officials believe that it was overestimated, though not by the full margin suggested by the first nine years of records. Inflows in the last four years averaged about 89 percent of the expected, long run average. 2.8 The annual inflow and release figures mask an evident problem of timing of releases for optimum water use for irrigation. ID indicated that hydropower production was given priority in the winter months, when the demand for irrigation was less than the water released. This contributed to water shortage during periods of higher irrigation demand. ID staff are skeptical that that priority will ever be reversed, especially in the face of electricity shortage. B39 2.9 OED has computed seasonal and annual irrigation system efficiency for the period 1990- 1994 (Annex 1). Average efficiency over the whole period was 52 percent, which combines a rate of 69 percent for the pre-monsoon, 51 percent for the monsoon, and 10 percent for the winter seasons (weighted by water releases).'2 While efficiency for the winter crops was low due to hydropower releases, the monsoon season efficiency was good and the pre-monsoon efficiency very high. 2.10 High efficiency levels might indicate that reservoir water is being used not only in the fields of first release but also lower down. In fact the rate of recapture of water is itself low, reflecting on the one hand the substantial depth of the water table in this dryland region of Myanmar, and on the other hand the lack of experience of farmers with conjunctive use of groundwater. There are some shallow wells near the rivers, but the OED team saw no evidence that farmers were pumping water for agricultural purposes. 12. While indicative of the real efficiency level, this analysis has some flaws. Complete meteorological data for each day was not available for the entire period. The crop water requirement was computed using ID's values for Mandalay Division, prepared from long-term monthly average evapotranspiration and rainfall records, and does not reflect the influence of seasonal evapotranspiration and rainfall differences or conditions specific to the Panlaung plain. B 10 3. Kinda Scheme: Operation and Maintenance A. Institutional Arrangements for O&M (1) Agency Level 3.1 The Cropping Plan. Irrigation systems in Myanmar are unique for the study area in the level of crop planning and control over crops planted. Targets reflecting national priorities are set by the Central Government and distributed according to agricultural potential among the State (or Division13) Agriculture Supervisory Committees (ASC). The State ASC use production statistics available at the State level to distribute production targets among District ASC for implementation. A similar process takes place at the District to provide targets for each Township. 3.2 At the Township level the ASC must deal directly with the farmers and decides for each irrigator how many acres of each crop he or she is to cultivate. In addition to distributing acreage to fulfill national targets, this decision is based on the ability of the irrigation officials to deliver water. To facilitate the process of meeting with farmers, Townships are divided into approximately four regions called Circles, each encompassing several Village Tracts (a rural area with one or more villages). An interim Circle ASC committee meets with farmers in the region to discuss and provisionally allocate the targets to each individual farm. 3.3 The present government of Myanmar is the State Law and Order Restoration Council (SLORC). It is represented at the Township and Village Tract levels by subordinate councils, which exercise authority within those jurisdictions. This structure is parallel to but fully coordinated with the ASC structure, and helps plan, approve, supervise and enforce the ASC decisions on water rights and planting obligations. The chairman of the Village Tract Council (VTLORC) chairman is a key figure in implementing the cropping plan. 3.4 Currently, depending on the location of fields relative to the irrigation supply, farmers in the Kinda Project are assigned to grow either paddy, cotton, and/or sesame on most of the land they farm. They are free to select and declare the crop of their choice on the one or two acres that are not assigned under the targets. Generally the ASC tries to accommodate the mix of crops requested by farmers but flexibility is constrained by target crops. Planning at the Circle level allows some adjustment of crop assignments to match soil characteristics. 3.5 Farmers must cooperate with the State Plan. If farmers disagree with their assignments, the ASC can transfer the assignment to other farmers and proportionally reduce the irrigation allocated for those who do not agree to fulfill the plan. Recalcitrant farmers may be penalized by the Township or VTLORCs. 3.6 Management ofAllocations and Deficits. Water allocation initially is directly tied to crop assignments and the water is distributed solely on the basis of estimated irrigation requirements of the different crops. As the season unfolds, one of the important tasks of the ID 13. The administrative units below the nation-state are called Divisions and States, the former occupying the Burmese heartland, the States representing mostly minority groups. The word State in this text can mean either. B 11 representatives on the ASC committees is to match crop water demand to progressive changes in reservoir supply. The reservoir level and average annual inflow information for the months of the cropping season are the basis for recomputing expected supply. The field level staff of ID, MAS, VTLORC and other institutions continuously supervise agricultural work at the farm-level on behalf of the ASC. They check whether the farmers cultivate their prescribed crops and are getting adequate irrigation supplies. When shortages are anticipated they help decide which farms, watercourses or entire tertiaries will be cut off. 3.7 The State has full authority to modify the allocation. Farmer rights in irrigation systems in many parts of the world mandate that even a dwindling irrigation supply be shared equitably, even to the point that inefficiencies increase and most crops are lost. In Myanmar the State or District MAS can reduce the area irrigated, even after crops have been planted, to make efficient use of the available irrigation resource. While this enables production to be maximized, some farmers may lose all their irrigated crops. 3.8 Penalties. The penalties as prescribed by law for farmers not meeting their production target are severe and include possible loss of land-use rights or even imprisonment for willfully disregarding the crop plan. However, if production fails because ID could not deliver the allocated quantity of irrigation at the right time, farmers can appeal for exemption from payment of the designated tax on production (the procurement quota, see para 3.23). They cannot appeal for compensation for crops lost. Extreme sanctions are usually avoided in the absence of a legitimate appeal, unless the recalcitrant is a repeater. Nevertheless the frequency of punishments and even jailings obviously reduce the incidence of abuse and conflict. 3.9 Operational Procedures Introduced by the Project. Project design documents show that the expected annual irrigation supply under nornal rainfall conditions would generally not be sufficient for crops of rice and cotton over the entire cultivable area. The targeted annual cropping intensity for the whole of the Kinda scheme was 160 percent. To make best use of the available water the Kinda authorities, through international technical assistance, introduced a modern water management system for use by ID. 3.10 A computer program called Water Allocation, Scheduling and Monitoring (WASAM) was installed and calibrated, and staff trained in its use by the end of 1987. The objective of the WASAM program was to: 1) allocate the irrigation supply by calculating the weekly required discharge in each canal section of the irrigation system-through the tertiary turnout; 2) prepare a report of expected weekly total water requirements of each part of the Kinda system for operators to follow in adjusting release of water from the reservoir and distribution among the canals; and 3) monitor the actual water distribution and compare it with planned distribution. Since the appraisal and technical assistance teams had not anticipated a shortfall of water, WASAM was not designed as a mechanism to reallocate supplies in times of scarcity. The information it provided could be used to facilitate those decisions. 3.11 WASAM was introduced throughout the LMC, but the consultants also sponsored an intensive pilot operation in 20,000 ac below the RMC. New tertiary canals with gated control offtakes were introduced, running along the distributary canal, to which three or four watercourses were connected. These watercourses had no controlled offtakes before rehabilitation and had taken off directly from the distributary. B 12 3.12 In the consultants' final report (1987) at the end of 13 months of assistance they concluded that the foundation had been laid for "modern water management" at the scheme. Work accomplished included establishment of a Water Operation Center and installation of a computer and WASAM software. The necessary calibration of structures for discharge measurement was well underway with several teams trained in calibration procedures. However, the report concluded that further development and training was necessary to realize the full benefit of the program and achieve "equitable" distribution of the limited irrigation supply as efficiently as possible, and to achieve the cropping intensity of 160 percent as planned. 3.13 Main, Distributaryand Tertiary Operation. The responsibility for system operation and infrastructure maintenance down to the tertiary level rests with ID. GOM through budget allocations to ID's Mandalay Division provides funds for O&M. The Executive Engineer of the Kinda System is in overall charge. Assistant Engineers in each township supervise the irrigation distribution from the main canal to distributary and tertiary canals. Under each Assistant Engineer are from one to three Canal Inspectors who with their staff of assistant inspectors and Zonemen are resident in the command area. Each Zoneman is assisted by up to 5 assistants. At Kinda, most of the positions are occupied, and by experienced personnel. There are few vacancies. 3.14 Rotations. Prior to construction of the Kinda Dam there was generally simultaneous delivery of water in the principal Kinda (RMC) and Panlaung canals. Only during periods of drought did they rotate delivery between these canals. Without a reservoir for storage, water was lost to the scheme if not used, so whenever possible during the rainy season these canals were operated at maximum discharge. This allowed much of the command area of the canals to receive water continuously. As supply diminished between rainfall events, rotation was first practiced within these canals and then among them only if necessary. 3.15 The Kinda Irrigation Project system, including the LMC command, was also designed for continuous water supply to the main canals and distributaries. It was planned that tertiary canals should always be operated close to full discharge to enable equitable distribution among the watercourses within the tertiary. To achieve full discharge in the tertiary, rotation was planned among tertiary outlets of the distributary. For the RMC/Panlaung system, where there were few tertiaries apart from the pilot area, continuous water delivery was to be maintained. The pilot area was to experiment with finely calibrated releases at all offtakes. 3.16 Maintenance. Except for a small number of dump trucks and a backhoe, there is virtually no mechanical equipment used in system maintenance. The work is labor intensive. ID has a pool of permanent laborers and their supervisors. They generally live near their area of assignment. They are assigned to patrol a segment of canal and remove debris, silt, and weed from the canal and banks. In some segments of the various canals this labor is sufficient to maintain the canal in good condition. In other segments it is impossible for such a small labor force to keep up with problems and periodically larger labor gangs of farmers are hired to assist. Until the early 1 990s ID carried out all the maintenance by labor gangs on daily wages, which resulted in very high costs. 3.17 Faced with budget cuts in 1992, GOM required farmers to provide mandatory labor for all earth work and routine maintenance including cleaning. The cut in funding resulted in little money to pay labor gangs, and the need to increase mass mobilization of labor to carry out O&M above the watercourses. B 13 3.18 While ID determines the number of laborers and organizes and supervises the work on site, it is primarily the staff of the Township LORC members that organize villagers to work on the main canal. These mass mobilizations generally require all adult males from specified villages to attend work at a site determined by the ID staff. The length of canal or amount of cleaning is frequently assigned to villages in proportion to the number of adult males in the village. Landless laborers and villages without access to land irrigated by the canals are included in the mobilization of labor. The rationale is that they get some benefit from irrigation. 3.19 The Village Tract LORC organizes cleaning and repairs of the distributary canals. Again, all farmers who live in the Village Tracts which are served by these canals are expected to take part whether they operate irrigated land or not. 3.20 In short, whereas ID is technically responsible for canal maintenance down to the watercourse level, farmers are now asked to assist in most parts of the system. Corvee labor has a long tradition in Myanmar and, at least on these irrigation schemes where community benefits are visable, does not raise objection. (2) Irrigator Level 3.21 Responsibility. Farmers are responsible for construction of all channels below the tertiary canal, and maintenance of the tertiaries as well as the watercourses and ditches. In much of the previously irrigated area, where the watercourses connect directly to the main or distributary canals, farmers do all maintenance up to the main canal. Tertiary and watercourse maintenance is a collective activity. Villagers who do not irrigate are not obliged to participate at these levels. Segments of watercourses are generally assigned to individual farmers for cleaning in proportion to the farmer's landholding in the watercourse command area. Though the "proportionality" standard for water sharing often collapses under stress of water scarcity (para 3.36), for maintenance it is preserved. A typical irrigator may spend 9 days a year cleaning canals as a member of a group (apart from his ditches): two days each on watercourses and tertiaries and five days on the distributaries and main canal. 3.22 If a farmer cannot attend to watercourse maintenance work he is responsible for sending a replacement. Any man over 18 years is acceptable. A woman is not acceptable as a substitute as "women cannot do heavy work like maintenance of canals." Failure to contribute labor makes farmers in most areas of the system liable to a fine. The current rate is 100 kyat14 a day, about twice the daily wage. Irrigators typically do not pay cash to their watercourse group for O&M, apart from a contribution for food and festivities on corvee days. There are exceptions to the system. One group of LMC farmers explained to OED that their watercourse leader collects money from each irrigator and hires labor to clean the watercourse. 3.23 Cost Recovery. Irrigators pay a flat annual fee of 10 kyat (US$0. 10) per acre. This is in addition to a land tax of 2.5 kyat per acre that all farmers pay. These fees have not been adjusted for years and obviously cover practically none of government's O&M expenses. However, government raises substantially more than the O&M outlay from the irrigators through Myanmar's compulsory crop purchase program, and the Bank has considered that mechanism an acceptable though imperfect alternative. Quotas are established for each crop-irrigated and 14. About US$ I at current (street) rates. B 14 rainfed-based on area cultivated. The quotas have been reduced since 1988 as part of the liberalization program. Currently, irrigators must sell 12 "baskets" per acre of paddy to government at the compulsory price, while rainfed paddy growers must sell 5 baskets (a basket of paddy weighs about 21 kg). For the average Kinda irrigator, the formula means he/she currently sells about 17 percent of his crop to government at a controlled price that now amounts to about 25 percent of the free market price. He/she can sell the rest of his surplus above subsistence in the marketplace. S In the past the quota was much more onerous. 16 Of course, the low, flat annual water fee also encourages planting paddy over other crops. 3.24 Leadership. Irrigation management at the watercourse level is done by leaders selected by the farmers. The nominee must be approved by the Township LORC. ID has a rule-and usually farmers try-to select a person with land in the tail-end of the watercourse, to encourage fair water distribution among farmers. The leader's main duty is to interact with ID staff to ensure sufficient water for farmers of the watercourse to cultivate crops according to the crop plan and to oversee fair distribution of water to each cultivated holding. If there is conflict the leader contacts the Assistant Canal Inspector of the area for assistance. In parts of the old system, watercourses are longer and divided into one or more areas with separate leaders for each. In the LMC the watercourses are uniformly small by design. 3.25 The leader generally manages watercourse cleaning, though there are cases where Village Tract officers organize and supervise maintenance. In general leaders are exempt from physical work on irrigation canals. They plan and direct the work and attend to food and drinks. 3.26 In many watercourses there are few if any meetings where farmers contribute to group management of O&M activities. Since maintenance primarily involves labor, no financial accounts are kept and there is no need for a secretary/accountant. Again there are exceptions, and several farmers described meetings where planning was carried out as a group and decisions were made by consensus of the watercourse members. But, as a rule, the "group" concept is poorly developed. 3.27 While there is some degree of self-governance at the watercourse level and in a few cases even at the tertiary canal level, organizing specifically for irrigation management has not been encouraged until very recently. The Village Tract authorities, who may or may not be irrigators, are often responsible for helping to mobilize resources for maintenance and to manage conflicts. From the sample of farmers interviewed it was clear, as one would expect from the long history of sustained irrigation, that the previously irrigated areas have considerably more local organization and governance capability than in the new areas of the LMC. 3.28 In the LMC the tertiary canals have a leader selected by the watercourse leaders and ratified by the Township LORC. His responsibilities are similar to the watercourse leader except at one higher level of association. He organizes with the watercourse leaders the tertiary's 15. In principle, the quota applies only to the monsoon crop. Marketing of the new pre-monsoon irrigated paddy crop is not controlled (although planting is!). However, Government can apply quotas to the off-season crops if it needs to do so to meet national production targets. Actual practice may differ from state to state. 16. D. Vermillion of the Intemational Irrigation Management Instate (IIMI) comments: "In many ways state intervention in irrigated agriculture in Myanmar resembles that in China. China also uses mass mobilization of compulsory labor for irrigation maintenance and until recently still required a percentage of the crop to be sold to govenmment purchasing boards.' (Personal correspondence, January 19, 1996). B 15 maintenance and mobilizes the farmers on the tertiarty to clean it each season. In the Sama and RMC there are also cases where leaders are selected to represent more than one watercourse to direct maintenance and oversee irrigation distribution for the entire area. B. Operational Performance 3.29 The Pilot. The pilot project for improving irrigation access by constructing tertiary canals in the RMC met considerable farmer resistance. The pilot activity was to demonstrate improved irrigation distribution efficiency and equity by allowing better control over the water. Many farmers, however, were not open to the changes. Farmers who lost precious land to new canals that crossed their farms resisted the move. More important, the new canal configuration required changes in distribution practices that conflicted with the established traditions, however inequitable, and caused sufficient conflict that the pilot arrangement was not repeated. 3.30 WASAM Subsequent to the consultants' work in installing the computerized WASAM program, civil unrest made it necessary to move the computer to a safer location for a period of time. This disrupted operation of the WASAM program. However, there is little evidence that subsequent recalibration has taken place or that any systematic measurements below the main turnouts are made. Discharge measurements are made at only a few key structures. Other factors also contributed to abandonment of the WASAM program. As mentioned above, the pilot project, where WASAM had been controlled by the technical assistance team, was discontinued. Further, implementation of the program proposed for Kinda is data-intensive, requiring extensive field reports that include: on a weekly basis field wetness and actual area irrigated by each crop, and continual monitoring of actual discharge at many locations. This level of data collection overtaxed the capacity of ID staff and duplicated long-established procedures for recording cropped area. 3.31 Most important, WASAM's emphasis on equitable water sharing did not support the policy of top-down, target-driven, goals for maximizing production. Since the state owns all land, giving farmers only the right of use, it attempts to operate the agriculture system as a single integrated farm where decisions are made to maximize the returns to water and land. Equity in the distribution of irrigation and agricultural benefits is of secondary importance. GOM is interested in achieving its production targets, and places less importance on individual farmers' fate with regard to production. ASC will not hesitate to stop water to the tail end areas if water is not sufficient, despite the fact it had assigned targets and crops to those lands at the beginning of the season. Government does not take any responsibility for uncultivated land and crop failures due to water shortage. Farmers who planted according to targets and lost some or all of their crops can expect a reduction in their procurement quota but no compensation for costs incurred or income foregone. Township and Village authorities are sympathetic with farmers who suffer from these harsh conditions, and do not take decisions to cut off tailenders lightly. For their part, the deprived farmers appear to accept these decisions as appropriate to the circumstances. The WASAM program below the distributaries is largely irrelevant in this scenario and has been replaced by less data-intensive manual procedures. 3.32 Reservoir and Main Canal Operation. As noted above, the reservoir status and reservoir inflow expectation based on historic averages are used to establish the cropping plan and water allocation each season. Rainfall recorded in the command area and at the dam site may be significantly different from the rainfall distribution in the catchment area, so reservoir status is independent of the conditions in the farmer's field. ID staff monitor the reservoir, and township B 16 level MAS and ID staff monitor soil moisture and crop conditions. But, rather than weekly as proposed by the WASAM program, the irrigation status is reviewed on a monthly basis. Once the crop plan is fixed the irrigation allocation is adjusted only to accommodate deviation from the expected water supply. Farmers cannot request changes in the allocation based on subsequent decisions to modify their crop plan. 3.33 Release of the irrigation supply through the hydroelectric plant is working smoothly. The only problem reported regarding reservoir operation relates to a policy for determining release of water for power generation at rates of discharge and at times that make it impossible to fully use the water for irrigation. In 1994 government accepted that ID requirements should take precedence over power requirements in decisions on water release. These priorities have yet to be fully respected. Water released for power generation in the months December to January, when ID has only limited needs for supplying farmers growing onions, wheat and other winter crops, largely escapes down the Panlaung to the Irrawaddy. 3.34 Rotations. The shortage of reservoir water, and problems in enforcing effective rotation at the tertiary level, have prompted a change in plans. Both the RMC and Panlaung canals are currently operated using a rotation which gives the RMC all the water for nine days and the rest of the Panlaung system all the water for the next nine days. In each system, the upper regions get water for five days and then the lower regions for four days. During the "on" portion of the rotation for the lower region of the main canals all of the upper distributaries are closed, a relatively easy situation to monitor. 3.35 For the LMC, ID now divides the 125 km LMC into three "sections" and, when there is sufficient water to supply all three, rotates water between them, closing all distributaries along the stretches of the two out of turn. In the past five years, since the distributaries of the lower reaches of the LMC were completed, that has generally not been possible. Section III has often not been provided any water at all, especially during the pre-monsoon season. Section II begins at distributary 18 (DY18); Section III begins at DY26 (see Map 5). ID admits that only farmers in Section I can expect secure water supplies in the pre-monsoon. Section II farmers are at varying degrees of risk, and the situation of Section III farners is considered precarious. During the monsoon, many Section III farmers are still out of reach and effectively outside the command (para 4.2). 3.36 Rotations below the distributaries are operated with varying effectiveness depending on the location. The variability of this arrangement makes it difficult to describe a common practice. The impression from OED's group interviews is that rotations are generally effective along the distributaries with secured supplies of water, where the rotation is amongst tertiaries. Rotations amongst and within watercourses were not part of the original design, but have been adopted on occasion by the watercourse groups when they made sense to the group as a whole, either during the pre-monsoon season or when supplementary water was required during the monsoon. But these arrangements appear to break down in periods of water stress. Then, headenders take what they need before passing the water downstream and the tailend farmers invariably suffer. By contrast, rotations at the watercourse level is practiced mainly in the old system. 3.37 That the deprived farmers acquiesce in this situation, rather than press their claims, is one of the remarkable features of the Myanmar experience. It is a result of several factors, including an acceptance of the headend fanner's advantage, a mutual understanding that the B 17 headend advantage will not be exploited excessively, a tradition of social discipline supporting the status quo, LORC's effective control over retaliatory action, and the absence of any sense of shared concern for consistently equitable treatment in distributing irrigation water. On the LMC, the lack of durable rules for proportional sharing of available water contrasts with the very real communal commitment to proportional contributions to watercourse maintenance based on the size of holdings. 3.38 However, at and below the tertiary level, under conditions of scarcity, rotation gives way to another form of rationing. In this case, as mentioned above, ID, ASC and LORC authorities anticipating a shortage simply block all tertiaries and watercourses below a certain point. These administrative actions can cut off entire watercourses, or farms at the tail of watercourses. 3.39 Although prolonged drought periods spelled disaster for large segments of the command areas roughly one year in ten prior to dam construction, farmers in the tail end of the RMC declared that was better than with the dam and the rotation system currently practiced, where water is to be delivered to them only 4 out of 18 days. Farmers in the lower parts of the last few distributaries of the RMC no longer have access to even supplemental irrigation for rainy season paddy. They blame this on LMC which, because of the size of the command area, they say receives continuous water release from the reservoir. 3.40 Irrigation Department staff disagree with the farmers' interpretation. They agree that less water is being delivered to the RMC and other Panlaung canals in the rainy season, but claim that sufficient water is released to those canals to provide supplemental irrigation to their entire command areas if all farmers use water efficiently. They see the problem as one of overuse in the upper parts of the canal depriving the lower portions of water. The fact that LMC farmers also complain-that their shortage is explained by excessive allocations to the RMC and other Panlaung canals-indicates a high degree of ignorance about the rationale for the present operational plan. 3.41 Department staff did agree, however, that the development of government farms which extract water from the RMC and LMC for over 7,000 ac-for cotton, mangoes and other upland crops-have diminished the amount of water that can be delivered to the tail areas. This condition is being rectified in part on the RMC by constructing a connection to the Zawgyi system for the tail distributaries of the RMC. This is possible because a major dam on the Zawgyi River is now nearing completion. 3.42 Delivery Problems. One constraint on problem rotations along the main canals is that the distributary, tertiary, and watercourse canals should have been designed for larger discharge in order to deliver the design volume to designated areas in a shorter period than would be required during continuous flow. The consultants installing the WASAM program reported that the canals were being destroyed by not operating them in continuous flow "as designed." However, inspection in 1995 only identified minor problems that can be attributed to larger than design discharge. 3.43 The LMC design discharge is 950 cubic feet per second (cfs). ID staff indicated that when the LMC canal is clean they can safely deliver up to 1,000 cfs. They also said that at 1,000 cfs the system would perform satisfactorily with continuous delivery up to about DY26 but that even with slightly more than the design flow and a clean canal adequate irrigation delivery past DY29 was not possible. The low gradient of the LMC cannot explain all of this apparent design B 18 error. It would appear as well that more irrigation than allocated in the original design is being delivered to upper distributaries. Since this land has only recently been developed for irrigation and is not yet fully leveled, some additional irrigation may be necessary until the fields are appropriately developed. The substitution of paddy for cotton and other field crops is another factor. OED suspects that yet another explanation is that the rate of infiltration in the ya land is higher than expected at design, and water is lost from the system that should have reached the lower distributaries. Since this water moves eastwards perpendicular to the LMC, it cannot be recaptured by farmers on the distributaries further north-the farmers who are now bearing the largest shortfall from design discharge. C. Maintenance Performance (1) Maintenance by the Agency 3.44 Dam. The reservoir and dam infrastructure is well maintained. Mechanical gates in the spillway which have never been used to release water because of less than planned inflow are in good working order. 3.45 Newly Irrigated Area. Physical control structures are in place in the LMC. At the main canal level they are operated by ID with little interference from farmers. Minimum routine maintenance of this recently completed system is keeping the infrastructure in good operating condition. All LMC structures observed-offtake gates, cross regulators, siphons, and escapes- were in good operating condition. Structures below the main canal are also in good working condition. The outlet side of the drop structures show some erosion that could be corrected with routine maintenance. Access roads along the distributaries are heavily used and in poor condition. Lack of maintenance of the roads does not threaten the canal system. 3.46 OED did see evidence of unauthorized pumping and siphoning over and even through the LMC embankments. In several spots deep trenches had been cut under the roadway through the embankment to allow farmers to install a private hose connected to a pump either inside or outside the canal. Elsewhere there was a proliferation of loose hoses draped over the roadway, siphoning or pumping to adjacent fields. These encounters were at the upper reaches of the LMC, in Section I where water supply was most secure, during a rotation to a section downstream. ID officials accompanying the OED mission demanded immediate removal of the illegal equipment. But it was convincing evidence not only of selfish behavior by advantaged irrigators, but of some laxness in ID's patrol schedule (the deepest cuts must have been there several days) and of the difficulty of maintaining complete control even with the formidable administrative apparatus of ID and LORC. 3.47 Apart from illegal abstraction, the only serious problem with the LMC is rapid weed growth in the canal. It is a major concern for ID. The weed growth chokes the canal and reduces delivery capacity. Though, according to ID, the clean canal delivers the design discharge of 950 cfs, it only takes twenty days of full sunshine in the pre-monsoon season for weed growth to reduce the discharge by 15 percent. Within fifty days of cleaning the canal the maximum possible discharge is about 600 cfs, a 36 percent reduction from design values. 3.48 ID tries to clean these weeds from the LMC, a new species not previously seen on the RMC system, but has not yet been able to control the problem. Because of the persistence of the weeds, mass mobilization has become routine. Before the pre-monsoon season starts and in B 19 principle every 50 days thereafter the canal is closed. About 3 days are required to drain the water from the canal. Then villagers have two or three days to remove the weeds and silt. Nearly 75,000 persons worked to clean over 140 km of LMC and RMC in December of 1994. Parts of the LMC were cleaned four times that year, generally with a much small number of persons. Mass labor mobilization for removing weeds has been effective, but it is difficult to schedule the necessary canal shutdown for 5 to 6 days every 50-54 days during the pre-monsoon season to enable cleaning. 3.49 While responsibility for maintenance of distributary and tertiary level canals technically lies with ID, lack of budget requires periodic assistance for maintenance from the farmers at these canal levels too. Even in situations where responsibility is not always clearly spelled out, the response by farmers has been supportive. 3.50 Previously Irrigated Area The new and repaired structures of the main canals are in operating condition. All the head regulator/drop structures inspected on the RMC were functional. Some have had only minor repairs for many decades and others were improved by the project. In a number of cases erosion downstream of the structure is threatening to collapse the stilling basin walls. However, given the long life of these structures it is clear that timely maintenance has been done in the past or they would have collapsed decades if not centuries ago. Extending the downstream lining beyond the turbulent flow region would solve the problem in most cases. 3.51 A few distributary canals have been fitted with new gated control and measurement structures by the project, but in most cases existing structures have been repaited or improved. The ungated pipe outlets from the main canal in these systems have not been modified. However, numerous gates that were not renovated are missing spindles or linkage between the spindle and wooden gate. In many cases the gate could be operated manually but this requires more than one person to lift and prop the gate open. There are also many ungated offtakes from the main canals of the previously irrigated area. Because of their numbers, these cannot be controlled by ID. Even the gated distributary and watercourse outlets in these systems are mostly operated by the farmers. 3.52 The main and minor canals on the right bank have steeper gradients than the LMC and there was little evidence of weed growth and few complaints that weeds hampered irrigation delivery. For the RMC and Sama canal, villagers near the canals generally spend only one or two days helping clean them. 3.53 Drainage System. The project included construction of 180 km of new drains. No evidence was found of drainage problems in either the new LMC or the old RMC and Panlaung system despite a dramatic increase in irrigation application. In some of the lowest areas of the old system the water table is within a few meters of the surface but farmers do not report water logging. (2) Maintenance by the Irrigators 3.54 At the watercourse and field channel level maintenance is adequate in that no connection could be established during field visits between maintenance performance and lower than expected cropping intensity and production. The condition of watercourses varies from those that are virtually nonexistent in the lower reaches of the LMC and lowest parts of the middle B 20 distributaries, to those that are well-developed and maintained in the areas where the water supply is reliable. Watercourses not receiving pre-monsoon water are maintained just prior to the start of the rainy season. Farmers reported that working as a group they can reshape and clean their watercourses in one or two days. 3.55 It was observed that individual farmers built field ditches to their plots growing non- paddy crops in the pre-monsoon season. These ditches were cleaned regularly. Farmers reported that they close field ditches running through their fields during the monsoon season to expand the paddy area and because they prefer field-to-field irrigation. 3.56 The only visible maintenance problem at the farmer level is the poor condition or absence of most project-supplied gates at tertiary as well as watercourse turnouts. Farmers rebuild the gates with banana stalks, earth and cloth when necessary. But the damage is evidence of the suspicions that most farmers behind the gates harbor about the utility of rotations. B21 4. Kinda Scheme: Agro-Economic Impacts A. Agricultural Impact 4.1 Net Cultivated Area The PCR assumed that the design in the Staff Appraisal Report (SAR) was still valid and assumed that by the year 2000 the entire programmed area of 195,000 ac would have access to irrigation, despite the delays in expanding the network in the early years. That assumption must now be abandoned. Not only are releases from the dam likely to be less than expected over the long run. The application of water to paddy planted on the high land of the LMC command implies that much more water than planned is being lost through infiltration through these better drained soils. The rate of infiltration through these soils also appears to be higher than expected.17 Added to these losses, the low-gradient architecture of the left main canal, and the weed growth it has stimulated, reduce the velocity and discharge to the lower reaches. Taken together, these factors indicate water supplies to the LMC will never provide full coverage up to appraisal design even if the dam were full. The roads through the lower reaches of the LMC-apart from in the vicinity of the Pyugan Tank-pass through a virtual desert, much of which is unlikely ever to benefit from the project. 4.2 Perhaps 20,000 ac will remain forever rainfed, though that will not be confirmed for many years. If the estimate is correct, it means that the project will have added about 87,000 ac of newly irrigated land to the 88,000 ac irrigated in the old scheme, for a total of 175,000 ac net irrigable and cultivated area. That is less than the 195,000 ac planned, but not an unimpressive result. 4.3 Cropping Intensities. The more important shortfall is not in the net cultivated area-the acreage under actual command-but in the gross cultivated area-reflected by the cropping intensity. That is where the real problems of the Kinda scheme are found. 4.4 In the areas on both the right and left banks where water has been secured throughout the year, the project's objectives have been fully reached. The Kinda reservoir has dramatically improved reliability of irrigation delivery to these parts of the command area. The effects of reliable water supply on the cropping intensity of a large number of farmers was forcefully revealed during group interviews. Increases in land prices (in the "illegal" land market) in areas with a secure water supply support that point. 4.5 Almost all of the farmers interviewed in several sessions in these favored areas had individual cropping intensities above 140 percent, with the average for groups of farmers from the different areas ranging from 180 to 200 percent. Those figures compare well with pre- project intensities for the old scheme alone, reported at appraisal to be 125 percent. The situation of these farmers does not represent the command area as a whole. However, the consistency of their cropping intensity suggests that, with adequate and reliable irrigation, the project design goal of 160 percent could easily have been achieved in all parts of the command area. 17. Para 3.43. B 22 4.6 Many farmers in these groups said they frequently face water shortages that affect yields and in some cases the crop selected, but few of them said that water was the main constraint in selecting more intensive cropping. Rather, unattractive market prices, cropping pattems imposed by the state, lack of labor, and the high cost of other inputs were more important factors. However, farmers with land in the lower reaches of the same distributaries, with infrequent and unreliable irrigation, were unanimous in saying it was the lack of irrigation that was the main limitation. 4.7 There is no routine reporting of agricultural data related exclusively to the scheme. The Irrigation Department assembles MAS data based on Village Tract reports of actual planted area and yields for all crops. These data come from four townships, but before they can be used, the Tracts inside and outside the scheme must be separated. These data purport to distinguish between irrigated and non-irrigated fields and are considered reliable. The PCR used the same source. 4.8 Table 4.1 presents the area cropped by season, for the 175,000 ac net irrigable and cultivated area under actual command, for the period from 1990/91 through 1994/95, the five years since the LMC distributaries were completed. Average cropping intensity for the period is 105 percent. For the last three years, the average is better, 1 17 percent. Comparing these figures with the average of the PCR estimates for these three years, 1 12 percent, 18 actual scheme performance looks rather good, although several years behind appraisal schedule. But the PCR assumes that intensities would continue to improve steadily from 75 percent prior to the project (including the rainfed left bank) to the SAR target level of 160 percent at full development. This projection can no longer be justified. In fact the trend of annual cropping intensities may have already leveled off. The PCR projections are undermined by the expected shortfall of water in storage, and by the preference for planting paddy over cotton and other upland crops. If the latter practice cannot be reversed, overall cropping intensities are unlikely to close much of the gap between the recent level of 1 7 percent and the level targeted level in the SAR of 160 percent. 4.9 Cropping Pattern. The choice of crops has a large impact on water requirements. In the pre-monsoon season, for example, ID estimates that it will need to deliver about 58 ac/inches of water for a crop of rice at Kinda compared to 30 ac/inches for cotton or sesame. Thus, any shift from paddy to these upland crops-where soil conditions permit-will double the effective area irrigated and raise the cropping intensity. In the monsoon season, the ratio is even higher- closer to three to one-though less irrigation water is needed for any crop in that season. 4.10 The area actually planted to each crop can be compared to the area planned during design. Table 4.1 shows the actual and planned annual cropped area for the major crops. The area planted to paddy was close to the design area, and even exceeded it in 1994. The area planted to chilies, sesame, and especially cotton averaged considerably less. As a percentage of total cropped area, the paddy/non-paddy difference is even more pronounced. 18. The PCR's year-wise estimates have been read from a trend line. Annual figures are not provided. The PCR data refer to the full 195,000 ac design command. Table 4. 1 Kinda: Seasonal Cropping Pattern and Intensity ,w3 Irrigable Crop Area (Acres) Total Cropping Intensity U Area Paddy Cotton Sesame Chillies Onion lWheat Beans Other Area Season Total 0 % % 1990/91 Premonsoon 175,000 15,617 13,033 21,089 9,442 9,084 68,265 39 Monsoon 175,000 68,837 7,593 76,430 44 0 Winter 175,000 6,400 1,770 1,529 1,127 10,826 6 Total 155,521 89 no 1991/92 Premonsoon 175,000 362 8,863 14,147 5,994 16,801 46,167 26 Monsoon 175,000 72,034 18,571 90,605 52 0 Winter 175,000 6,922 3,250 3,629 546 14,347 8 W. Total 151,119 86 1992/93 Premonsoon 175,000 14,635 21,876 32,516 22,696 5,223 96,946 55 a Monsoon 175,000 83,879 34,083 117,962 67 | Winter 175,000 8,486 1,497 1,767 1,863 13,613 8 6 Total 228,521 131 r 1993/94 Premonsoon 175,000 36,439 7,745 14,605 12,899 100 71,788 41 Monsoon 175,000 81,071 5,807 86,878 50 Winter 175,000 7,911 2,782 6,120 8,093 24,906 14 Total 183,572 105 1994/95 Premonsoon 175,000 36,208 13,021 19,889 11,799 5,270 86,187 49 Monsoon 175,000 78,132 10,521 88,653 51 Winter 175,000 5,449 1,950 10,828 8,314 26,541 15 Total 201,381 115 Source: MAS data using Village Tract information on actual planted area. Information assembled by the Irrigation Department. w B24 4.11 The most noticeable difference is the failure to establish cotton as the dominate crop on the LMC. There is less cotton acreage now than there was before the project.'9 If farmers can be persuaded to replace paddy with cotton where soil conditions permit on existing irrigated fields, the shortfall in cropping intensities from SAR targets would be substantially, though far from fully, closed.20 4.12 The average area planted annually during the period from 1993 to 1995 was 196,000 ac, a little less than two-thirds of the 317,000 ac planned. As discussed above, reservoir inflow remains below design by a factor of about 11 percent (para 2.7). If 11 ercent more water were available and if all of it was applied to non-paddy crops, the total area cropped woild increase to 228,000 ac or a little less than three-quarters of the 317,000 ac planned, a significant improvement though still well below the level expected at design. The deficit in irrigation supply thus joins the paddy preference to provide the main causes for the gap between expected and actual area cropped and the low cropping intensity. 4.13 The dominance of paddy is the result not only of the farmers' preferences but also of pressure from government. The gap between projected and actual cropping patterns is thus an indication of the failure of project designers and officials to adequately understand farners needs, as well as anticipate the overriding force of government's long-term policy objectives. By allowing rice prices to rise in the last several years, government has been able to satisfy farmers and its own policies simultaneously. 4.14 Yields. OED's estimates of paddy yields in the pre-monsoon (3.1 tons/hectare 21) and monsoon (3.6 tlha) seasons22 are about 80 percent of SAR projections for full development. The PCR gave a single estimate of 3.7 t/ha for both seasons. These figures compare with estimates of pre-project paddy yields of 2.8 t/ha for irrigated fields in the old system, and 1.5 t/ha on rainfed farms of the LMC. 3 OED's estimates of sesame yields are also about 80 percent of the SAR projection: 320 kg/ha vs 400 kg/ha (the PCR accepted the SAR projection for sesame). Those irrigated sesame yields can be compared with the SAR and PCR estimates of pre-project sesame yields-both about 200 kg/ha. 4.15 A much larger shortfall from projected yields emerges with the cotton crop. OED's estimate is 650 kg/ha, only 40 percent of the SAR projection for full development (1.7 t/ha) and no different than the SAR and PCR estimates of pre-project irrigated cotton yields. Thus, for cotton, not only has there been no growth in irrigated area, but the yields have remained static. 19. Cotton was incorporated in the planned crop rotation to raise the ERR and justify the project, against the objections of some Bank staff who felt that the assumption it would be taken up by fanners at the scale projected was too optimistic. At appraisal, non-paddy crops, led by cotton and sesame, were expected to occupy 66 percent of the total cropped area, mostly on the LMC. Cotton was expected to take a third of that (about 23 percent). But for the three crop years 199213-94/5 non-paddy crops actually took only 46 percent. Cotton took 7 percent, and most of that was planted in traditional zones in the old scheme. In short, paddy has crowded out upland crops in the farmers' rotations-compared with appraisal projections. 20. That refers to the 175,000 ac assumed in this report to be under command. 21. Yields are reported per hectare in this report. See Footnote 5. 22. OED's yield estimates are based on interviews with MAS and ID staff, as well as individual and groups of farmers. 23. These are PCR estimates. The SAR used lower pre-project estimates: 1.5 t/ha and 1.0 t/ha for the pre-monsoon and monsoon seasons. B 25 This deals a further blow to the SAR rate of return estimates, since almost one-half of all incremental benefits were attributed to cotton. 4.16 Production. Annual total production in tons of paddy, cotton, sesame and several other crops is shown in Table 4.2, supported by Annex 2. It compares projections at appraisal for full development, which was to be reached in 1993/94, and as calculated by OED in 1995 for the average of the last three years, i.e. the same time frame. Area and yield figures for the three main crops are as quoted above. The production total for each crop includes yield increases attributable to better water supply on fields in the old areas that were already irrigated. The table shows the remarkable concentration of incremental production on paddy. The anticipated expansion of cotton, sesame and the other crops failed to occur. The paddy crop grew by 100,000 tons, two-thirds of the expected 149,000 tons recomputed by OED based on appraisal area and yield estimates. By contrast, an incremental cotton crop of 67,000 tons projected at appraisal never materialized, and cotton production actually declined by 6,000 tons. Table 4.2: Kinda-Incremental Crop Production Estimates (tons) SAR OED Impact Paddy 148.7 99.6 Cotton 66.9 (6.3) Sesame 7.9 0.9 Chillies 20.4 8.3 Onion 20.2 4.4 Source: Annex 2 4.17 Since the basic canal infrastructure is now in place, and all available water is used and used rather efficiently (para 2.9), no substantial further expansion of total production can be anticipated. The exception would be if inflows into the reservoir return to appraisal expectations, which would provide an additional eleven percent to the water supply. The PCR acknowledged the shortfall in area planted and production at the time of writing in 1991, but assumed that ultimately the scheme would catch up with appraisal targets. That assumption is no longer defensible. B. Financial Impact, Farmers' Level 4.18 The average family holding of irrigated land is larger on the previously rainfed left canal than on the already irrigated right canal. Nevertheless, current information suggests that a rough average of about three hectares is acceptable for both. Crop budgets per hectare calculated by OED at 1995 financial prices and free market exchange rates (K100=US$1) indicate a net incremental income of about US$1,815 per year for a fully-irrigated three hectare upland farm on the LMC with two crops of paddy (Annex 3). Each crop returns about US$300 per hectare, after subtracting the value of a rainfed sesame crop that would have been grown during the monsoon season in the absence of irrigation. If the LMC farmer were growing irrigated sesame and cotton as had been planned for the pre-monsoon and monsoon seasons respectively, the total incremental return would be much less. In this case it would be only US$670 for the two crops (3 ha), reflecting the substantial difference between the present price incentives for paddy and B 26 other field crops. Presumably, this is the factor mainly responsible for the LMC farmer ignoring the appraisal cropping rotations and planting paddy on the ya-land.24 4.19 The paddyland farmer on the RMC enjoys a smaller incremental income attributable to the project, because he was already profiting from the older scheme. Assuming he used to grow one secure monsoon crop per year, and a pre-monsoon crop every second year, his incremental income would be about US$1,005. Total net income for the farmers on the LMC and RMC from paddy is the same, however, about US$2,010 from the three hectares. The equivalent of about 13 percent of that in paddy would be retained by the family for home consumption.25 C. Economic Impact 4.20 The SAR economic rate of return (ERR) for the whole of the Kinda scheme was estimated at 21.3 percent, including the power component. The PCR re-estimated the ERR at 14 percent, a decline attributed to higher investment costs (15 percent above projections), a lag in crop production benefits, and a fall in the price of rice. As mentioned above, PCR maintained that the SAR's full development production targets would eventually be met. 4.21 The ERR again must be revised sharply downward. For the impact study OED did not repeat the complete economic analysis based on farm budgets carried out at appraisal and completion. Rather, the SAR analytical framework was preserved, and adjustments were made to the cost and benefit streams to reflect the 15 percent increase in construction costs and, of far greater significance, the actual and projected shortfall in the incremental value of agricultural production. Production of the different crops was aggregated after adjusting each crop other than paddy by a crop/paddy price ratio to convert them to paddy equivalents.26 This allows the SAR and OED tonnage estimates to be compared. 4.22 Actual total production benefits are only about 40 percent of the appraisal projections.27 When the SAR crop production benefit streams are reduced by these amounts, with comparable28 reductions to the costs of crop production, and the higher investment costs are also incorporated, the ERR falls from 21.3 to 9.7 percent. 4.23 A final adjustment has to be made to reflect the decline in the traded price of rice since the appraisal report was prepared. The SAR used a 1990 rice price, based on 1980 constant 24. As discussed in the Overview to this irrigation impact study, a comparison of input/output price ratios from all three countries included in the study shows that Myanmar's paddy farmers have a distinct advantage and enjoy the most favorable ratios. 25. That corresponds to 2.5 tons of paddy, or 1.5 tons of rice. These parameters are based on per capita annual rice consumption of 250 kg, for a six member family. See Mitchell, Donald O., and Merlinda D. Ingco. November 1993. The World Food Outlook. International Economics Department, World Bank, p. 160. 26. The prices were taken from the PCR (economic prices, PCR p. 55). OED did not update the PCR prices, and to the extent there have been changes in relative prices the comparison should also be adjusted. That adjustment is immaterial, however, because of the vast difference between SAR production projections and OED's updated projections. 27. Appraised projections for groundnuts and pulses were ignored. These crops are grown in the LMC extension, but mostly without irrigation 28. See Footnote e, Annex 4. B27 dollars, of US$504/t. That converts to exactly US$700/t in 1990 dollars. The actual price in mid- 1995, again in 1990 dollars, was US$237/t, or 34 percent of appraisal expectations. Adjusting only the appraisal paddy price in the ERR calculations by this factor,29 leaving the other economic crop prices intact, would bring the ERR to 7.4 percent. This analysis was not pursued30 because it was obvious the project is only marginally viable in economic terms. Details of the calculations are provided in Annex 4. The benefit steam for the power component was left intact. Flood control and the fish catch from the reservoir31 are additional economic benefits which are not reflected in the analysis. 29. The adjustment should be by less than the difference in fob Bangkok rice prices, but the impact of that d;stor.ion has little effect on the outcome. 30. For example to adjust all the re-calculations to reflect the very recent increase in the Bank's rice price estimates for 1995-1996 by 17 percent. 31. About 70 full-time fishing families live on the margin of the reservoir, and deliver their catches to one of their own, a women licensed by the Central Govemment Commander in Mandalay. During the monsoon she is contracted to deliver to collecting army personnel of Central Command about 130 kg a day. Any surplus is retained by the fishermen. B 28 5. Tank Schemes A. Structure of the Irrigation Schemes 5.1 Kitnmundaung. The Kinmundaung subproject included a 25 meter-high earth dam, a reservoir fed from about 74 km2 of catchment, construction of a six km main canal, rehabilitation of an existing diversion weir, extension and remodeling of the right and left distributaries (each about six km long), and construction of the subsidiary irrigation network (Map 6). The system as finally approved would command about 5,000 irrigable acres, although original plans called for 4,000 ac. The storage capacity was expected to provide supplementary irrigation for the monsoon crop for all 5,000 ac, including 3,000 ac of paddy and 2,000 ac of other field crops (cotton and sunflower were proposed in the SAR). It was also expected to support 1,500 ac of pre-monsoon sesame and "up to" 1,000 ac of post-monsoon groundnuts. The average farm holdings of irrigable land was a little under 5 acres, and 1,100 farmers would participate. 5.2 The annual inflow to the reservoir was overestimated. Streams with highly variable discharge are extremely difficult to gauge reliably. Bank files during preparation attest to the concern of the hydrologists to get reliable estimates. Preparation had perforce to rely on flow measurements taken at the weir, not at the reservoir site. The SAR states that estimated runoff from the reservoir catchment based on records for the period from 1959-81 was 22,800 ac-ft. The expected water availability ("yield") from the reservoir was estimated to be 16,800 ac-ft. The actual average inflow for the 1986-94 period was only 8,700 ac-ft, about 50 percent of the expected irrigation supply, and the average release for the same period was 5,900 ac-ft, about 35 percent of the expected supply. The maximum inflow in one year was 72 percent of the expected supply. The spillway has never been topped. The scheme is clearly overdimensioned in relation to the catchment. When water is needed, the irrigation infrastructure in the lower reaches of the system remains dry and nonfunctional. The water shortfall is significantly more severe at Kinmundaung than at Kinda. It appears to be attributable to the miscalculation of reservoir inflow rather than the period of reduced rainfall, though that has aggravated the problem. 5.3 Azin. The Azin subproject includes a 28 meter-high earth dam, a reservoir fed from about 6 km2 of catchment, and construction of a main canal, two major distributaries and the subsidiary irrigation network (Map 7). The distance from dam site to the end of the longest distributary is about ten km. The system as designed was to command about 2,000 irrigable acres of paddy land and 850 ac of orchard. It was expected that the 2,000 ac would be farmed b3 about 450 families, averaging four-five irrigable acres each. The irrigated orchard land was to be divided between a state farm of 250 acres and 160 individual holdings averaging nearly four acres each (totaling 600 ac). Unlike Kinda and Kinmundaung, the catchment and reservoir at Azin have proven to be more than adequate to irrigate the design area, and the spillway is topped every year. B. Operation and Maintenance (1) Kinmundaung 5.4 Water Allocation. The planning process is the same as described for Kinda. The crops specified by the targets in Kinmundaung are paddy and sesame. There is little if any fanner B 29 input to the crop plan. Farmers must plant the specified crops or face penalties that could include loss of cultivation rights or imprisonment. 5.5 Punishment. In 1994 twenty persons were arrested and fined. All were fined K500 (US$5), five for breaking the article of the water code which prohibits damaging the canal, the others for failing to attend to maintenance as requested. Some of the persons were reported by the watercourse leaders. Some were reported by ID staff. Several were arrested by the chairman of the Village Tract LORC and taken directly to the police without consulting ID. 5.6 Rotations. An important difference between Kinda and Kinmundaung is that the water shortage in the Kinmundaung tank limits distribution of supplementary monsoon supplies as well as pre-monsoon applications. As indicated in para 5.17, only 2,500 ac of the planned 5,000 ac receives monsoon irrigation regularly, and only about 500 ac receives pre-monsoon irrigation. Rotations in the monsoon season at Kinmundaung are administered on an ad hoc basis, depending upon the rainfall. Nevertheless, it appears that at Kinmundaung as at Kinda these rotations at the tertiary and watercourse levels are respected by the farmers as long as supplies are sufficient-over the area approved by ASC and ID-but break down when water is short. Then, as at Kinda, headend farmers use as much water as they reasonably believe they need while tailend farmers go short. Their group structure is not cohesive enough to guarantee proportional distribution, and neither is it enforced by ID or the Village Tract LORC. The farmers who are to get nothing are usually told that in advance of the season. 5.7 There is no "rotation" in the pre-monsoon season at Kinmundaung. ID and the Village authorities decide which small block(s) of farmers will receive the privileged supplies and when the releases will be made. Because the 50 ha project research farm is located in the command area of the RMC at Kinmundaung, the RMC is favored for water release in the pre-monsoon season and farmers adjacent to the station can count on being selected. 5.8 Leadership. The process of election of watercourse leaders, and ratification by the Village and Township LORCs, is comparable to the Kinda model. An informal system was already in place covering the approximately 20 percent of the project farmers who benefited from irrigation supplied from the old diversion weir. However, in the OED interviews there were few references to rules, roles or other indicators of farmers playing an active part of management. "Participation" in scheme decision making is limited to the leaders and those other irrigators who are members of the Village Tract authorities. 5.9 Maintenance. It appears that the agencies and irrigators handle most if not all maintenance tasks without difficulty. Before the dam there was considerable silt carried into the system which farmers say enhanced fertility, even though they had to desilt the canals. After the dam there is very little silt deposited in the canals. The main maintenance activities therefore are cleaning weeds and making minor repairs to embankments where crabs, rats and other animals have caused damage. Many of the lower reaches of canals have never been used. But even there, because of low rainfall there has been little deterioration in the canal and its structures though the channels are overgrown and often hidden by weeds. 5.10 Overall the canals inspected by the OED team below the distributary level including watercourses and field ditches were not as well cleaned as in Kinda. There appeared to be more than two months of weed growth and siltation. Nevertheless, the water was moving well enough to serve the fields still requiring supply. It is when the flow drops below that point that the B 30 farmers clean the canals. The incentives for concerted action well in anticipation ofpre- monsoon supplies are stronger than for monsoon supplements-because the pre-monsoon crop depends on the canals-but that affects at most ten percent of the design command area. Officials as well as farmers claimed the irrigators maintained the canals when necessary. How the canals look does not matter to farmers as long as they get their water. 5.11 Much of the work on the larger canals is done by labor hired by ID, although, as at Kinda, farmers respond without objection to instructions from the Village Tract LORC to clean the mains, minors and distributaries. One farmer interviewed with fields 3,000 m from the main canal on LMC Minor I is 1,000 m beyond the limits of supply through that channel and has only received water in his fields once since the canal was built. However, in 1994 he had to spend three days on two occasions cleaning the minor. There is sufficient labor living in the design command area to clean all canals in a few days if necessary. The weed species that proliferates in the Kinda LMC has not appeared at Kinmundaung, where the major canals have a steeper gradient and the water a greater velocity. There was adequate demonstration that the VT LORC can accomplish such mobilization to conclude that canal cleaning does not limit system performance. 5.12 The dam and its control structures are in good condition. Nothing was seen by the mission or reported that would indicate any problem with operation of the reservoir. The diversion weir, RMC and LMC structures are also in good working condition. Where Minor I from the RMC had breached an embankment several years ago because the embankment had not been properly compacted, and although ID intends to repair and line the minor, farmers simply bypassed the break using an existing watercourse. Most gate emplacements and other masonry control structures below the distributaries were functional, although the wood gates (stoplogs) were everywhere missing. The few gates in the system operated by ID's Assistant Canal Inspector were operational. Accepting farmer-installed banana stalks and earth as gates, then all structures are functional. (2) Azin 5.13 Crop allocations, watercourse groups and rotations are organized as at the other sites in Myanmar. No O&M problems of any sort were reported or observed. The overall rotation provides the orchards water two days a week, and the paddylands five days a week. Each paddy farmer gets water twice a week. Tailend farmers are served first, and there are few complaints about headender abuse. No water is released into the main canal during the four monsoon months: the rains are sufficient. Orchard farmers receive irrigation the other eight months; paddyland farmers only during the pre-monsoon growing period. Complaints of water shortages by the orchard and paddy farmers are rare. However, MAS is currently piloting a third crop in 500 ac out of the 2,000 ac paddyland total. MAS foresees the day when the present abundance of water will end, and competition for irrigation will emerge. It is trying to persuade orchard farmers to shift to sprinklers and other water saving technologies in anticipation of that event. 5.14 The dam, main and two principal distributaries were all in good condition. The distributaries are lined to the boundary between the villages and the paddyfields. At the time of OED's visit the main had just been cleaned and was immaculate. Watercourses in the paddy area are structurally sound but often choked with weeds, when, for example, the summer crop is being harvested. The OED visit coincided with the start of the monsoon season, and the farmers were unconcerned with the appearance of the channels. In anticipation of each summer season ID B 31 organizes one substantial cleaning of the larger structures, and farmers help with three other cleanings during that season. They provide this labor when instructed to do so. They also patrol the canals, including the distributaries, and call for ID's intervention only when they cannot make the repairs themselves. 5.15 According to ID and MAS the ditches initially constructed by these paddy farmers-who had never worked with irrigation before-were poorly made. The agencies had to step in to instruct and assist the farmers in remodeling the channels. In the last three years the paddy canal network has been extended to the full 2,000 ac. Those farmers that had opted not to constiuct the ditches but relied on field-to-field flooding have now almost all made private connections. The last 85 acres to be connected, in four separate pockets on the fringe of the scheme, were to be completed in 1995. 5.16 For the first time at the three project sites visited in Myanmar, OED found a large collection of wooden gates that were in good condition and said to be used as intended. Since they were not needed at the time of the visit, they had been pulled out of the masonry gate structures and stored in the yard of a member of the water user group at the edge of the village and close to the scheme. This is further evidence of the Azin scheme operating as planned. C. Agro-Economic Impacts (1) Kinmundaung 5.17 Over the last four years ID has provided, on average, supplementary monsoon irrigation for about 2,500 ac, all of it dedicated to paddy. The SAR indicative plan called for 3,000 ac of paddy and 2,000 ac of cotton and sunflower. Thus, only half of the area commanded by the project infrastructure has actually received irrigation during the rainy season. In the pre- monsoon season the fraction is much smaller. In 1994 ID supplied 500 ac of paddy; in 1995 it supplied 310 ac of paddy and 250 ac of sesame. Had all the pre-monsoon water been applied to field crops other than paddy, the 1995 total would have been about 900 ac, up from 560 ac. The SAR estimate for the off-season was 1,500 ac of sesame plus up to 1,000 more acres of groundnut. The SAR did not anticipate using water for paddy in the second season. 5.18 The SAR indicative plan for developing a monsoon cropping system that included irrigated cotton was also frustrated. The only cotton planted in the scheme is at the research farm. The sunflower and groundnut components of the SAR rotation were also not taken up by the farmers. As long as paddy dominates the cropping pattern, the intensity of land use will be commensurately lower than the potential. 5.19 For 1995, the intensity of irrigated farming was 61 percent for the design command area (5,000 ac), or 122 percent for the actual command area (2,500 ac), figures that can be compa.ed with the appraisal target of 150 percent for 5,000 ac.32 32. The SAR cropping pattern tables show a minimum 130 percent intensity, ranging up to 150 percent depending on arnual water availability for a 1,000 ac post-monsoon groundnut crop. In the SAR ERR calculations, the full groundnut component was included without qualification. B 32 5.20 Most of the crops irrigated during the off-season, whether it be paddy, sesame or other, can be counted as incremental. But that refers only to about 500 ac. The monsoon benefits, enjoyed by about half the farmers within the scheme perimeter, are more substantial. During interviews, farmers within reach of the irrigation supply claimed that their monsoon paddy yields had doubled, an increment of about 0.8 tons per acre (2 t/ha). 5.21 A reestimated rate of return was worked out for the Kinmundaung subproject, based on the SAR ERR framework. The sharp reduction of 60 percent in cropping intensity substantially decreases the ERR. This loss is compounded by the decline in the price of rice, as described in the Kinda analysis. The recomputed ERR for Kinmundaung is negative. The SAR estimate was 13.6 percent. The PCR reestimate, issued in 1991, is 12.3 percent, based on the assumption that the appraisal cropping pattern and intensity would be achieved. 5.22 Paddy yields for the farmers who have reliable water for the monsoon season are about the same as at Kinda, and incremental net income per hectare per crop is also estimated at about US$300. The average irrigated farm size is 2 hectares; thus for those privileged farmers enjoying full water supply for the two seasons, the net incremental farm income from irrigated fields is about US$890. Total net income from those two-hectare farms is US$1,340, two-thirds the earnings of the Kinda farmers (see Annex 3). A little under 20 percent of the paddy would be retained for home consumption (Footnote 25). (2) Azin 5.23 Average annual releases from the Azin reservoir during 1992-94 were 99 percent of projections. Against the SAR target of 850 ac of orchard, an estimated 838 ac are presently under irrigation, including all 250 ac on the state farm. The acreage of paddy benefiting from irrigation has grown steadily since the first release to the lowlands was made in 1991. In the 1994/95 dry season, 1,914 ac of paddy were irrigated. The total was expected to reach the scheme potential of 2,000 ac in 1995, as the last farmers dug their ditches and entered the network. 5.24 Summer season paddy yields have been about as high as expected, ranging between 3 and 5 tons/ha (60- 100 baskets/acre). MAS uses an average of 1.7 tons/acre (80 baskets/acre, or 4 t/ha) in its reports, which compares with the SAR projection of 1.8 tons/acre. That is consistent with the OED interviews, and is 20 percent higher than actual yields at Kinda and Kinmundaung. This is all incremental production attributable to the project. Nothing was planted on these fields in the dry season before the project. Conversely, no supplementary irrigation is needed or released for the monsoon crop, and no benefits from that season are attributable to the project. 5.25 The private orchards, comprising varying mixtures of mango, durian, pomelo and other fruit, appear to be maturing on schedule. Planted in 1989, the first fruit were harvested in 1994. Full maturity will not be reached for all trees until 2009. The plantations have many vacancies, but these supposedly are being or will be filled by replacement seedlings. MAS reckons that the farmers will have no trouble reaching the SAR yield projections for durian and pomelo, but that the mango estimates were exaggerated to begin with and the potential is about half the SAR figure. None of the 187 farmers presently in place (up from 160 at appraisal) have had prior experience with orchards of this size (4 acres), and the Azin project was seen as a pilot for subsequent irrigation schemes in this part of the country. MAS is satisfied that the experiment has succeeded. B 33 5.26 These projections of fruit production, although reasonable and consistent with MAS expectations, may prove to be overly optimistic. The orchard program got off to a bad start because the parastatal bank which the government had agreed would provide term credit to support the plantings,33 never set up the necessary lending line. All of the private plantations were self-financed, apart from a subsidy on the seedlings, with some risk to the quality of the plantings due to inadequate application of fertilizer and other inputs. MAS believes the damage was small. Fruit prices have been rising rapidly in domestic markets in the last five years, giving MAS confidence that fears at appraisal that the market would be quickly saturated were unwarranted. Nevertheless, the harvest has just begun, and the real test of marketability on domestic markets has yet to come. 5.27 Rates of return have been re-estimated for the Azin subproject, again based on the SAR ERR framework. There is no reduction necessary for cropping intensity, where performance meets the SAR target. If MAS succeeds with its pilot project, and a third crop is added throughout the paddylands, the intensity will increase above the target. However the ERR is affected by the decline in the international price of rice. OED re-calculates the ERR for the Azin investments, including the Mudon town water supply component, at 6 percent.34 The SAR estimate was 12.3 percent. The PCR reestimate was 7.3 percent, reflecting the first, dramatic phase of the rice-price collapse and higher orchard investment and operating costs. If the dam had been completed and the water released on schedule three years earlier, OED's reestimated ERR would have been only marginally higher, about 8 percent. If the rice price had not declined, the ERR would have been 11 percent. 5.28 Net incremental earnings from the second irrigated paddy crop, for the typical two hectare (four-five acre) Azin farmer are US$840. Incremental earnings for the orchard farmers have not been recalculated, pending progress of the plantation and the success of pest control and infilling. 33. And to support as well the households during the trees' immature period, to the extent the intercrops did not meet all their needs. 34. The Mudon water supply benefit stream was adjusted to reflect the very slow buildup in demand for household and standpipe supplies. See the PAR. The PCR was prepared before the shortfall in demand appeared. B 34 6. Conclusions: Influence of O&M Performance on Agro-Economic Impacts 6.1 Kinda. As demonstrated by Chapter 3, O&M performance by ID as well as the irrigators is satisfactory. The primary problems were the overwhelming dominance of paddy and the uncontrollable weed proliferation in the LMC. Nevertheless, the OED team considered, and ID and MAS officers were asked to assess, the relative importance of five O&M practices at Kinda where sub-optimal behavior could be depressing total irrigated crop production. Three are operational deficiencies; two are maintenance deficiencies: * supporting inefficient cropping patterns, especially by supplying water to paddy in seasons and fields where larger areas of ya-land crops could otherwise be cultivated and where more water would escape by infiltration; * inappropriate and erratic rotation practices at all levels of turnout to subsidiary channels. This includes a concern about ID's failure to continue the sophisticated control system WASAM; - overusing water on the fields, particularly but not exclusively on paddy; * inadequate maintenance of the irrigation structures-the canals, gates and other controls-resulting in excessive losses in conveyance; and * insufficient cleaning of canals and watercourses, also resulting in losses in conveyance to the fields. 6.2 The tendency of established water rotations at Kinda to break down under stress, when headenders take more water than an equitable rotation would otherwise have allowed, encourages the preference of headenders for paddy. At present, that leads to an inefficient outcome: apart from the inherent desirability of diversification scheme-wise, the headenders occupy higher, better-drained fields with a comparative advantage for crops requiring less water. If the rotation worked as planned, more water would pass to tailend farmers with fields that can only grow paddy. But that outcome is attributable to crop preferences and relative crop prices, not a failure of rotations. It is easy to imagine a situation where headender crop preferences and comparative advantage are aligned, and overindulgence of headenders because of ineffective rotation would be economically justified. Rotation at Kinda is an instrument of equity, not maximization of value added. Thus there is no predictable relation between the failure of rotations and production impact. In any case government overrides all conventional rotation practices-precisely in order to maximize production-by simply cutting off sections of canal altogether. WASAM does not enter this picture. It was too data-intensive and sophisticated to survive without the consultants. 6.3 Oversupply of water to paddy fields-above the crop's evapotranspirational requirements-is a factor, but appears to be less important than in other countries in OED's regional impact study. The village-level presence of MAS and other public agencies is impressive in Myanmar. Given the government's preoccupation with crop production targets, and given the recognized shortage of water in the reservoir, excessive waste of water is not B35 accepted behavior in the Panlaung system and never has been. Headenders will have more opportunity to overuse water, but the level of abuse seems to be tolerable. 6.4 With reference to the maintenance of structures, the masonry works everywhere, and the undershot gates leading off the main canals and most distributaries, are kept in good condition by ID. However, at the levels operated by the farmers-the turnouts to the watercourses and from them to the field ditches-project supported metal and/or wood gates have been removed. The OED team was especially alert to the high visibility of damaged and missing gates. This problem has the same effect as the breakdown of rotations, in encouraging headenders to grow paddy. But, as with rotations, the direct impact of missing gates on production is unpredictable. 6.5 The farmers behind the gates do not want them. Farmers downstream from the gates do. When the agency replaces the gates, the intent is usually to shift water from the first group to the second. At Kinda, ID feels it is futile to try, because it is so easy for the first group to remove them again. Meanwhile the farmers behind the gates have time-proven methods for replacing the gates when they want to block the water. Most of the masonry gate housings are intact, and narrow enough to allow the farmers to substitute stalks, earth, cloth and other local materials. So, what appears to an outsider to be a wasteland of degenerate gated controls, is in fact a substitute system that works well enough to carry out normal rotations when the canals are full and the only rotations that are acceptable when they are not. 6.6 Cleaning of canals and watercourses seems to be timely and sufficient to achieve the objective of keeping the water moving, with the single and substantial exception of the LMC. 6.7 It is not possible to measure the impact of the LMC weed problem on production. One would have to separate four factors that are all contributing to reduce supply to the tailend watercourses on tertiaries in the upper reaches of this canal system, and to the distributaries on the lower reaches: (1) the water shortage itself; (2) the inherited design feature, which gave the LMC such a low slope; (3) the weeds that have come up as a consequence of low velocity flows, and ID's inability to control them; and (4) losses of water from the LMC system altogether due to infiltration down (percolation) from the better drained fields that receive it. Efficiency rates are higher on the RMC system, because it is mostly low paddyland and infiltration losses are smaller. ID reckons it could push twice as much water through the LMC were it not for the weeds. To do so, however, it would have to divert water from the RMC system. If this were a water abundant situation, where the requirements on both the RMC and LMC could be met, the weeds on the latter might depress overall production. With water scarcity, the lower discharges through weed-clogged canals are mostly compensated by increased production on farms which the water does reach. 6.8 In the face of these considerations, government's insistence on the paddy priority is inconsistent with its other behavior. Paddy is not the best use of water resources in the pre- monsoon season at Kinda, nor of the higher land on both main canals in the monsoon. But government is now enforcing paddy production through its crop targeting program, and encouraging farmers to plant paddy in non-scheduled fields as well. It could have, but chose not to, enforce the expansion of the cotton crop-as dictated by project design. That might have been equally undesirable. This issue is listed in para 6.1 as an "operational" issue, presumably amenable to improved O&M performance. In fact changes in crop preferences and pricing policies are required, not exhortation to better O&M. B36 6.9 Kinmundaung. The tank subproject is a repeat of the Kinda LMC experience. The water shortfall is more severe, and nearly one half of the farmers who expected to benefit do not get supplied, even in the monsoon season. The preference for paddy in fields that could support crops with lower moisture requirements, and government's current policy of concentrating the small amount of water available for pre-monsoon cropping also on paddy, combine to lower net economic benefits. 6.10 The other familiar operational and maintenance problems listed in para 6.1 all appear to be under control. There is no unusual weed growth, so in that sense the O&M deficit is even less alarming than at Kinda. However, water scarcity so completely overshadows all other problems that it is difficult to say much about the current level of performance with operation and maintenance practices. 6.11 Azin. Poor O&M performance is not at issue in Azin. The paddylands receiving water have no better option than paddy in both the monsoon and summer seasons, so the crop allocation is optimal. MAS is promoting a third season, but paddy is left out of that rotation. Structures, gates and canals are in satisfactory condition. The rotation between orchards and paddy functions well, and rotations within the 2,000 ac paddy zone also are handled routinely. 6.12 The Azin experience supports the claim that under conditions of water abundance, O&M problems recede. Persons familiar with fanner behavior on the south coast as well as the central dry zone add another explanation, that the residents of Mon State are better farmers than the others, even if they have never irrigated before. B 37 7. Other Conclusions 7.1 Water Scarcity, Abundance and O&M The Myanmar case studies provide two dramatic illustrations of over-dimensioned systems and inadequate water supplies. Exceptional low rainfall explains part of the water shortfall. But miscalculations of potential inflows to the reservoirs appear to have been made, especially at the tank. ID claims this is exceptional, and that other irrigation systems even in the central-dry zone are not suffering water shortages to the same extent. The inclusion of the counter-example of Azin is a reminder that O&M issues that present themselves forcefully in situations of scarcity tend to recede in importance in situations of abundance. 7.2 The Economics of Paddy Irrigation. The decline in the traded price of rice since appraisal rates of return were calculated-by a factor of about 65 percent35-is enough to drive any paddy irrigation project well below the opportunity cost of capital. Azin is reaching its production targets, but even there the projected net incremental production is too low to overcome the low rice-price. OED now re-estimates the Azin and Kinda ERRs to be about 6-7 percent, and Kinmundaung's ERR to be negative. Diversification away from paddy into the specified field crops of cotton and sesame, and into higher quality crops, would have improved the results at Kinda and Kinmundaung. But the shortfall in cropping intensity and net cultivated area at both schemes would still have kept these results marginal at best. 7.3 This impact study does not follow the lead of the two PCRs. They argued that the cropping patterns as planned at appraisal, with a sharply diminished role for paddy, would eventually prevail, not because relative prices had changed but because improved extension services, and fertilizer and chemical availabilities, would lift farmers onto a higher plane of technical fanning. In 1992 the Kinda PCR said that "no better assessment can be made under the current uncertain prevailing conditions in the country."36 Three years later conditions have improved, the farming systems seem to have stabilized, and the prospects of a radical shift in the cropping pattern without some major change in government policy has receded. To OED it seems more realistic to extend the present trend. 7.4 Water Control. There are many ungated offtakes from the main canals of the previously irrigated area at Kinda. Because of their numbers, these cannot be controlled by ID. Even the gated distributary and watercourse outlets are largely operated by the farmers. Especially in periods of water scarcity, this situation requires local institutions to control water delivery. Rules, roles, and practices protected by traditional rights have developed over the centuries and still provide the underlying governance even though a modem irrigation department and formal government structure dictates cropping patterns from the top. As a result, relative to the newly irrigated area of the LMC, farmers in the previously irrigated area make more efficient use of the irrigation system, according to ID. 35. The increase in the rice price recently projected by the Bank for 1995-1997-17 percent in 1995 falling to 5 percent in 1991 and 2 percent thereafter-is not reflected in this figure. It would not have a significant effect on the conclusion. 36. PCR, p. 15. B 38 7.5 However, as in irrigation systems in other parts of the world, it was observed that local organizations are more successful in managing equitable water delivery if collective action is necessary to accomplish operation or maintenance tasks of mutual benefit to all members. In the Sama Canal, for example, farmer watercourse groups are successful in using communally recruited water guards to move water to the tail region of the main canal, but are not similarly successful in sharing the water equitably within tertiaries and watercourses once it reaches the tail. Head to tail differences were higher within the watercourses than head to tail differences between watercourses or between distributaries in main canal operation. Collective action is required from all watercourses to hire enough water guards for the main canal. However, within the watercourse, farmers at the head of the watercourse do not depend on farmers in the tail for most operation and maintenance activities and can take a greater share of the water. A strong organization is not necessary to carry out cleaning, and groups come together smoothly for this activity. By contrast, organizing to distribute irrigation is disadvantageous for the headend farmers, and they neither want nor have been forced to cooperate. 7.6 Missing: the Impact of Maintenance on Group Cohesion. At some schemes in other countries, particularly farmer-owned schemes with long conveyance canals demanding collective action on clearing and repair for any members to benefit, maintenance requirements have the effects of encouraging members to associate at all levels of scheme O&M. That is not the case at Kinda and Kinmundaung. Construction of the dams has nearly eliminated silt deposits in the canals. Maintenance consists of weed removal and channel-bank repair. With the support of ID and the group leaders, Village Tract officials are usually expected to manage maintenance of the main and distributary canals-cleaning and repair. This works in favor of the irrigators. Not only does it reduce the burden on them of policing compliance and enforcing sanctions against free riders, it increases the labor force and lightens their load by including landless and nen- irrigators in the work force. 7.7 The LMC control structures at Kinda are of high quality, in good repair, and receiving adequate routine maintenance. Since ID does not impose strict regulation of irrigation distribution below the main canal outlet, it has not been necessary for farmers in an advantaged headend position to break structures to assure access to water on demand, with reduced benefits as one moves down the distributary and tertiary canals. In many distributaries no irrigation can be delivered to lower tertiaries in the pre-monsoon season, and even in the monsoon season the timing of deliveries to tailenders is much later than to farmers in the headend. With low maintenance responsibility and no need to monitor and control water delivery from the main canal, there are even fewer incentives than in the previously irrigated area for farmers to cooperate in equitable water distribution among tertiaries and watercourses. 7.8 Mass Mobilization. There is one maintenance activity in these irrigation systems that is unique in the study area: periodic mass mobilization of people to clean the main and distributary canals. While at Kinda this is important in the previously irrigated area, it has become a critical factor in weed control in, and hence operation of the LMC. Farmer-managed irrigation systems in many parts of the world routinely mobilize irrigators to clean and repair their main canals as well as the watercourses. Irrigation agencies have almost uniformly been unable or unwilling to follow this example. Instead, water-user fees are collected by the central government and budget allocations used by the agencies to maintain canals. 7.9 Mass mobilization for irrigation maintenance builds upon the practice of District and Village Tract authority to use citizen labor for public projects. By joining local government B 39 authority with the organizing and supervision capability of ID there is an effective mechanism for mobilizing labor. In addition, mass mobilization puts the burden of paying for maintenance squarely on the community receiving most benefits. Participation by individuals is not voluntary, but at least for these irrigation structures the benefits are visible and compliance is almost universal. 7.10 It would be useful to examine ways of institutionalizing such mass labor mobilization within the management of the canals by using only irrigators to do the work. Such maintenance responsibility could easily be more closely linked to irrigation benefits by proportioning the work to farmers according to their irrigation allocation. The right to irrigation water coupled with a responsibility for maintenance might provide incentive for improved governance and result in changes in cropping pattern and more equitable irrigation distribution. However, the magnitude of the necessary mobilization needed makes it difficult to establish and enforce local control without strong state support. 7.11 State Interventions: an Inconsistent Instrument. Myanmar stands out in OED's regional impact study by the degree of state intervention in the irrigation cycle, especially in planning cropping patterns and rotations, in blocking canals when water is scarce, and in successfully mobilizing corvee labor. Such intervention has been part of the conditions of rural life in Burma since dynastic times. Interventions at this level are driven by policy considerations, and are not always consistent with the state's economic interests: for example, whenever ASC allocates paddy to headenders, which is a common practice, and, if it anticipates a water shortage, ya- crops to tailenders-the opposite of what soil conditions would dictate. 7.12 What is remarkable is that the state does not intervene where, for many farmers, intervention is most needed-in protecting tailender rights to water against the selfish behavior of headenders on the same tertiaries. Whether irrigator groups will emerge with sufficient cohesion and mutual respect to ensure equitable treatment for tailenders is open to question. The fact that in the oldest part of the Kinda system, along the Panlaung diversion canals, the members of a watercourse still cannot guarantee equitable treatment illustrates, perhaps, the limits of self-governance. In many other parts of the world, the same breakdown of apparently agreed group covenants during periods of water shortage is found.37 In one case-Taiwan-the irrigators reverted to a two phase system, where the groups took decisions during periods of water abundance, but a public agency took control of rotations during periods of extra stress. Government-presumably the Township and Village Tract LORCs-has this choice in Myanmar. It has shown itself effective in imposing the rule of order. But it chooses not to enter the irrigation cycle at this point. 37. But not everywhere. D. Vermillion points out "Breakdown of group agreements during periods of water stress is common but not universal, as the warabundi, Balinese subak rotational distribution, Sri Lankan betma and Indonesian factor K systems often work reasonably well during periods of water scarcity." (personal commnunication, January 19, 1996) I B41 Kindat irrigation system seasonal and annual system level efficiency. Annex 1 Monsoon ( July-31 November) 1990 1991 1992 1993 1994 Water Supply Reservoir release (A-ft) 508,362 401,117 679,806 522,068 380,613 Irrigated area (A) Rice (monsoon) 68,837 72,034 83,879 81,071 78,132 Other 7,593 18,571 34,083 5,807 10,521 Crop water requirement (A-fl) Rice (monsoon) 204,331 213,821 248,981 240,646 231,922 Rice (premonsoon) 3,293 76 3,086 7,683 7,634 Other 10,124 24,761 45.444 7,743 14,028 Total Water Demand (A-ft) 217,748 238,659 297,510 256,071 253,584 Syst'mievel Efficiency for monsoon (%) 43 59 44 49 67 Average Annual Efficiency 1990-94 (%) 51 Premonsoon (I February-31 June) 1990 1991 1992 1993 1994 Water Supply Reservoirrelease(A-ft) 438,404 131,738 303,156 552,601 332,927 Irrigated area (A) Rice (premonsoon) 15,617 362 14,635 36,439 36,208 Cotton 13,033 8,863 21,876 7,745 13,021 Sesame 21,089 14,147 32,516 14,605 19,889 Chilies 9,442 5,994 22,696 12,899 11,799 Other 9,084 16,801 5,223 100 5,270 Crop water requirement (A-ft) Premonsoon rice 72,762 1,687 68,187 169,775 168,699 Cotton 33,364 22,689 56,003 19,827 33,334 Scsame 53,988 36,216 83,241 37,389 50,916 Chilies 27,885 17,702 67,029 38,095 34,846 Onion 5,568 6,022 7,383 6,883 4,741 Other 20,439 37,802 11,752 225 11,858 Total Water Demand (A-ft) 214,007 122,119 293,594 272,194 304,393 System Level Efficiency for premonsoon (%) 49 93 97 49 91 Average Annual Efficiency 1990-94 ) 69 Winter (1 December31 January) 1990/91 1991/92 1992/93 1993/94 1994/95 Water Supply Reservoir release (A-It) 107,678 111,706 223.666 94,752 45,668 Irrigated area (A) Onion 6,400 6,922 8,486 7,911 5,449 Wheat 1,770 3,250 1,497 2.782 1,950 Beans 1,529 3,629 1,767 6,120 10,828 Other 1,127 546 1,863 8,093 8,314 Crop water requirement (A-ft) Paddy (monsoon) 36,885 38,598 44,945 43,441 41,866 Onion 6,197 6,703 8,217 7,660 5,276 Wheat 0 0 0 0 0 Beans 1,481 3,514 1,711 5,926 10,485 Other 0 0 0 0 0 Total Water Demand (A-tt) 7,678 10,217 9.928 13,587 15,762 System Level Efficiency for winter(%) 7 9 4 14 35 Average Annual Efficiency 1990-94 (%) 10 Summary 1990/91 1991/92 1992/93 1993/94 1994/95 Total Annual Supply (A-t') 1,054,444 644,561 1,206,628 1,169,421 759,208 Total Annual Demand (A-ft) 439,433 370,994 601,033 541,852 573,738 System Level Efficiency for Year(%) 42 58 50 46 76 Overall System Average Annual Efficiency 1990-94 (%) 52 Note: January 1995 inflow was estimated for the winter 1994/95 season based on December 1994 data. Water demand was estimated using the Irrigation Department average monthly irrigation requirement chart since adaquate rainfall and evapotranspiration data was not available to compute crop water demand based on actual conditions. Area, Yield and Production Estimates 1993/94' Cropped Area ('000 acres) Yield (tons/hab) Incremental Production (tons) Economic b PreProject Incremental Total PreProject Post/Project PreProject Incremental Total Price A. Arca' Aread Area Area' Area Area Ratio (E + N) (1) (2) (3) (4) (5) (6) (7) (8) (9) Staff Appraisal Report 1980 Paddy Existing Irrigated Area 50.7 51.1 101.8 2.1 4.5 49.3 93.1 148.7 1.0 Newly Irrigated Area 5.6 5.6 1.7 4.5 6.3 Cotton Ec 23.5 42.9 66.4 .57 1.7 10.8 29.5 44.6 1.5 N' 7.3 7.3 .25 1.7 4.3 Sesame E 13.3 22.8 36.1 .19 .40 1.1 3.7 7.2 1.1 N 20.6 20.6 .11 .40 2.4 Chillies E 9.8 12.0 21.8 .56 .78 0.9 3.8 6.2 3.3 N 8.6 8.6 .34 . .78 1.5 Onion E 5.4 13.3 18.7 4.1 5.0 2.0 26.9 28.9 0.7 N 0 0 - 5.0 OED Impact Study 1995 Paddy E 50.7 53.8 104.5 2.8 3.6 16.4 78.4 99.6 1.0 N 5.6 5.6 1.5 3.6 4.8 Cotton E 23.5 (16.5)' 7.0 .69 .65 (0. 1) (4.6) (4.2) 1.5 N 7.3 7.3 .47 .65 0.5 Sesame E 13.3 (11.6)' 1.7 .22 .32 0.1 (1.0) 0.8 1.1 N 20.6 20.6 .12 .32 1.7 Chillies E 9.8 (2.8) 7.0 .32 .77 1.3 (0.4) 2.5 3.3 N 8.6 8.6 .32 .77 1.6 Onion E 5.4 1.9 7.3 6.2 6.7 1.1 5.2 6.3 0.7 N 0 5.1 6.7 Other E n.a. N n.a B 43 Annex 2 Notes: a. Full Development for SAR; average of 92/93 - 94/95 for OED Impact. b. Cropped area in acres, to correspond to ID scheme measurements. Yields in tons per ha, converting at 2.47. c. E is irrigated, N is dryland. d. Total expansion or reduction ( ) throughout scheme command area, entered on one line (the E line). e. For E, if Column 2 is positive, multiply Column I by difference between Columns 5 and 4. If Column 2 is negative, multiply Column 3 by difference between Columns 5 and 4. For N, multiply Column I by difference between Columns 5 and 4. f. If Column 2 is positive, multiply it by Column 5. If Column 2 is negative, mutiply it by Column 4. g. Cotton and Sesame: assume all reduction comes out of Pre-project irrigated area. Souces: SAR Areas and Yield OED Areas-Table 4.1 Yields-Field survey for Post/Project Paddy, Cotton and Sesame PCR for all others (pp. 26-27) Kinda: Farm Income Calculations (3 Ha, Farm) Production Gross Income Cost of Production Net Income Area Yield' Production Farm Gate Price' Income Per ha Total Cost Total Per ha Crop (ha) (ton/ ha) (ton) (K/ ton) (K) (K) (K) (K) ($, approx) ($, approx) 3a Left Main Canal I Pre Project w Premonsoon None Monsoon Sesame 3 0.2 0.6 47,700 28,620 3,000b 9,000 19,620 195 65 Total Net Farm Income 195 Post Project Premonsoon Paddy 3 3.1 9.3 13,150 122,295 10,550c 31,650 90,645 905 300 Monsoon Paddy 3 3.6 10.8 13,150 142,020 10,550 31,650 110,370 1,105 370 Total Net Farm Income Paddy 2,010 Premonsoon Sesame 3 0.32 0.96 47,700 45,792 4,800c 14,400 31,392 315 105 Monsoon Cotton 3 0.65 1.95 37,500d 73,125 6,000' 18,000 55,125 550 185 Total Net Farm Income 865 Incremental Net Income Premonsoon Paddy 900 300 Monsoon Paddy 915 305 Total Net Income Paddy 1,815 Premonsoon Sesame 315 105 Monsoon Cotton 355 120 Total 670 Right Main Canal Pre Project Premonsoon Paddy 1.5 2.5 3.8 13,150 49,313 10,550 15,825 33,487 335 225 Monsoon Paddy 3 2.5 7.5 13,150 98,625 10,550 31,650 66,975 670 225 Total Net Farm 1,005 Income Post Project Premonsoon Paddy 3 3.1 9.3 13,150 122,295 10,550 31,650 90,645 905 300 Monsoon Paddy 3 3.6 10.8 13,150 142,020 10,550 31,650 110,370 1,105 370 Total Net Farm Paddy 2,010 Income Incremental Net Income Premonsoon 570 300/75 Monsoon 435 145 Total 1,005 Notes: a. OED Survey Data. See Annex 2. d. Acreage of controlled and free price b. OED Survey, reduced for unirrigated conditions e. Low ir.nut cotton budget c. OED Survey f. Assumes I crop every other year. Kinda: Economic Rate of Return Calculations (Cash Flow in Million Kyats) 1 2 3 4 5 6 7 8 9 10 I1 12 13 14-22 23 24 25 26-35 36-50a ERR Staff Appraisal Report Project Benefits I Agriculture 5 23 51 70 101 258 317 359 386 417 445 469 500 500 500 500 500 500 la (ofwhich,paddy) 5 23 51 66 74 106 115 123 130 137 139 140 143 143 143 143 143 143 2 Power 23 51 35 32 35 38 41 41 41 41 41 41 61 93 72 41 41 Project Costs 3 Capital, Agriculture & Power (115) (257) (286) (240) (131) (31) 14b 4 Recurrent, Agriculture' (1) (4) (9) (15) (24) (53) (74) (94) (113) (131) (147) (161) (169) (169) (169) (169) (169) (169) 5 Recurrent, Powcr (3) (3) (3) (3) (3) (3) (3) (3) (3) (3) (3) (3) (3) (3) 6 TotalProjectNetBenefits(TPNB) (115) (254) (245) (113) (42) 76 238 278 303 311 324 336 346 369 389 420 400 369 355 21.3d OED Impact Adiustments Increase all Capital costs by 15% 7 Line 3, adjustment (-) (17) (39) (43) (36) (20) (5) 8 Line 6, adjusted (TPNB) (132) (293) (288) (149) (62) 71 238 278 303 311 324 336 346 369 389 420 400 369 355 19.4 And Reduce Agricultural Benefits by 64% 9 Line I, adjustment(-) (3) (14) (30) (42) (61) (155) (191) (2:6) (232) (251) (268) (282) (301) (301) (301) (301) (301) (301) 10 Line4,adjustment(+)' 1 2 4 7 9 23 31 39 47 55 62 70 71 71 71 71 71 71 11 Line8,adjusted(TPNB) (132) (295) (300) (175) (97) 19 106 118 126 126 128 130 134 139 159 190 170 139 125 9.7 And Reduce Paddy Benefits by 66% 12 Line Ia, adjustment (-)f (1) (6) (12) (16) (18) (25) (28) (30) (31) (32) (33) (34) (34) Line 1,adjustedTPNB (132) (296) (306) (187) (113) 1 81 90 96 95 96 97 100 103 126 156 136 105 89 7.4 Notes: a. SAR figures compressed into one column b. Power Rehabilitation Costs Averaged c. 95% on cropping d. Figures on Line 6 actually give 21.6 ERR, 21.3 is in the SAR ERR. e. 70% of 60% reduction (costs deleted) f. 66% of 34% (actual paddy) Annex C IRRIGATION O&M AND SYSTEM PERFORMANCE IN SOUTHEAST ASIA: AN OED IMPACT STUDY REVIEW OF THE DAU TIENG IRRIGATION PROJECT VIETNAM June 27, 1996 Operations Evaluation Department Abbreviations and Acronyms Bank World Bank BVND Billion Vietnamese Dong DTIE Dau Tieng Irrigation Enterprise EIA Environmental Impact Assessment ERR Economic Rate of Return HCMC Ho Chi Minh City IDA International Development Association IIMI International Irrigation Management Institute MWR Ministry of Water Resources O&M Operation and Maintenance OED Operations Evaluation Department PAR Performance Audit Report PAS Provincial Agricultural Service PCR Project Completion Report PIMC Provincial Irrigation Management Company SAR Staff Appraisal Report WUG Water User Group Contents 1. Introduction A. Background ............................3 B. Characteristics of the Command Area ............................4 C. Structure of the Irrigation System ............................5 D. Water Availability and Efficiency ............................7 2. Operation and Maintenance A. Institutional Arrangements for O&M ...........................9 1. Agency ...........................9 2. Irrigators .......................... 10 B. Operational Performance ...........................11 C. Maintenance Performance .......................... 14 1. Agency .......................... 14 2. Irrigators .......................... 16 3. Impacts A. Agricultural Impact ....................................................... 18 B. Financial Impact, Farmers' Level ....................................................... 21 C. Economic Impact ....................................................... 21 1. Command Area ....................................................... 21 2. Indirect Impacts ....................................................... 23 D. Environmental Impacts and Resettlement ....................................................... 23 4. Conclusions: Influence of O&M Performance on Agro-Economic Impacts ............... 25 5. Other Conclusions ....................................................... 27 Annexes 1. Planned, Reported and Actual Irrigated Areas ................................................... 31 2. Reestimated Production of Paddy and Groundnuts ................................................ 32 3. Reestimated Crop Budgets ................................................... 33 4. Reestimated Economic Rates of Return ................................................... 34 5. Comments on the Dau Tieng Scheme WUG and Cost Recovery Program ............ 35 Maps 8. Project Envisioned at Appraisal (IBRD 28133) .......................... ........................ end 9. Assumed Development through 1988 (IBRD 23308) ........................................... end 10. Actual Development through 1995 (IBRD 27560) ............................................... end 11. Actual Development through 1995, Canal Detail (IBRD 27559) .......................... end This report was prepared by Edward B. Rice (Task Manager), with support from Vinh Le-Si (Bank), Robert Yoder, Jayantha Perera, Annemarie Brolsma, Sinee Chuangcham and Tran Kim Thanh (Consultants), who visited this project in November 1994 and June 1995. Afi Zormelo and Megan Kimball provided administrative support. C3 1. Introduction A. Background 1.1 The Dau Tieng dam is on the Saigon River 65 kilometers north west of Ho Chi Minh City (HCMC). The Saigon River is one of a number of rivers that flow southeastward towards a confluence with the Dong Nai River, which empties into the South China Sea. The Dong Nai basin lies adjacent to, but is not part of the Mekong River delta. Dau Tieng is the largest irrigation scheme in Vietnam, and the only one where an agency of the central government maintains control of the headworks and main canals. The reservoir area above the dam was described as jungle and bush and lightly populated before the war; the command area below the dam was more densely settled but depopulated during the war by massive American bombing, ground warfare and evacuation. The irrigation scheme was to occupy parts of five districts of Tay Ninh Province and the Cu Chi District of HCMC Province. Both these areas were politicall-y charged during the conflict: Tay Ninh Province not only bordered Cambodia but was the center of the militant Cao Dai religious sect; Cu Chi was the headquarters of the National Liberation Front of South Vietnam. Apart from the Dau Tieng site's potential for a large-scale irrigation scheme, resettlement, recovery and stabilization of these sensitive areas was an important strategic objective for government. 1.2 Dau Tieng was the Bank's first operation in Vietnam, approved in 1978 twenty-one years after Vietnam joined the Bank. The renewed partnership was politically attractive to Bank management as well as to Vietnam. In March 1977, nine months after reunification, the Borrower and the Bank began discussing several prospects for a first IDA credit. Dau Tieng took priority. A cadre of irrigation engineers from north Vietnam had already been relocated permanently to the south in 1976 as the Southern Institute for Hydraulic Survey and Design, to prepare and implement this and other irrigation schemes. It is one of the few Bank operations where the pressure to lend was admitted without embarrassment by Bank management. Bank staff were encouraged to bring the appraised project to the Board as soon as possible. 1.3 The Project Completion Report (PCR)' and the Project Performance Audit Report (PAR)2 discuss in detail the inadequacy of the topographical surveys and maps and other essential design data that was assembled in this accelerated preparation/appraisal exercise. The low relief of the area, and the 1:25,000 U.S. Government maps in use (accurate only to +/- 2 meters), compromised the accuracy of planned canal alignments. Some of the technical design errors were corrected after Board approval and before construction commenced; others were corrected during construction, in a series of "ingenious" adjustments which the PAR highlights; still others were not then detected, and help explain the remaining, built-in engineering flaws which continue to limit scheme performance, although many of them are now gradually and finally being addressed. Fortunately, the serious design problems relating to the headworks and main canals were caught and corrected before construction. 1. Report No. 8239, December 11, 1989. 2. Report No. 9993, October 16, 1991. C4 1.4 IDA Credit 845-VN of $60 million was signed on August 24, 1978. The project was co- financed by an additional US$10 million equivalent each from OPEC, the Kuwait Fund and the Government of the Netherlands. In 1984 the Kuwait Fund approved a supplemental US$20.4 million credit. The IDA Credit was extended twice and closed on December 31, 1986.3 Total project costs were estimated at US$110 million at appraisal, and US$124 million at completion. The first water for irrigation was released from the dam for the 1984/85 dry season. Thus, at the time of this impact study in 1995 there had been ten years of on-farm development. B. Characteristics of the Command Area 1.5 The project as first proposed by government would have irrigated 172,000 ha: (1) from two main canals (east and west) taking off from the dam (72,000 ha); and (2) from at least eight pumping stations on the rim of the reservoir, supplying higher land designated for sugar cane (100,000 ha). The Bank recommended the sugar component be postponed indefinitely and, because of a shortage of donor funds, divided the rest into the appraised project of 42,000 ha and a second stage of 30,000 ha (see end-Map 8). For reasons explained in the PCR, 14,000 ha of the original stage I area were subsequently dropped from the design, while a comparable area from stage 2, which government started on anyway without anticipating IDA funding, was later incorporated in "the project" and made eligible for Bank reimbursement (Map 9). Thus, the 42,000 ha total area of the first stage was preserved. Government continued to expand the scheme in the rest of the stage 2 area using its own funds. For this impact study the distinction between the two stages is artificial, because they are now geographically integrated into a single scheme (Map 10). With the exception of an extension of a primary distributory from the east main canal in Cu Chi District,4 by an aqueduct presently (1995) under construction, the basic canal network is complete (Map 11). The scheme occupies the right bank of the Saigon River, and the drainage channels carry the excess water toward the Vam Co Dong River running parallel to most of the scheme's southern boundary. The majority of the proposed sugar land has been planted with rubber and other dryland crops and is no longer in the grand design. 1.6 With the 100,000 ha sugar block deferred and later deleted, the dam was over- dimensioned in relation to the two-stage 72,000 ha irrigation scheme. The surplus has been released directly into the Saigon and Vam Co Dong Rivers, and serves to support other irrigation downstream (called "indirect" irrigation from the reservoir) as well as repel salinity intrusion. In 1993, after further review of the potential for expansion of the Dau Tieng gravity scheme and the record of water-use to date, government decided to limit expansion of the present irrigated area of stages I and 2 from 45,000 ha in 1995 (see para. 3.2) to 60,000 ha.5 Almost all of the room for that expansion lies just beyond the tail ends of the distributaries in the existing network. There are no new blocks of land under consideration, except for the 2,000 ha to be developed at the end of the aqueduct. 1.7 The command area is characterized by broad plains of gently rolling topography with elevations between 6 and 29 m above sea level. All the soils in the project area were formed 3. The balance of US$303,000 was canceled in July 1989. 4. Cu Chi District is not within the Bank "project" area. 5. The 60,000 ha includes about 3,000 ha which are already being supplied, with official approval, by small, private pumps directly from the main (and, to a lesser extent, the primary) canals. The other 57,000 ha are or will be gravity fed. Cs from alluvial sediments. They are not high quality soils, and exhibit varying levels of acidity. About 75 percent of the command area had been committed to rainfed paddy in monsoon seasons before the project, though much of this was abandoned during the war. Farmers along the Saigon and Vam Co Dong Rivers and their tributaries abstracted water in the dry season for adjacent paddy fields and gardens, but there was no prior irrigation network (as there was further downriver nearer HCMC). The remainder was used for upland crops, fruit trees and sugarcane, also under rainfed conditions. In the lower part of the Mekong delta paddy farmers are now able to harvest two rainfed paddy crops a year, a rotation made possible after the introduction of shorter season rice varieties. However that rotation is not possible in the project area, where the soils are not as good. Pre-project paddy farming was mostly limited to one rainy season crop. 1.8 Tidal action moves seawater up the flat gradient of the rivers. Increased flow from irrigation drainage together with regulated reservoir releases to the Saigon and Vam Co Dong Rivers has pushed the saline intrusion about 30 km downstream from the before-project maximum penetration. Gravity diversion and pumping from these rivers indirectly provides Dau Tieng reservoir water to irrigate over 41,000 ha in five provinces (para. 3.22, first bullet). Those indirect benefits can be added to the direct benefits of the Dau Tieng irrigation scheme. 1.9 The SAR assumed that most of the project beneficiaries would be grouped into production cooperatives and cultivate their lands collectively. A production cooperative, farming 150 ha, was considered the preferred socio-economic unit in the rural sector. In the north of Vietnam the system had spread rapidly. But the establishment of production cooperatives in the south since reunification had been slow, and, after the economic reforms sanctioned by the Sixth Congress of the Communist Party in 1986 got underway, and especially after de-collectivization was approved in 1989, further progress stopped. The PCR in 1989 noted that the extent of cooperative development fell substantially short of appraisal expectations. In 1995 the prevailing mode of production is one in which independent farmers cultivate on average one hectare of irrigated land, leased on long term from the state, using water provided by a state- owned irrigation system. The 150 ha unit is preserved only as the dimension of the Water User Group. 1.10 The physical and economic recovery of this area since the war has been remarkable. Cu Chi District, once the site of a major American airbase, of a legendary grid of guerrilla tunnels beneath it, and of devastation everywhere, is now, along with many parts of Tay Ninh Province, considered among the most progressive agricultural regions in the country. Rapid economic growth in the HCMC metropolitan area has not yet led to the massive emigration of labor from the irrigated farms to the city, as it has around other Southeast Asian cities. Farming families remain more or less intact, though some young men go to the city for work during the day, returning to the fields in the evening. But conditions are changing rapidly, and the relative isolation of scheme villages in the midst of rising industrial and commercial prosperity 50 kilometers away will not last long. A corridor of business enterprise is already developing along the highway from HCMC through Cu Chi town and the middle of the scheme to the provincial center at Tay Ninh. C. Structure of the Irrigation System 1.11 A 2.1 km-long earth-fill dam about 26 m high together with a 27 km-long dike form the reservoir. The reservoir has a storage capacity of about 1I,100 Mm3. The reservoir catchment is C6 2 2,700 km , 17 percent lying in Cambodia. The estimated mean annual inflow to the reservoir is about 1,800 Mm3. Government says that actual inflows are on target.6 1.12 The west main canal is 39 km long and the east main canal 41 km long. The canal network terminology used in the rest of this report conforms to local usage. It calls the first distributaries off the mains the "primaries," the next the "secondaries," and, below the "tertiary" turnoffs, either tertiary distributaries or, in limited terminal commands, the watercourse itself with its irregular grid of farm ditches. That is the standard; in small sub-networks, some of the levels are missing. None of the big canals-main, primary and secondary-are lined except to protect weak sections. Twenty-two percent of the Dau Tieng scheme is in Cu Chi District, on the east main canal. The other 78 percent is in Tay Ninh Province. 1.13 Both the east and west main canals are designed for a maximum discharge of 70 m3/sec. The area now planned for irrigation from both canals will require about 40 m3/sec at full development (60,000 ha), giving the main canals excess capacity which to some extent simplifies their operation. The gentle slope of the command area allowed the main canal to be laid out with few check and drop structures. Off-takes from the main canal, with the exception of siphon pipes installed by individual farmers with agency permission, are all hand-operated gated rectangular orifices. 1.14 Primary canals were designed and constructed with simple metal sliding gates providing offtakes to secondaries or directly to field watercourses. Subsequent operation exposed problems with structure and canal placement for optimal distribution. Inadequacies of topographical detail at appraisal, and failure to consult with farmers and local officials, are a large part of the explanation. Some of these problems were corrected by remodeling structures; some problems cannot be corrected without major reconstruction. The provincial irrigation management companies (PIMC) have selectively approved additional offtakes from primary and secondary canals to compensate for many of the earlier errors. 1.15 The Ministry of Water Resources (MWR, formerly the Ministry of Water Conservancy) was the implementing agency for the project and responsible for construction of all sections of canal to the tertiary turnoffs, that is, to blocks of about 150 ha. The PIMCs were responsible for construction of the field-level distribution system down to blocks of about 20 ha. Farmers were responsible for digging the ditches. One of the defining characteristics of the Dau Tieng scheme is that the tertiary canals and subsidiary watercourses and ditches are poorly developed in relation to the primaries and secondaries (which since 1990 have been under PIMC control). Development of the field-level distribution system is only about 50 percent complete in the design command area. At the fringe of the irrigated areas and especially in lower regions where only paddy is grown, the farmers generally rely on field-to-field irrigation. There are exceptional locations in the paddy zone where the network of field channels is dense and provides each bunded paddy plot with direct access to water. In most locations, however, water is delivered to all fields within 200 to 300 m of primary and secondary canals, but channels to take water beyond that point are missing or incomplete. The engineering of field-level structures where construction is underway is done by provincial technical staff, who then supervise the 6. The Staff Appraisal Report projected an inflow 30 percent larger. Govemment claims its original estimate was indeed 1,800 Mm3, but that this had been set aside by the Bank in favor of the larger figure after an analysis based on another, similar basin. C7 farners in layout and construction. Farmers dig the channels but receive assistance for construction of concrete structures. 1.16 Construction of the drainage system was the responsibility of the provincial governments and started about two years after the main and primary canals. Only the main and a few of the primary drains have been completed. This consists primarily of cleaning and enlarging natural drains. Plans and designs are in place for a complete drainage system. The logic being followed for drain construction is to stay abreast of need. Priority plans prepared each year are for areas where fields are frequently inundated by flooding or the height of the water table makes cropping impossible. The PAR commented on impending problems with waterlogging, evident in 1990. The most serious of those problems appear to have been corrected by drain construction, though small areas on many of the longer distributaries have gone out of production. In some locations drains are not essential until further field development is complete, and it is cost effective to delay construction. In other locations, where additional drains in the upper reaches of the system would provide irrigation supplies to farmners on lower lands, delay in drain construction does have economic costs. D. Water Availability and Efficiency 1.17 The maximum monthly rate of irrigation release necessary to irrigate the planned 60,000 ha when the design command area is fully developed is well within the capacity of the dam and the main canals. Measurements are not available for discharge in primary and secondary canals so comparison of actual and planned capacity at those levels is not possible. However, judging from reported problems of irrigation delivery in some of the primary canals, it is unlikely that all of the primary and secondary canals can accommodate the increase in water delivery necessary to serve the planned area. In other words, the water constraint is in the delivery capacity of the primary and secondary canal systems. 1.18 The Bank is currently appraising an irrigation rehabilitation project for Vietnam that would finance construction at ten sites in the country. One of them is an incomplete irrigation scheme begun in 1960 between the Saigon and Vam Co Dong Rivers and between Cu Chi and HCMC, the Hoc Mon/North Binh Chanh Irrigation Project totaling 12,000 ha. During preparation all available meteorological and hydrological records were used to assess water availability from the Dau Tieng reservoir, in the near and distant future, to substitute for and then supplement supplies for the Hoc Mon Project from another dam yet to be built in the Dong Nai basin (the Phuoc Hoa dam). The report concluded that any development of the Dau Tieng irrigation scheme beyond 55,000 ha before the other dam was built would preclude releases sufficient to fully satisfy plans for currently agreed abstractions for HCMC, for salinity control in the Saigon and Vam Co Dong Rivers, and for irrigation requirements of the Ioc Mon Project. On the other hand the report implies that, even in years with lower than average rainfall, the Dau Tieng irrigation scheme would have sufficient water for irrigation of the full command area, up to that 55,000 ceiling. The ceiling of 55,000 ha is 10,000 above the area reported by the PIMC as irrigated in 1995, but 5,000 ha less than the maximum scheme perimeter of 60,000 ha. There is room therefore for further extension of the Dau Tieng network. However, if Hoc Mon goes ahead on schedule but the new dam is delayed, by the year 2000 the abundance of water now enjoyed at Dau Tieng will end. 1.19 In many irrigation systems water delivery depends upon adequacy of the design, and the quality of operation and maintenance. Expansion of the capacity of the primary and secondary C8 canals at Dau Tieng is an expensive way of extending the scheme to full command. For Dau Tieng farmers the relatively easy access to groundwater provides an alternative irrigation supply option. Since the start of the project, the water table appears to have risen from about 5-8 meters to about 2-3 meters throughout much of the command area. 1.20 The PCR estimates an efficiency rate for the canal system at Dau Tieng of 40 percent. An OED review of data bearing upon the measurement of actual efficiency of water use within the Dau Tieng project reveals a slightly more positive situation. The review was restricted to the east canal command. Assuming that the reservoir release was the only source of irrigation water for the entire 25,000 ha reported as irrigated from that canal in 1994, then the overall efficiency that year was about 44 percent. For the dry season (November through March) the efficiency appears to have been well over 70 percent. These levels are surprisingly high, especially when set against the poor operational performance of the canal network. 1.21 The analysis for the Hoc Mon Project assumes an efficiency rate of 60 percent. Though the basis of that calculation conforms to the normal methodology, experience at Dau Tieng suggests it may be optimistic. The basis for the Dau Tieng 44 percent efficiency estimate includes another factor, the recovery of water applied near the canal network that moves through the drains, fields and ground to farmers at the low end. 1.22 These impressive water-efficiency rates are not enough, however, to yield a satisfactory re-estimated rate of return, as shown below. Also, were it not for this return use of the water, the efficiency rating and overall performance standard of the canal operation would be dismal. 7. It is impossible to assess the irrigation efficiency of the west canal because unquantified watcr released from the dam through a channel direct to the Vam Co Dong River is included in release figures for the west canal. c9 2. Operation and Maintenance A. Institutional Arrangements for O&M 1. Agency 2.1 Management of the reservoir and main canal is the responsibility of the Dau Ticng Irrigation Enterprise (DTIE), created in 1987. The General Manager of DTIE is appointed by the MWR, and a Management Board regulates the Enterprise and looks after the interest of the water users. The Board allocates the reservoir water among the provinces, following review of their water plans and other claims such as HCMC urban and industrial uses and requirements to repel saline intrusion. While irrigation is DTIE's primary responsibility, it runs several other productive enterprises. It is responsible for developing recreation and tourism and other income- generating activities that use the reservoir. It owns and manages hotels and restaurants, in HCMC, Tay Ninh city and near the reservoir. In 1994, about 3 billion Vietnamese Dong (BVND) net profit from the reservoir enterprises was used to cover part of DTIE's 17 BVND (approximately US$1.5 million) budget for operation, maintenance and improvement of the dam and main canal structures. Another 1.2 BVND was generated from DTIE's construction groups. Only 1 BVND was available to DTIE from water fees. The deficit is made up by the MWR. It is government policy now at national and provincial levels to oblige public companies to reduce subsidies as much as practicable. 2.2 DTIE has established a monitoring and communication network to manage water discharge and regulation in the main canals and maintenance of all structures. Each of the Enterprise's 16 field stations, which are located at head regulators and other major structures on the main canals, have resident staff responsible for operation and minor maintenance such as grass and weed removal from canal banks and greasing of gates. 2.3 The PIMC, public companies controlled by provincial governments, are responsible for operation and maintenance of the primary and secondary canal network below the main canal offtakes. For example, the Tay Ninh Irrigation Management Company manages 40 primary and over 1,000 secondary canals with a total length of 1,500 km and over 10,000 structures. In principal, PIMCs control all infrastructure down to and including the tertiary turnoff-on average to units of 150 ha. Farmers are responsible below those points. But there are many canal configurations supplying terminal command areas of less than 150 ha next to the main canals, where primaries feed tertiaries directly. In these cases the farmers themselves take control of the small, primary turnoff. In Tay Ninh Province the PIMC O&M activities are administered from five Sectional Offices, plus a Sub-Section for terminal systems within the perimeter of Tay Ninh city. Since 1992, the boundaries of Sections have followed the primary and secondary canal hydraulic boundaries. Previously, they corresponded to district boundaries. The PIMC do not have an administrative relationship with DTIE. Rather they contract to pay DTIE 20 percent of the irrigation fees they collect from farmers for the irrigation delivery service DTIE provides. 2.4 The intent is that the PIMC become self-supporting. In Tay Ninh Province the PIMC's income comes primarily from farmer's fees. In 1994 a total of 1.0 BVND (approximately US$90,000) was collected and retained by the PIMC. The 3.3 BVND budget for 1995 called for C 10 income of 1.7 BVND from contract fees and 1.6 BVND to be paid by the provincial government. Thus the budgeting process of the provincial government, to some extent, controls the nature and timing of maintenance activities. Since 1993 the PIMC requires prepayment of 30% of the farmer's contract fee in order to mobilize maintenance in the early part of the season, before disbursements are made by the provincial governments. 2. Irrigators 2.5 PIMC assist farmers in service areas of up to about 150 ha to form Water User Groups (WUG). The WUG are also organized along hydraulic boundaries and are responsible for irrigation O&M in their service area. Occasionally, they assist with cleaning of secondary canals. The term "WUG" refers both to all the irrigators within the service area, and to a small team drawn from the area who manage the system. Each WUG elects a leader whose appointment is ratified by the PIMC. The term is for one year, and can be renewed. The other members of the WUG team are also elected and each is responsible for a block of about 20 ha. In principle, individual irrigators contract with the WUG leader or another team member for water delivery and pay user fees to the PIMC for the service. The "contract" spells out the irrigator's rights and responsibilities. The leader liaises with village and hamlet committees with regard to agricultural work and seeks their support in recovering water fees and settling irrigation-relatee disputes. 2.6 A WUG is a legal body, recognized by the PIMC, and comes directly under a primary- canal manager at one of the PIMC Section Offices. From time to time the manager issues directives for the WUG to carry out. The manager signs a contract with the WUG leader before issuing water to the WUG's command area. If there is no functioning WUG in the command area, PIMC may sign the contract with the hamlet leaders and collect water fees directly from farmers. It is the WUG leader's responsibility to identify farmers who are irrigating and convince them to sign a contract and pay fees. The leader signs water fee contracts with individual farmers and collects water fees from each of them. A leader gets 20 percent of the total collection of water fees as his/her team's commission. Of this amount, the WUG leader pays 2 percent to the village committee and 2 percent to the Section Office. 2.7 Fees are proposed by the PIMC, approved by the provincial government, and conform to MWR guidelines. They are fixed by area irrigated and not by volume of water supplied. They refer to the full year. They vary depending on which seasons (of three) are covered and whether the water is gravity fed or pumped. They do not specify the crop. The dry season is the one most frequently covered. The fees are low: even if collected from all irrigators, they would not come close to covering the PIMC O&M budgets. In 1995 the highest gravity irrigation fee was the equivalent of US$32/ha for dry season paddy. 2.8 The WUG leader has no powers to punish free-riders or those who damage irrigation structures. He must seek assistance from the village committee to deal with wrong-doers. A WUG leader can mobilize his fellow farmers to provide unpaid labor to maintain irrigation infrastructure, over and above the mandatory 15-day-a-year each adult Vietnamese farmer must contribute to public works. In some areas of the Dau Tieng system the village committee allows the 15 day obligation to be used partly on irrigation works; in other areas the village uses it all on other works, forcing farmers to contribute additional days to the irrigation system (or in some circumstances pay a cash equivalent). Cil 2.9 Where primary and secondary canals are long, the WUG may be formed at the tertiary canal level with the PIMC managing the primary and secondary canals. However, in cases where a secondary or even a primary canal serves less than 150 ha, the WUG may be fully responsible for operation of the irrigation delivery at that level but receive assistance from the PIMC with maintenance of the structures. The WUG relates closely to the PIMC, but plays a subordinate role to the village leadership wherever their authorities overlap 2.10 The WUG at Dau Tieng are not grass-roots organizations. In many cases they are not organizations at all. In effect, the WUG leader is the field extension of the PIMC Section Office and there is little interaction among group members outside the team. While the WUG leader has an important role, the WUG itself has almost no responsibility or authority. The PIMC may rearrange the boundaries of a WUG, remove a leader, and appoint one of its own employees instead. The Tay Ninh Provincial Irrigation Company wants to reduce the number of WUGs in the project area. Fewer WUGs mean a smaller supervision burden on the Section Office. The Company believes that if a leader is given a larger area, he/she would collect water fees efficiently as the broader commission would give him/her a more attractive income. Such an autocratic policy would diminish farmer participation and increase more external intervention and control over irrigation and agriculture. 2.11 For the OED team, with an admittedly small sample of farmers interviewed and little time to observe irrigation practices in the field, the lack of organized irrigation activities was striking. Neither group discussion nor farmer interviews identified strong cooperative effort to construct lower level canals and manage water distribution. When water delivery in the primary canal was reported to be inadequate, there was little evidence that farmers used organized group effort to improve the supply. One hard-hitting 1994 annual report from a Section Office in Tay Ninh complained that during "tense" times of water stress and irrigator conflict, the leaders disappeared. Instead, the farmers seek individual solutions by pumping from drains or groundwater or by damaging structures elsewhere. 2.12 In many countries the objective of forming WUGs is to provide a rneans for transferring responsibility for some O&M costs to farmers. While the Dau Tieng WUG identifies farmers and establishes links to a leader via a contract that carries rights and responsibilities, that leader functions as an extension of the PIMC rather than a representative of farmer interests. His role is viewed by farmers as tax collector rather than a spokesman and organizer for irrigation services. The Dau Tieng WUG system as presently structured does not foster local governance that would help build the institutions and management of activities that is essential to more effective operation and maintenance of the system. B. Operational Performance 2.13 The Dau Tieng canal irrigation system is operating well below its potential. An abundance of water in the reservoir, and oversized main canals filled to levels at or near design capacity, are important, unexploited advantages. They are offset by operational deficiencies in the primary and secondary canal systems and the limited extension of the tertiary and watercourse network. 2.14 DTIE releases water into the east and west main canals at rates adequate to maintain water levels to supply an area at least 50 percent greater than fields presently served. The head regulators on the main canals are set to provide water to the primaries "on demand," that is C 12 without limits other than the capacity of the canal system below the mains. There is no rotation from the mains canals. 2.15 A number of factors explain the shortfall in flow below the mains. First, design errors, such as sub-optimal placement of outlets to primary and secondary canals, and inappropriate alignments, sizing and/or slope of the canals, that can be traced to the original hasty construction. Dau Tieng is an exception to the practice in other countries of either oversizing the whole of the system to give extra flexibility to scheme management, or building it at a size correspording exactly to the water requirements of the expected crop rotation. At Dau Tieng the mains are oversized and many of the primaries and secondaries are undersized (partly because farmers plant paddy in areas mapped for upland crops.) Second, failure of DTIE to supply the mains consistently at full design level, despite ample water availabilities at the headworks. A few centimeters drop in the water level in the mains can have a large effect on supply through the primary gates. 2.16 Third, weeds and silt deposition in primaries and secondaries that are not cleaned at the required frequency. The sedimentation problem may be attributable in some distributaries to poor survey work and design error, which has left sections with inadequate or non-uniform gradient. More generally, the weed and silt problem is explained by the inadequate budget of the PIMC, and the absence of an institutional apparatus that can quickly press farmers into service to assist the PIMC. 2.17 These first three factors reduce the flow below design capacity. The next three factors increase the demand for water above design expectation. 2.18 Fourth, the preference mentioned above for farmers to substitute paddy for other field crops compared with the original plan, abstracting more water at the head-ends of all the tertiaries and smaller watercourses than the hydraulic design can accommodate. Fifth, the tendency of many irrigators to plant several crops in the same area, despite different moisture requirements, and then irrigate up to the levels of the most demanding crop. This is a common dry season situation. 2.19 A sixth factor is the absence of any common concern for water rotation below the mains. The general appearance of the system is of water provided on demand at each level, except on the longer distributaries where adequate supplies do not reach the most distant watercourses. Gates on the primaries and secondaries are usually left open for most of the cropping season, unless the irrigators below (behind, i.e., served by) the gate want them closed. The process of calibrating discharge measurement and controlling all gated flows to design specifications, wetness reports and cropping patterns, which is the target if not the practice in irrigation systems in other countries, is not even recognized at Dau Tieng. 2.20 PIMC and Section reports, as well as group leaders interviewed by OED, discuss the attempts to impose rotations. They are usually planned to start from the tail-end of the secondary canal and move upward. The rotations are administered by WUG leaders acting in concert. The control and distribution of water on rotation is complicated by the fact that many of the secondary canal gates are missing. As a result, straw, wood blocks and clay are used to block gates. But such blockages can easily be removed by farmers who want more water on their holdings. C 13 2.21 However, although rotations are practiced, they are uncommon and often unsuccessful. In general, farmers do not consider rotations an effective method of water distribution. in interviews they indicated that in most cases when rotation was really needed, to preserve equitable distribution of water in short supply, it has not been successful because farmers in advantaged positions always take water first anyway and the rotation soon breaks down. When rotations are practiced in a water-abundant regime, simply to increase the predictability of the delivery system, they work. When they are needed most-during water shortages-they do not. 2.22 Competition between WUGs on a common secondary, and between farmers within a WUG along a common watercourse, seem less under control in Dau Tieng than the other Southeast Asia country case studies examined by OED in the regional impact study. Group interviews and discussions with agency staff brought out repeated complaints about advantaged farmers drawing water in excess of needs. This includes such obvious methods as siphoning from, or digging a ditch through, a major canal. 2.23 But if the level of complaint is relatively high at Dau Tieng, it does not indicate a state of anarchy. Many of the complaints focused on the period of the New Year celebrations of Tet, in January/February. The Dau Tieng system is then fully closed down for four days, at the peak of the dry season. When the releases start again, the headenders take a disproportionate share, thereby prolonging tailender anxiety That commentators single out specific, time-bound events of this sort suggests that the periods of real stress between farmers in the canal network were few. The frequency of deliberate damage to structures also seems to be much lower at Dau Tieng than at schemes in the other countries in OED's regional review. 2.24 Abundant water in the Dau Tieng system provides a substantial benefit to almost all irrigators within the area of command, reducing sharply the appetite for retaliation. Even tail- enders, and a large number of the farmers with fields beyond the direct reach of any of the ditches, have been able to add a second crop despite their disadvantaged position-because of the flows of water in drainage channels, from field to field, and, of increasing importance, the rising groundwater table. 2.25 The imperfections of canal operations are reflected in the failure of PIMC and the WUGs to sign contracts covering the whole of the command. Of the 45,000 ha total reported as irrigated in the 1994/95 dry season within the Dau Tieng scheme perimeter, only about 15,000 ha, or 33 percent, was under contract. Information provided OED by village and agency officials suggests that the degree of under-reporting by the contractees themselves of their own crop area supplied directly by the scheme is about 30 percent. Thus, the coverage by canal irrigation of area farmed by contracted farmers is about 20,000 ha (roughly 130% of 15,000).8 There is a gap of 25,000 "irrigated" hectares (45,000-20,000) farmed by irrigating families without contract. 2.26 These farmers are "free riding," but not all maliciously so. Most of them are farmers with uncertain access to sufficient year-round water-within a scheme perimeter that was supposed to provide just that. The contracted farmers occupy the band of land close to the secondary and tertiary canals. Beyond that the PIMC cannot guarantee water on demand, even 8. Provincial field agents in consort with village authorities report routinely on land holdings, field sizes, cropping patterns, crop conditions and whether plots are irrigated. These data do not distinguish source of irrigation. That is, irrigators are not distinguished by whether they access the canals, drainage channels, gravity flows from irrigated fields at higher elevation, or shallow hand-dug wells. C 14 though the mains are full. Disadvantaged farners are in a position to refuse to sign contracts, because the contractual relation is supposed to guarantee delivery of water sufficient to supply designated crops and areas. PIMC is as reluctant as the farmers to sign under these uncertain conditions, because the contract obliges the agency not only to reduce fees if the supply is short but to compensate the fanner for crop losses. One altemative, to charge lower fees in non- assured areas, has not been tried. 2.27 The ability to collect the fees stipulated in a contract is another indicator of system performance. In Cu Chi District about 70% of the fees are collected and in Tay Ninh Province the PIMC has a similar level of fee collection.9 The other 30 percent are not paid for several reasons. The farmer may decide as the season approaches not to grow a crop because of poor market prices or household cash and labor constraints. If they do not grow a crop they refuse to pay. It appears that the two most common reasons, however, are that (1) farmers refuse to pay if they consider poor water delivery adversely affected crop yield, and (2) they are satisfied with the status quo, receiving a benefit without paying for it. Nevertheless, it is safe to assume that the fee-paying farmers in the contract area received the most reliable and sufficient irrigation. 2.28 Of course, contracts and fee collection are not exclusively dependent on the reliability of water supply. Persistence, hard work and the administrative skills of the PIMC, WUG leaders and village authority contribute to success in bringing free riders under contract. But the OED mission was told everywhere that a reliable water supply is the dominant factor. The two districts in the scheme with the best collection records, Cu Chi and one of the Tay Ninh districts close to the headworks, are said to owe that achievement to the mixture of effective water delivery throughout the district and strong administration. 2.29 The good side of this otherwise disappointing situation is the scope for improved operational control and water management as farmers and agencies gain more experience. Few of these farmers have more than ten years of irrigation experience. Rotations and appropriate water applications can either be learned or imposed, and so far government has shown no intention of forcing the pace. Nevertheless, everywhere OED was told that management is getting better. C. Maintenance Performance 1. Agency 2.30 In contrast to the operational record, maintenance standards are quite satisfactory. The reservoir facilities and main canal structures are in good condition. All gates and structures under DTIE's control examined by the OED team were in working order. Weeds and sediment in the main canals are presently not a problem. In 1994/5 DTIE spent over US$1.3 million to repair and improve structures.10 In 1994, as part of preparation of the irrigation project downstream 9. There are reports of fee collections at Dau Tieng averaging as little as 10%. But that refers to collections against the total design command area, and not the contract area 10. The priority large-scale repairs were: replacement of seals on the spillway gates; repair of the reinforced concrete spilIway structure using grout injection to fill cracks; strengthening the upstream slope of the dam by reinforced concrete; raising by one meter the level of the dam where wave action during wind storms threatened overtopping; and repair of several main canal off-takes and side spillways. In addition, the DTIE carried out considerable minor maintenance such as painting steel components; removing and refurbishing the reservoir outlet main gate; C 15 from Dau Tieng (para. 1. 18), a consultant engineer assessed the dam, dike and appurtenances. His findings were consistent with earlier evaluations, that the structures were basically sound and well maintained. The engineer on OED's impact team came to the same conclusion. 2.31 Three distinct levels of maintenance by the provincial companies were identified. Each year the Tay Ninh PIMC through its Section Offices carries out routine maintenance. This consists of minor repair and cleaning of weeds and silt from the primary and secondary canals. The Tay Ninh company spent about 0.2 BVND on routine maintenance in 1993-94 and mobilized about 9,000 person-days of labor for cleaning the secondary canals. While some level of routine maintenance is generally done each season, preparation for the critical winter/spring irrigation program that begins in late November is intensive. The main canal is shut down on October 31 for two weeks to facilitate routine maintenance of all canals. Secondary canals are only reopened if upon inspection the Section Office staffjudge the secondary, tertiary and field- level canals to be in good condition for water delivery. Tertiary or field-level canals are considered the property of the farmers and their maintenance is not otherwise ironitored by PIMC. 2.32 In addition to routine maintenance, emergency repairs to prevent failure or to rebuild structures that have failed are carried out throughout the irrigation season. In their annual report, one Section Office in Tay Ninh claimed they had reinforced canal embankments in 32 locations during one season. The repairs were carried out in time to prevent failure. 2.33 The third level of maintenance is to carry out more substantial repairs in select&d primaries as the budget permits. These may be reported as "rehabilitation," but the majority deal with the cumulative impact over several years of inadequate cleaning, after which the canal profile has been seriously compromised and must be reshaped. Nevertheless, this is still a job requiring mostly labor rather than heavy equipment. The task does give the PIMC the opportunity to correct some of the inherited design errors. In addition to removing silt and weeds to restore the original canal profile, weak sections of canal are rebuilt and offtake, drop, and spillway structures are repaired as needed. Budget conditions do not permit the company to enlarge the canal profiles and thereby significantly expand the flow below the mains. That would entail moving one of the embankments. 2.34 The structures and canals comprising the primary and secondary distribution system are in most cases functional. Most gates operated by the PIMC are in working condition. The team saw some gates that were not functional, and structures with gates missing altogether. That would appear to be the minority. The structures and canals show signs of wear and there is evidence of tampering but, for the most part, not wanton destruction or decay and abandonment. Sediment and weeds cause the greatest operational problems. Periodic seasonal cleaning is not sufficient to adequately overcome these problems. 2.35 In short, there is no evidence that maintenance of physical infrastructure above the tertiaries is materially deficient. However there is ample evidence that routine and emergency maintenance as well as rehabilitation are continually being carried out, though not at the scale strengthening embankments, roads, the main dam and dike; establishing new grass on embankments; removing 130 termite colonies from the dike; and maintaining access roads. The fact that the dam had to be raised one meter answers a question posed in the PCR as to whether the original dam height of 28 feet above sea level was excessive. The dam had to be raised because of water turbulence caused by unusually high winds. C 16 necessary to keep the system from slowly deteriorating. Structures do fail and must be rebuilt. But the larger problem is caused by the weed and sediment that reduces conveyance capacity in the primaries and secondaries. Labor would have to be mobilized, or financed from PIMC budget more frequently than at present to significantly remove that problem. 2.36 One of the impressive features of the maintenance program is the interest and involvement of PIMC engineers in field conditions. In both the Tay Ninh PIMC and the Cu Chi District Irrigation Management Company the engineers play a hands-on role in anticipating, observing and responding to deficiencies. The fact that they could readily identify the sections along all canals with design flaws or excessive siltation, and could talk as well about a schedule already agreed for attacking most of those problems, indicates that the problems are gradually being overcome. Also, even where farmers on the mains and primaries were siphoning or pumping water directly, that access was restricted, patrolled, and charged in the contract. 2.37 The work of the PIMC is impressive. The allocation of limited funds between operations and maintenance also is appropriate. The low turnover of PIMC staff provides a valuable resource of experience that is lacking in most irrigation systems. Staff understand their jobs and take them seriously. Establishing a job performance contest among WUG leaders is one example of an innovation to improve management. 2. Irrigators 2.38 Farmer maintenance of their canals and structures reflects a different set of incentives than those that face the agencies. There is no evidence that the irrigators were incapable or neglectful of maintaining their watercourses when cleaning and minor repair were necessary to keep the water moving. The team observed ditches in all conditions, from recently cleaned to clogged and overgrown. But the timing of OED's visit was out of phase with the periods of acute water demand, and the claims that the clearing of the watercourses would be carried out in time for the next supply were convincing. The fact that the field network has not been extended at the rate anticipated, so that only 50 percent of the potential grid is actually in place, is not a maintenance problem. It is explained by other factors: reluctance of farmers in the network to permit neighbors to dig channels through their property to connect to the system; absence of mechanisms or motivation at the village level to induce farmers on the network to allow outsiders right of way, especially if competition for water and losses to those already connected might result; reluctance of farmers outside the network to connect and risk having to enter a contract if they feel they are reasonably well served by drains, field-to-field flow and groundwater; and a lack of confidence that headenders will cooperate. 2.39 Balancing the interests of advantaged and disadvantaged farmers plays a big part in explaining the one category of maintenance where the farmers do poorly-which is the lack of care almost everywhere for the gates that were part of the original architecture. Many of the gates, in structures designed for gates, from the tertiaries on down are missing, and the structures are often in a state of disrepair. This is not a failure of maintenance through indifference, incompetence or lack of funds. This is the result of deliberate interference or deliberate neglect. In either case it is a sign of protest against the main purpose of the gates, which is to restrict access to water by farmers below the structure. 2.40 The social bonds between headenders, tailenders and farmers on the fringe in the villages within the Dau Tieng perimeter are too weak to oblige advantaged families freely to reduce their C 17 favored position, or for the disadvantaged to successfully press their claims in the WUG or village inst