Madagascar - Second Irrigation Rehabilitation Project

Document of The World Bank FOR OFFICIAL USE ONLY Report No: 20756 IMPLEMENTATION COMPLETION REPORT (26440) ONA CREDIT IN THE AMOUNT OF US$21.2 MILLION TO THE REPUBLIC OF MADAGASCAR FOR A SECOND IRRIGATION REHABILITATION PROJECT MARCH 30, 2001 Rural Development 2 Country Department 8 Africa Region This document has a restricted distribution and may be used by recipients only in the performance of their official duties. Its contents may not otherwise be disclosed without World Bank authorization.  CURRENCY EQUIVALENTS (Exchange Rate Effective December 2000) Currency Unit = MALAGASY FRANC FMG I = US$ 0.000151 US$ I = FMG 6551 FISCAL YEAR January I December 31 ABBREVIATIONS AND ACRONYMS AFD French Development Agency (Agence Frangaise de Developpement) CAS Country Assistance Strategy EDF European Development Fund ERR Economic Rate of Return GoM Goverment of Madagascar ICR Implementation Completion Report MoA Ministry of Agriculture O&M Operations and Maintenance PAE Environment Support Project (Projet d'Appui A l'Environnement) PNVA Agricultural Extension Program Support Project (Programme National de Vulgarisation Agricole) PPI Small Scale Irrigation (Programme Pdrimetres Irrigues) PSR Project Status Report RSDP Rural Sector Development Program SAR Staff Appraisal Report WUG Water Users'Associations Vice President: Callisto Madavo Country Director: Hafez Ghanem Sector Manager: Joseph Baah-Dwomoh Task Team Leader: Ousmane Seck  FOR OFFICIAL USE ONLY CONTENTS Page No. 1. Project Data I 2. Principal Performance Ratings I 3. Assessment of Development Objective and Design, and of Quality at Entry 4. Achievement of Objective and Outputs 3 5. Major Factors Affecting Implementation and Outcome 8 6. Sustainability 9 7. Bank and Borrower Performance 12 8. Lessons Learned 14 9. Partner Comments 15 10. Additional Information 15 Annex 1. Key Performance Indicators/Log Frame Matrix 16 Annex 2. Project Costs and Financing 17 Annex 3. Economic Costs and Benefits 19 Annex 4. Bank Inputs 20 Annex 5. Ratings for Achievement of Objectives/Outputs of Components 21 Annex 6. Ratings of Bank'and Borrower Performance 22 Annex 7. List of Supporting Documents 23 This document has a restricted distribution and may be used by recipients only in the performance of their official duties. Its contents may not be otherwise disclosed without World Bank authorization.   Project ID: P00 1522 Project Name: IRRIGATION II Team Leader: Ousmane Seck TL Unit: AFTR2 ICR Type: Core ICR Report Date: April 3, 2001 1. Project Data Name: IRRIGATION II L/C/TF Number: 26440 Country/Department: MADAGASCAR Region: Africa Regional Office Sector/subsector: Al - Irrigation & Drainage KEY DATES Original Revised/Actual PCD: 03/31/88 Effective: 12/14/94 12/14/94 Appraisal: 03/19/93 MTR: 07/29/97 07/29/97 Approval: 07/12/94 Closing: 06/30/2000 09/30/2000 Borrower/lmplementing Agency: GOVT. OF MADAGASCAR/MINISTRY OF AGRICULTURE Other Partners: STAFF Current At Appraisal Vice President: Callisto Madavo Edward Jaycox Country Manager: Hafez Ghanem Francisco Aguirre-Sacasa Sector Manager: Joseph Baah-Dwomoh Nils Tcheyan Team Leader at ICR: Ousmane Seck Christopher Trapman ICR Primary Author: Ijsbrand H. de Jong 2. Principal Performance Ratings (HS=Highly Satisfactory, S=Satisfactory, U=Unsatisfactory, HL=Highly Likely, L=Likely, UN=Unlikely, HUN=Highly Unlikely, HU=Highly Unsatisfactory, H=High, SU=-Substantial, M=Modest, N=Negligible) Outcome: S Sustainability: UN Institutional Development Impact: M Bank Performance: S Borrower Performance: S QAG (if available) ICR Quality at Entry: U Project at Risk at Any Time: No 3. Assessment of Development Objective and Design, and of Quality at Entry 3.1 Original Objective: The project's objectives were: (a) promoting growth by increasing agricultural output through better water management and expanded agricultural extension; (b) reducing poverty by improving productivity to increase both rural incomes and food security; (c) enhancing environmental management by developing techniques for assessing environmental irrigation rehabilitation works and improving local assessment capacity; and (d) strengthening project implementation capacity by transferring management of irrigation schemes to water users.  Both the Second Irrigation Project (PPI-2) and its predecessor the Irrigation Rehabilitation Project (PPI-1) were inspired by high recurring Government expenditures on operation and maintenance (O&M) of irrigation schemes under the responsibility of the Ministry of Agriculture (approximately 270,000 hectares in 1985). For example, in 1985, at the start of PPI-1, half of the operational budget of the Ministry of Agriculture was allocated for O&M of these schemes. In 1994, at the start of PPI-2, half of the Ministry's investment budget was allocated for rehabilitating irrigation schemes. In addition, the PPI-2 was undertaken because long-term donor contributions to irrigation rehabilitation failed to have an impact on production and productivity levels in irrigated agriculture, and had not achieved sustainable irrigation O&M. Rehabilitated irrigation schemes continued to require government O&M assistance. High levels of expenditure continued to be needed for rehabilitation, because of lack of maintenance, high erosion and sedimentation levels, and recurrent cyclones and flooding. Accordingly, the PPI-2 project aimed to transfer O&M of selected irrigation schemes to Water User Groups (WUGs), to promote sustainable irrigation and reduce dependence of irrigation schemes on government O&M funds. Farmers, organized in WUGs, were to operate and manage their own irrigation schemes, with the Government acting as service provider. This was to be facilitated by infrastructure investments in better water management, following a 'participatory and progressive' approach. The participatory and progressive approach is spelled out in detail in the Staff Appraisal Report (SAR), and distinguishes three phases. During the first phase, water users define, through detailed studies, the works to be executed during subsequent phases, by themselves and by contractors. The efforts of water users serve as indicators of their long-term commitment to O&M. When the commitment of water users has been demonstrated, the second phase can begin during which a plan for agricultural development and the different stages of irrigation rehabilitation is formulated. A detailed design of the rehabilitation works that are beyond the capacity of the water users is prepared. Finally, during the third phase, works are carried out by water users and contractors, respectively. The contribution of water users determines if further works will be executed by the contractor. In the SAR, the savings. to the Government of Madagascar (GoM) of the irrigation management transfer were estimated at USD 500,000 per year. In order to ensure the sustainability of the transfer, the SAR recognized that the following issues, among others, had to be addressed: developing local capacity and training of WUGs to ensure proper management, implementing environmental works to reduce maintenance needs, and expanding extension to increase the capacity of farmers to cover O&M costs. Under PPI-1 and PP;-2, 43,710 ha out of 270,000 ha originally under Govemment O&M responsibility have been rehabilitated and transferred to WUGs. O&M of the balance of (270,000 - 43,710 =) 226,290 ha still remains Government responsibility. Government expenditures on O&M of irrigation schemes are therefore still high. The objective of transferring O&M responsibility and scheme ownership to WUGs is therefore still appropriate and in line with the Government strategy for the sector. 3.2 Revised Objective: During a 1996 supervision mission, large scale irrigation was included in the project, and project coverage was expanded from the initially identified nine zones to the national level. Land security was added as activitiy, as well as outsourcing of WUG mobilization to NGOs. This was done in view of the slow disbursement of the project, and to increase its impact. In September 1998, funds were set aside for financing emergency activities to control the outbreak of migratory locusts which had infested almost four-fifths of the country. 3.3 Original Components: The project consisted of six components: (a) consolidating the rehabilitation and transfering O&M responsibilities for 16,000 hectars of irrigation schemes rehabilitated during the first project, including emergency repairs of cyclone damages; (b) assessing and executing environmental protection works to ensure the sustainability of rehabilitated irrigation schemes, including preparing guidelines for emergency rehabilitation works; (c) rehabilitating 4,000 hectars of new irrigated schemes; (d) strengthening the Ministry of Agriculture's extension, rural - 2 -  works and programming and planning services, at the provincial, district and local levels to improve local supervision of rehabilitation efforts and to support the WUGs; (e) training of project staff and farmers with an emphasis on developing local capacity to ensure sustainable activities for subsequent phases of PPI-2; and (f) monitoring and evaluation of project impact and verification of accounts. 3.4 Revised Components: During a 1996 supervision mission, the project was reorganized. Rehabilitation and O&M transfer of large-scale schemes, as well as national coverage of the project, was included in order to improve the project's impact. In addition, the issue of improving land security was included to facilitate management transfer to WUGs, as well as to take away an impediment to implementation of environmental activities. In September 1998, 4.21 million USD were set aside for financing emergency activities to control an outbreak of migratory locusts. 3. 5 Quality at Entry: PPI-2 built on the results and recommendabons of PPI-1. The project's objectives were also consistent with the Bank CAS and with Government priorities. The project became effective in December 1994, and focused on the emergency repair works after the four 1994 cyclones. At the same time, the project worked on the elaboration of an intervention strategy. However, project disbursement was slow during the first two years. This was mainly related to the long time that the project had spent on elaborating the participatory intervention strategy, and the unpredictability of the outcome of such a strategy. The project was consequently reorganized to increase project impact by (i) no longer exclusively focusing on medium-sized schemes but also on large-scale schemes, and (ii) by enlarging the coverage of the project to the entire country. The slow initial disbursement and subsequent reorganization of the project raises a number of issues related to the quality at entry. The key question that can be asked is why the project was reorganized for reasons of slow disbursement. While the 'participatory and progressive' project approach provided a sound framework for employing participatory infrastructure delivery as a vehicle for mobilizing WUGs, this process depended to a large extent on beneficiary commitment. Project outcomes of participatory processes are notoriously unpredictable and need appropriate, flexible and demand-driven funding operations to ensure that the level of commitment of stakeholders determine the pace of investments, and not whether or not disbursement is delayed. It is felt that a more flexible project disbursement schedule could have assisted the project in preventing that disbursements considerations determine project activities. In addition, project outcomes were impacted by the absence of cohesion in the SAR between the different sectors involved in agricultural growth. Links between irrigation, extension, watershed management, market access and input supply were to a large degree fortuitous and not part of a holistic approach. In fact, extension and environmental activities were covered by separate projects (PNVA and PAE, respectively), and collaboration between the three projects was not very effective in the first years of project implementation. Although reference is made in the SAR to the extension and environmental project, no strong mechanism for collaboration or coordination of activities between the three projects was identified. Activities to improve market access and input supply have not received a systematic treatment in the SAR. This resulted in a negative impact on the sustainability of project outcomes. The Quality at Entry of the project is therefore rated as unsatisfactory. 4. Achievement of Objective and Outputs 4.1 Outcome/achievement of objective: Performance indicators were established in the SAR and reviewed during the first supervision mission. A summary is presented in Annex 1. Performance indicators were not revised throughout the existence of the project. For a discussion on their relevance, see section 4.2.f. An extemal impact evaluation of the - 3 -  project was carried out as part of the Government contribution to the ICR (see Annex 7). Although the project suffered from delays during the first two years of its existence, disbursements and acreage covered by rehabilitation significantly increased following the 1996 supervision mission. At the end of the project, a total of 25,295 hectares were rehabilitated, as compared to 20,000 hectares foreseen in the SAR. An additional 15,385 ha, already covered by the PPI-1 project, received further support following them 1994 cyclones. Variation orders were executed on an additional 3,030 hectares, also covered under the PPI-1 project. (a) Promoting growth by increasing agricultural output through better water management and intensified agricultural extension. This objective has been achieved. Between 1994 and 1999, main season rice yields in rehabilitated schemes increased by an average of 36%, from 2.2 tons per hectare to 3.0 tons per hectare, close to the target of 3.3 tons per hectare. Weighed average adoption rates of improved rice technologies amounted to 53%. The area under improved water management covered 25,295 hectares at project completion, more than the target of 20,000 hectares. The average main season production amounted to 75,885 tons, as compared to the SAR target of 66,000 tons. More than three quarters of the area that was rehabilitated had irrigation of good or average quality. Acreage of dry season rice and other cultures increased by close to 4,000 hectares and 1,000 hectares, respectively. Actual cropping intensity equaled the target of 1.3 on average, as compared to 1.0 at project start. (b) Reducing poverty by improving rural incomes and food security. The objective has been achieved. The average income increase from agriculture in schemes that benefited from project interventions was estimated at 203 US$/hectare. Food security was marginally improved through an increase in non-rice coverage of close to 1,000 hectares. (c) Enhancing environmental management by developing techniques for assessing environmental irrigation rehabilitation works and improving local assessment capacity. This objective was partly achieved. An Environmental Unit was created in 1994 at the level of the Rural Works Directorate, and later directly linked to the project. In addition, local capacity for environmental assessment was reinforced through the establishment of 15 Regional Environmental Offices (at the level of the regional project offices) and 64 Local Environmental Units (at scheme level) in 1997. A guide for environmental appraisal of irrigation rehabilitation programs was published by the project in 1996. Government officers received training in environmental protection and techniques, including environmental diagnosis, agro-forestry, watershed development, environmental protection of hydraulic structures and development of agro-biological sites for improved soil fertility management. Agro-biological experiments were conducted to demonstrate how to increase rice production levels by greater use of organic manure in rice fields. These demonstrations made clear that with simple tools (covering fields with straw, minimum tillage, plowing under straw), lowland rice production increased to 11-14 tons per hectare, upland rice yield could increase by up to 600%, and variability of yields could decrease by up to 300%. The number of farmers adopting the recommended technology was very low because the new technology required more cash (up to 300% more) and more labor (around 35% more) as compared to more traditional practices. However, throughout project implementation but in particular following the expansion of project activities to large scale schemes, it became clear that environmental issues related to sustainable irrigation development needed to be addressed in a more comprehensive way. In fact, the opposite occured and project objectives were not amended to reinforce local capacity to increase project impact on the environment. In the end, the impact of environmental activities was just too small to make a difference. Interestingly, and contrary to the claim that the environmental component failed to make an impact because of the limited budget allocation, only 30% of the budget line related to this objective was disbursed. (d) Strengthening project implementation capacity by establishing operational procedures for transfering management of schemes to water users. This objective has been achieved. Procedures for management transfer were established and implemented by the project. At project completion, 148 -4 -  WUGs undertook O&M activities. This responsibility was taken over from Government by these WUGs since project start. Out of these 148, only 14 WUGs have benefited from actual irrigation management transfer. This low number of 14 WUGs having benefitted from formal irrigation transfer (not to be confused with the 148 WUGs actually undertaking O&M works) is related to slow registration procedures and the relatively late signing of an agreement between the project and the Ministry of Land. This in turn was due to the fact that priority was given (rightly so) to de facto O&M capability, and that formal O&M transfer came more as an after thought. The Ministry of Land had not been involved in the project. At project completion, Government capacity to continue the process of irrigation management transfer has been strengthened by the establishment of operational procedures. These include informal strengthening, training and exchange visits, as well as the more formal part of registering WUGs. (e) Locust control. Following the invasion of four-fifths of Madagascar by locusts, USD 4.21 million was allocated in 1999 to contribute to the financing of emergency activities to control the outbreak of migratory locusts. The locust control program, executed by the National Committee for the Fight against Locusts, came to an end in late 2000, when the problem was under control. In conclusion, project outcomes were generally achieved. However, the sustainability of irrigation O&M remains a concem, a conclusion similar to the one that was reached at completion of PPI-1 and that inspired the PPI-2 preparation process (see section 6). The rating is therefore satisfactory. This is based on the following assessment: * the target acreage rehabilitated was exceeded; * the rehabilitation strategy was a sound example of participatory infrastructure delivery with the aim to mobilize WUGs for taking responsbility for O&M; * productivity levels increased and were close to target levels, total production exceeded the target production, the cropping intensity met its target of 1.3, and adoption rates of new rice production technologies were satisfactory; * the locust control objectives were met with the recent termination of the locust emergency activities. 4.2 Outputs by components: As mentioned above, the overall assessment is satisfactory. By project components, the assessments are as follows: (a) Consolidation of rehabilitation and transfer of O&M responsibilities on 16,000 hectares of irrigation schemes rehabilitated under the first phase project, including emergency repairs of damages caused by cyclones. This component was successful mostly in tenms of acreage. In total 15,385 hectares were rehabilitated following the devastation of the 1994 cyclones. The total costs of the operation amounted to USD 1.8 million, or USD 117 per hectare (as compared to USD 207 per hectare in the SAR). Larger scale schemes benefited from lower per hectare investments as compared to smaller scale schemes (e.g., Analavory (140 ha) with 740 USD per hectare, as compared to Soavina (1,200 ha) with 35 USD per hectare). Transfer of O&M responsibilities was not fully achieved and the 148 WUGs that were created and received assistance were, at project completion, not fully able to carry out O&M in a sustainable way. This will be discussed in section 6. (b) Assessment and execution of the environmental protection works needed to ensure sustainability of rehabilitated irrigation schemes, including the preparation of guidelines for implementating emergency rehabilitation works. The Environmental Guide prepared by the project assisted in assessing environmental works needed for the sustainability of irrigation schemes. A variety of environmental protection works were implemented. These included the establishment of nurseries for vetiver and azolla, planting trees and other plants to reduce erosion, and fertilizer tests. Environmental protection works to enhance irrigation infrastructure achieved some localized successes (vetiver planting), but the component failed in general to have an impact on the sustainability of the irrigation schemes. A number of causes were identified in successive PSRs: (i) the fact that the scale of environmental degradation exceeded the implementation and analytical capacity of the project (e.g., forest fires), (ii) environmental protection does not only depend on training and sensitization, but also on an appropriate institutional context, in particular with respect to land tenure, and on the concerted effort - 5 -  of all involved (iii) the scope of the environmental works was not modified when, at the time of the 1996 supervision mission, it was decided to include large scale schemes in the project. On many occasions, project activities related to environmental protection remained isolated attempts (sometimes only measurng a few square meters) to control erosion in watersheds of several thousands of hectares. Not surprisingly, the negligible results and high labor investments discouraged farmers from undertaking additional environmental protection works. (c) Rehabilitation of 4,000 hectares of new irrigation schemes. This component exceeded by far its target. In total, 25,295 hectares have been rehabilitated. In addition, the costs per hectare are markedly lower than foreseen at the time of appraisal (154 USD per hectare as compared to the originally foreseen 1,778 USD per hectare). The reasons are: * The scope of works was smaller and corresponded to only one cycle of the participatory and progressive approach. Full implementation of all works for a per hectare amount that would correspond to the one anticipated in the SAR, would only occur after several cycles of participatory and progressive project approach. * Overall average investments were distorted because per hectare investments in large-scale schemes were generally lower than per hectare investments in small-scale schemes (e.g., Ranomena Iboaka (252 ha) with a per hectare investment of 1,800 USD or Reambe (94 ha) with 1,624 USD per hectare, as compared to Sahamaloto (6,937 ha) with 72 USD per hectare and Anony (10,100 ha) with 147 USD per hectare). (d) Strengthening the Ministry of Agriculture's agricultural extension, rural works and programming and planning services, at the provincial, district and local levels in order to improve local supervision of the rehabilitation and the support to the creation and performance of WUGs. This component was successful in general. After the reorganization of the project, all project activities were implemented through the regional directorates of the Ministry of Agriculture, including Rural Works and Extension. These regional directorates operated with their own decentralized operational project budget. Regional Environmental Offices were established at regional project office and scheme level and were reinforced through training and technical assistance from the center. These offices were strenathened by gaining experience from implementation of activities. However, in contrast lo programming and planning of project activities, launching of bids and selection of contractors remained comparatively centralized. In many cases, civil works were outsourced to contractors from the c, pital whereas competent local contractors were available for at least some works. Further strengthening of provincial and district capacity could have been achieved by outsourcing civil works in a decentralized manner. This would have accelerated administrative procedures, reinforced the development of small local contractors and assisted in strengthening the decentralized Rural Works Directorate. (e) Training of project staff and farmers on developing local capacity and talents to ensure sustainable activities for subsequent phases of PPI-2. Project staff were trained in agricultural policy, project management, administration, finance and procurement, organization of farmers, participatory appro3ches and environmental diagnostics, techniques and protection. Farmers were trained in SRI, cropping practices, soil fertility management and composting, cultivation of dry season crops, management of WUGs, administration and bookkeeping. National and international study tours were organized. Government staff of regional directorates were trained in bidding procedures. The impact of training has been largely positive, in particular the national level exchange visits between farmers and WUGs. (f) Monitoring and evaluation of project impact and verification of accounts. Project monitoring and evaluation was done satisfactorily. Budget expenditures on this line item far exceeded budget allocations (see Annex 2). A large number of performance indicators on WUGs progress and performance were monitored, including membership, payment of O&M fees, labor contributions to O&M, and contribution to investment costs. However, the valuation of the farmers' contribution and heir maintenance work was not always done in a systematic way. E.g., conflicting data were used on the value of one labor day. - 6 -  Many other performance indicators as identified in the SAR appeared not to be appropriate for the project, for example: 'irrigation area fully managed by water user groups' was a performance indicator, but there is no indication of what was understood by 'fully managed', nor was there any attempt to define this; 'volume of incremental triticale production' was a performance indicator, but the project never seems to have taken any particular measures to increase production of this crop, and no production data were collected. In addition, no output indicators were developed or monitored. (g) Locust Control. Following the invasion of four-fifths of Madagascar by locusts, USD 4.21 million was allocated in 1999 to contribute to the financing of emergency activities to control the outbreak of migratory locusts. Expenditures were made to purchase insecticides and materials, rent airplanes, and pay services and part of the administrative budget of the National Committee for the Fight against Locusts. Out of the funds allocated by the PPI-2 project to purchase insectides, 785,798 USD of have been spent on Fipronil, 532,550 USD on Deltamethrne, 458,903 on Propoxur and 278,648 USD on other pesticides. No information was available on the expenditures for other items under the locust control program. The relevant Amendment to the Development Credit Agreement specified the environmental concerns that the Borrower needed to respect. This included a prohibition on spraying insecticides in environmentally sensitive areas, use of quick knock-down and low persistence insecticides, mitigation of safety and health risks, and monitoring and evaluation of the impact on fauna and flora and of safety of handling operations. In this respect, use of Fipronil has received criticism in view of its relatively low knock-down effect, its high persistance and therefore its impact on non-targetted populations, in particular insectivores and, as a secondary threat, their predators. It has been recommended to no longer use Fipronil as full cover or barrier spray. The program was implemented in full cooperation with all donors. This financing contributed to controlling the invasion (that risked compromising the results of the on-going operations in the Agricultural and Livestock sector). An audit of the expenditures under this budget line raised issues related to sole source purchases made prior to the non-objection of a procurement accredited staff member. However, there have been no cases of misprocurement. 4.3 Net Present Value/lEconomic rate of return: The economic rate of return of the project calculated at appraisal was 27%, based on a cost/benefit stream over 10 years. The ERR calculation considered all the investments under the first PPI as sunk costs and hence did not include any of this initial investment costs. As part of an impact evaluation of irrigated perimeters, conducted in the year 2000 by independent consultants, some ERR calculations were made for about 10 perimeters rehabilitated under both PPI 1 and PPI 2. These ERR calculations included the costs of the PPI rehabilitation (excluded from the SAR calculations). Their estimates show rates of return of between -5 to 23 percent with only two out of the 10 rehabilitated schemes obtaining an ERR of 20% or more. Close to 40% of the schemes show an ERR of more than 14%, and 60% of the schemes have an ERR above 10%. Apart from these individual perimeter ERRs there has been no separate calculation of an ERR for the project as a whole. It is difficult to compare the ERR calculated in the SAR with those estimated as part of the impact evaluation even though the impact evaluation ERRs give a good base for estimating the impact of the investments. Using the same assumption used in the SAR that the initial PPI investments were sunk costs, the ERRs calculated by the independent consultants would have been considerably higher than what they found. Thus comparing the ERR calculated at appraisal and those at completion (using the same methodology) one could conclude that returns were satisfactory though the appraisal ERR calculation may have underestimated the costs. The external evaluation also revealed that the relationship between investments and returns is not linear. For example, some small works can solve a problem crucial to overall scheme functioning and thus decrease the time needed to bring total investments to maturity. Programming and prioritizing rehabilitation works therefore are important instruments to increase returns on investments, in particular in view of the progressive character of the project approach. - 7 -  4.4 Financial rate of return: No financial rate of return was calculated for this project. 4.5 Institutional development impact: The institutional development impact of the project is rated as modest. The positive results that were achieved are as follows: 1. VWUGs were established for all schemes that were rehabilitated. All 148 WUGs thus established or now responsible for O&M, all WUGs undertake at least some of the O&M work, and some show an excellent performance. The sustainability of the WUGs is discussed in section 6; 2. Programming and planning activities were implemented by the Regional Directorates of the Ministry of Agriculture, including Rural Works and Extension, and Regional Environmental Offices, established under the project. All of the above benefited from training and technical assistance. This experience is of particular importance in view of the continued process of decentralization in Madagascar, and the need to create decentralized programming and planning capacity; 3. Following the reorganization at the time of the 1996 supervision mission, project activities were outsourced to NGOs and private sector, in particular mobilization of WUGs, civil works and land titling. In view of the importance of private sector involvement in sustainable irrigation development, this outsourcing is a crucial contribution to local capacity building; and 4. Project M&E was conducted in a satisfactory way. At the same time, the project had some negative institutional development impacts, or failed to have any positive effects: 1. Notwithstanding extensive training, WUGs appear only able to assume responsibility for O&M in a sustainable way to a limited extend (see section 6 for further details); 2. Although decentralized units played an important role in the conduct of studies and preparation of documents, the actual launching of bids and selection of contractors remained a relatively centralized opetation, more so than the quality of local contractors justified. 5. Major Factors Affecting Implementation and Outcome 5. 1 Factors outside the control of government or implementing agency: Due to the invasion of four-fifths of Madagascar by locusts, USD 4.21 million was allocated in 1999 to contribute to the financing of emergency activities to control the outbreak of migratory locusts. Another factor outside the control of the government was the occurrence during the 1999/2000 rainy season of three cyclones that caused severe damage in a number of schemes that were rehabilitated under the project. In total, 22 schemes were damaged, covering a total area of over 9,000 hectares. The rehabilitation works were estimated to cost USD 2.3 million but, due to lack of project funds, they have not been executed under PPI-2. IDA released supplementary credit after PPI completion. 5.2 Factors generally subject to government control: One of the risks that was identified at the project start was that without extension, production levels would not increase, as has happened during the first project. Measures were therefore taken to encourage extension staff to deliver extension service in rehabilitated schemes. In practice, however, extension staff suffered from a lack of motivation due to delays in payment of salaries of non-permanent staff. The collaboration between PNVA and PPI was in the beginning of the two projects characterized by competition, rather than teamwork. This did improve towards the end of the projects. Finally, rapid erosion of O&M recovery rates resulted from, among other things, a lack of legal incentives to take appropriate action against defaulters. This has played a role in the often weak performance of WUGs. The need to enact appropriate legislation is therefore of paramount importance for WMJGs establishment and proper functioning. 5.3 Factors generally subject to implementing agency control: - 8-  During the first two years of its existence, the project spent excessive time on the establishment of an intervention strategy, the key elements of which were already elaborated in significant detail in the SAR. In the end, the strategy as developed by the project missed crucial elements, including in particular: * A strategy for withdrawal from the villages, with successive benchmarks for withdrawal discussed with stakeholders at the start of the intervention. This would have avoided situations in a number of schemes in which, the project had difficulty in 'cutting the umbilical cord' with WUJGs, thereby involuntarily perpetuating and deepening the dependence of these WUGs on the project; * A strategy for implementing the three year's participatory and progressive intervention strategy towards the end of the project, when there was insufficient time to carry out the full cycle. One category of delays in project implementation involved the official transfer of responsibility for O&M and ownership of irrigation and drainage infrastructure to WUGs, because the Ministry of Land delayed issuing titles to benefiting WUGs. This resulted from the fact that formal O&M transfer (as opposed to de facto capability for O&M) was not part of the project, and hence the Ministry of Land was not officially involved, nor properly facilitated. Long delays in adopting a modus operandi resulted from administrative problems and limited operational capacity at the level of the Ministry of Land. As a consequence, only 14 out of 148 WUGs officially benefited from formal irrigation management transfer at project completion. Similar considerations apply to the issuing of ownership titles to users following the 1996 supervision mission to strengthen the impact of environmental activities. Administrative delays of over. 24 months in the responsible Ministry have resulted in only 220 users having received 529 titles for 130 hectares, concentrated in just one scheme. As before, this was related to the oversight of the Ministry of Lands as partner in this process, and the resulting lack of facilitation. According to the Credit Agreement of the Project, a study was planned to propose guidelines for the creation, financing and operation of a permanent emergency rehabilitation fund to deal with damage caused to irrigation schemes as a result of natural disasters. However, this study was never done. When the Terms of Reference of the study were drafted, there was insufficient time to conduct it, and the study was therefore cancelled. 5.4 Costs andfinancing: At the time of appraisal, the cost of the project was estimated at 25.7 M USD, with an IDA contribution of 21.2 M USD, a beneficiary contribution of 1.7 M USD, and a Government contribution of 2.8 M USD. The latest cost is 24.5 M USD, financed by 20.4 M USD from the IDA credit, 0.6 M USD from the beneficiaries, and 3.5 M USD from the Government. This includes the amounts allocated to locust control and cyclone relief works (see Annex 2 for details). 6. Sustainability 61 Rationalefor sustainability rating: This section discusses the sustainability of the project. Project sustainability was negatively affected by the following factors: * the decision to include large-scale schemes, and the fact that the validity of the participatory and progressive approach is questionable for larger-scale schemes; * similarly, the environmental sustainability was compromised by the inclusion of large-scale schemes; * the allocation of project funds to locust control; * the occurrence of cyclones in 1999; * the absence of a time frame for the intervention, and the subsequent perpetuation of project assistance to WVUGs; * long delays in official O&M transfer and registration of land, as a consequence of a lack of involvement of the Ministry of Land; * the weak level of upfront contribution and O&M costs; * the fact that most schemes have only benefitted from the first cycle of the participatory and progressive approach. The rating for sustainability includes the following considerations: (i) the project approach, (ii) the -9-  commitment of stakeholders to investing in improving water management, (iii) the functioning of WUGs and (iv) the environmental sustainability. (i) The participatory and progressive approach of the project is an excellent example of how irrigation infrastructure delivery can be used as vehicle and incentive for WUG mobilization. It allows for a step-by-step process, where project and WUG are in cooperation. It also allows for periodic progress monitoring and evaluation by both parties, and is sufficiently flexible in case either one of them wants to amend, accelerate, slow down or abandon the process. PPI-2 experimented with the question 'what is the minimum investment in irrigation infrastructure sufficient to mobilize stakeholders to assume responsibility for O&M?' Other projects have often answered this question by assuming that only complete rehabilitation of irrigation schemes and investments in the order of 5,000 USD per hectare were necessary to trigger farmer mobilization for O&M. However, the PPI-2 has demonstrated that similar results can be obtained with per hectare investments that are much lower. This is an important lesson for future irrigation investment projects. Modest investments can create momentum for management transfer, on the condition that they are accompanied by proper mobilization and training of farmers. This is evidenced by the fact that at least some WUGs have achieved impressive results in terms of O&M. This suggests that, in principle, the approach and the low overall investment levels can lead to sustainable results. it should be pointed out, however, that the approach is likely to work better in small- to medium-sized schemes, and probably less so in large-scale schemes. The scope of work should be visible to have an impact on the water management of all members of the WUG so as to be an incentive for WUG mobilization. It is questionable wether this is the case in large-scale schemes, in particular in view of the very modest per hectare investments in these large-scale schemes. When the project decided to include large-scale schemes, it should also have amended the investment approach towards WVUG mobilization. (ii) An indicator for sustainability is the level of participation in investment costs. Up-front contributions help convey the message that farmers own the scheme and are responsible for O&M, and screen beneficiaries for genuine interest in assuming responsibility for O&M. On average, between 6% and 12% of the rehabilitation works were financed through farmers' contributions. The level of these contributions is nominal as comparad to similar projects elsewhere (e.g., in Mali, beneficiaries pay 25% upfront), will only contribute marginally to a sense of ownership and will only screen beneficiary commitment to a limited extent. Better rEsults in terms of sustainability would have been achieved with higher up-front farmers' contribution to investment costs. (iii) The extent to which WUGs were able to organize and conduct O&M on the rehabilitated infrastructure is of key importance to achieving sustainable results. This sustainability is at risk if, during project implementation, payment of irrigation water fees is not firmly and lastingly established at the 1 00% level. It is difficult to see how recovery of O&M fees would regain a sustainable level afterproject completion, if recovery is significantly below that level during implementation. Recovery of O&M fees scored generally low throughout project implementation, with only few positive exceptions for a small number of schemes. The PSR report of June 15, 2000 calculates actual cost recovery at 59% for recovery of maintenance costs, 65% for recovery of operational costs of the WUG and 75% for recovery of manual maintenance costs, the balance being deferred maintenance. The total financial contribution is on average on the order of USD 6 per hectare, excluding manually executed maintenance. The nominal level of the fees (as compared to similar projects elsewhere (e.g., Mali: USD 85 per hectare), and as compared to the benefits that beneficiaries derive from the investments) reinforces this argument. In addition, proper O&M was estimated to cost between USD 23 and USD 38 per hectare (although some questions have been raised whether this has been estimated correctly). With an average fee of USD 6 per hectare, it is evident that continued govemment support of maintenance will be unavoidable, even if farmers pay 100%. An additional issue that affected project sustainability is the fact that project assistance to schemes has in some cases taken a long time to reach completion. These long delays were justified by the fact that farmers were not yet capable of ensuring O&M sustainability. However, the result was often the opposite - 1 0 -  of what was intended to be achieved: farmers were spoon fed and became used to outside assistance. Paradoxically, thus, more financial means does not always mean higher sustainability, and it often even means the contrary. Among the weakest WUGs under PPI-2 are those that saw project assistance perpetuate. A study to guide the establishment of a permanent emergency rehabilitation fund planned in the SAR and intended to contribute to the development of a long-term approach to sustainable irrigation intervention programs in Madagascar, was not conducted. This study was to address issues related to the long-term reduction of aid dependence of the irrigated sector, of high relevance in view of Madagascar's permanent exposure to cyclones. The fact that this study was not done is a missed opportunity to contribute to the long-term sustainability of farmer-managed irrigation in Madagascar. (iv) The environmental sustainability of the project remains questionable. Few results and few positive impacts were achieved, and their replicability seems to be localized. The continuing environmental degradation that affects almost all of the schemes will most likely endanger their long-term physical integrity. In many cases, the scale of the degradation and sedimentation is beyond the financial means of the farmers. The outcomes therefore suggest that it is unlikely that the project's achievements, impressive as they may be in terms of area coverage, are sustainable. In fact, out of 19 schemes analyzed in the external evaluation, only five are rated as having a high level of functioning. The reasons for these results are related to (a) the design of the SAR and (b) implementation and supervision. (a) The SAR lacked a holistic approach towards sustainable agricultural development, including a meaningful integration of extension and other ways to enhance production, watershed management and irrigation. Extension activities were undertaken under a separate project (PNVA) resulting in an inefficient communication and collaboration with PPI. Similar considerations apply to the collaboration between PPI-2 and PAE. Most of the environmental protection works that needed to be undertaken in order to protect schemes from erosion and sedimentation exceeded by far the capacity of local populations, and in many cases consisted of not much more than a few square meters of vetiver in a watershed of thousands of hectares. The unfavorable comparison of labor investments and retums failed to trigger a self-propelled process of environmental protection. In addition, the project approach and funding mechanism used were found to be inconsistent, in the sense that the 'participatory and progressive' project approach depended to a large extent on beneficiary commitment. Project activities needed flexible and demand-driven funding operations to avoid the situation in which disbursement considerations determine project activities, and to ensure that the stakeholders are in the driver's seat. As one of the options, the establishment of decentralized, flexible and participatory rural investment funds (as discussed during the June 2000 watershed program pre-identification mission) should have been considered in the SAR. (b) With respect to implementation and supervision, the project was delayed by excessive time spent on the elaboration of an intervention strategy, most details of which were already spelled out in the SAR (with the exception of a time frame for the activities). Furthermore, the strategy that was established failed to specify a strategy for withdrawal from the villages, including a number of clear benchmarks for withdrawal as indicated by stakeholders. Finally, the project lacked a completion strategy on how to implement the three year's participatory and progressive intervention strategy towards the end of the project, when there is insufficient time for the full three year intervention cycle. It should however be pointed out that, although Madagascar has a long irrigation tradition, recovery of O&M charges is relatively new in the country and this may help to explain the relative low recovery levels. Other countries, where O&M cost recovery has been introduced, have typically shown a slow take-off and have achieved acceptable recovery rates only after many years and projects later. Sustainable O&M in Madagascar will require a number of years of consistently and persistently transmitting the same message by all parties involved. It will also require meticulous analysis of lessons learned, and adjusting the message and approach in the light of these lessons. Finally, it needs an - 1 1 -  holistic approach to sustainable rural development, including irrigation, extension, environment, market access, soil fertility and input supply. The fact that some villages have attained impressive levels of cost recovery based on a solid intemal WUG organization, demonstrates that it is possible to go beyond the level that was attained in the majority of the schemes. In summary, the rating of unlikely sustainability is based on the following considerations: * Notwithstanding the participatory and progressive approach adopted by the project, cost recovery for O&M remains weak, farmers only nominally contributed to investment costs, and project interference often lasted longer than necessary; * The project achieved limited or no environmental impact and did not reduce erosion and sedimentation levels; * No studies were conducted on issues related to broader and long-term sustainability of irrigation investment programs. 6.2 Transition arrangement to regular operations: In view of its important cost savings, the process of management transfer and O&M cost recovery is likely to be continued under the initiative of the Ministry, albeit, possibly, at a slower pace. Experience has shown that a combination of infrastructure investment and management transfer can be highly synergetic, and has the highest likelihood of lastingly leading to the intended results. Completion of works in some schemes was shelved at the time of the project's closure. In view of the above, it is expected that the GoM will continue completion of these works. 7. Bank and Borrower Performance Bank 7. 1 Lending; One of the problems with which participatory irrigation intervention projects are faced in general is the disconnect between participatory project implementation and disbursement. Often, credit disbursement is at odds with the longer time that a participatory approach requires to mature. A more adaptable disbursement schedule is needed. It can be questioned, in this particular project, if a more flexible disbursement schedule would not have been more appropriate to allow the establishment of a meaningful participatory process, adapted to the pace of the beneficiaries. The current disbursement schedule did not seem to leave the option open to accept disbursement delays. As discussed in the section on quality at entry, preparation and appraisal underestimated the risks associated with the possible disconnect identified above, and failed to identify an appropriate, more flexible disbursement procedure, adapted to a participatory project implementation. Because of this disconnect, the Bank's role in lending is rated as marginally satisfactory. 7.2 Supervision: During the first two years of implementation, disbursement was relatively slow for reasons of elaborating an intervention strategy. While the SAR anticipated that two years would be used to obtain beneficiary commitment, it is unclear why this needed to take two years, since most of the anticipated approach had been spelled out in detail in the SAR. In fact, only the time line for the various steps in the approach was missing in the SAR, and this could have been established within a relatively short period of time. With more pro-active supervision (e.g. through the organization of a workshop at project start to solve project intervention approach issues), the slow start could have been avoided. The decisions made during the 1996 supervision mission following this slow disbursement were decisive for subsequent project implementation. These decisions aimed at increasing the impact of the project, and included measures related to the implementation of works (decentralization, outsourcing to contractors, national coverage, small-, medium- and large-scale irrigation) as well as to the mobilization and organization of WUGs (outsourcing of activities to NGOs, capacity building of NGOs, decentralization). However, the results were not as anticipated. Whereas these decisions indeed sped up project implementation and disbursement, they also shifted attention to some extent away from - 12-  reinforcing capacity building of WUGs and focused on quantity more than on sustainability. Especially because of the inclusion of large-scale irrigation rehabilitation, environmental activities became insufficiently focused to have an impact, and therefore were ineffective. In addition, the progressive and participatory approach in large-scale schemes was no longer valid. As mentioned in section 3.5 and 7.1, one of the root problems of the project was the disconnect between disbursement constraints and the participatory project approach. Low disbursement may indeed have inspired the project supervision team to increase short-term project coverage and accelerate disbursement. However, the decision to accelerate disbursement also appears to have been inspired by a biased interpretation of the project's goals in terms of targets rather than in terms of their sustainability. In this particular case, there is no evidence of attempts to opt for sustainability rather than for increasing disbursement. The focus on disbursement has biased what would have otherwise been a satisfactory performance on supervision, especially in terms of flexible response by the supervision team to emerging issues. Additionally, it is not clear from the supervision reports why a study to determine the feasability of establishment of a permanent emergency rehabilitation fund, originally foreseen in the SAR, was not conducted. In the first supervision report, the terms of reference of this study were mentioned, but at project completion this study had still not been done. Supervision performance can therefore be rated as marginally satisfatory. 7.3 Overall Bank performance. Overall, the Bank performance can be rated as marginally satisfactory. This is based on the following considerations: * the failure to recognize risks associated with the inconsistency between disbursement constraints and the participatory project approach during preparation and appraisal; * contrary to the overall objective of the project, the decision to increase project impact seems to have been inspired by achieving targets, rather than by achieving them in a sustainable way; * the study into a permanent emergency rehabilitation fund was not conducted. Borrower 7.4 Preparation: No information is available on the involvement of the Borrower in project preparation. 7.5 Government implementation performance: Govemment implementation performance was satisfactory overall. Although counterpart funds were not always provided at the right time, the actual expenditures show that counterpart disbursements were higher than originally foreseen. 7.6 Implementing Agency: The argument developed under section 7.2, and hence a similar performance rating, equally applies to the Implementing Agency, who is responsible for project implementation. There is no evidence in the project documentation that the Implementing Agency resisted the idea of including large-scale schemes, or enlarging the coverage of the project, nor is there any indication suggesting to amend the project approach with respect to WUG mobilization or to environmental sustainability. The same applies to the study into a permanent emergency rehabilitation fund that was not conducted. During the implementation of the project, a number of procurement problems surfaced and appropriate action was taken to correct these problems. 7.7 Overall Borrower performance: Overall, the Borrower performance can be rated as marginally satisfactory. This is based on the following considerations: * Borrower did not raise any objections against amending the coverage of the project; * On the positive side, at project completion, government contribution was higher than anticipated - 13-  originally. 8. Lessons Learned The lessons leamed through the implementation of this project in Madagascar built on the lessons leamed from the PPI-1. They are likely to have a strong relevance to future irrigation programs. * The project has experimented with the question of a minimum investment volume needed to create the required critical mass for the mobilization of farmers. Whereas other projects have often assumed that only complete rehabilitation of irrigation schemes and investments of at least 5,000 USD per hectare were necessary to trigger farmer mobilization for O&M, the PPI-2 has demonstrated that similar results can be obtained with investments in the order of 250-500 USD per hectare. This is an important lesson for future irrigation investment projects. Even modest investments can create momentum for management transfer, on the condition that they are accompanied by proper mobilization and training of farmers. Future projects can build on this experience in the sense that infrastructure development and institutional capacity building can be designed to move forward together, step by step, adapted to the desired rhythm of the respective farmers. * The project approach used should be consistent with the selected disbursement approach. This issue should have been addressed as early as the project preparation phase. If a participatory project approach is to be taken seriously, then the Government and the Bank should both be willing to accept that more flexibility in the duration of project implementation is necessary. In addition, projects, and the participatory process on which they depend, need to be given a longer time to mature. For reasons of slow disbursement, they should not be allowed to shift focus towards achieving targets, and away from achieving them in a sustainable way. * A time limit should be set to the duration of the outside intervention. The participatory approach should be well-structured with an agreed agenda and objectives among all stakeholders. In addition, for the approach to produce sustainable results, it is imperative that the WUG is in the drivers seat and indicates the moment when farmers are ready to move ahead to the next stage, the need for further training and the technical assistance needed. This was not the case under the current project, and may partly account for the unsustainability of the results. * The environmental component of the project has demonstrated that irrigation projects should not be developed as stand-alone projects. Rather, irrigation projects should be part of a multi-sectoral approach, integrated into watersheds management programs and including land tenure, credit, transport, access to marketing, input supply and extension. Sufficient budgetary and human resources allocations need to be set aside to enable the project to adopt a meaningful and holistic watershed approach. * For investments in irrigation to be profitable, intensification should be adopted rigorously by farmers, including use of high yielding varieties, double cropping and use of fertilizers. This should be included in the selection criteria for receiving continued assistance from the project. Extension services and soil fertility, now left for implementation to the PNVA, should become more holistically integrated with irrigation development. In addition, other components need to receive attention, e.g., market access. * In the current project, the executing agency has spent much time, sometimes excessively so, on the development of an intervention approach. Technical assistance should be given to limit this time, in particular since participatory approaches are in a permanent state of being further developed and amended on the basis of new experiences. It does not seem to be fair to rely on the executing agency to take the lead in applying the latest developments. * Programming and prioritizing rehabilitation works have an impact on the returns on investments. The extemal evaluation revealed that the relationship between investments and returns is not linear. For example, some small works can solve a problem crucial to overall scheme functioning and thus decrease the time needed to bring total investments to maturity. * O&M cost recovery was frustrated by a lack of legal incentives for defaulters, resulting in rapid erosion of recovery rates. The need to adopt appropriate legal measures is therefore of paramount importance for the establishment and proper functioning of WUGs. * Procurement of small works should be decentralized as far as possible, depending on the size and nature of the contract. Most local contractors should, in principle, be capable of executing simple - 14-  civil works and should be given an opportunity to participate in bidding. Bidding and bid evaluation should, as far as possible, be organized at the provincial or lower level. * Donor co-ordination between all parties involved (EDF, AFD) is a prerogative of the Govemment and a necessary condition to successful implementation of the project and accomplishment of its objectives. 9. Partner Comments (a) Borrower/implementing agency: The document prepared by the Govemment is attached to this report (Annex 8). The following is an excerpt from this report. The interest of the GoM in irrigation does not end with the completion of PPI-2. Sectoral objectives and intervention strategies remain the same, with modifications inspired by the lessons learned from PPI-2 and the current decentralization process. To achieve its objectives in this important sector, the GoM needs continued donor support. During the implementation of PPI-2, some results were achieved, but they had not yet come to full fruition at project completion: * WUGs were established, but their operations are not yet sustainable; * In a number of cases, technical feasibility studies for scheme rehabilitation were conducted, but actual rehabilitation works have not been carried out; * A number of schemes have only benefited from emergency repairs, and have not been rehabilitated in a more comprehensive way; * Management of a number of schemes has not yet been transferred to farmers; * In case where management has been transferred, some WUGs find it as yet difficult to be autonomous; * Some schemes where management was transferred have not yet benefited from land titling. It is important to consolidate the achievements of PPI-2 and to continue to work in the directions set out during implementation of the two projects so as to achieve the objectives in full. Without this, the irrigation sector policy started in 1985 would fail and be in contradiction with activities undertaken so far. This would result in a confidence crisis between beneficiaries and the government impacting on future operations in the rural sector. In the meantime, the GoM will continue to support WUGs through relocation of staff, undertake operational costs, and protect strategic hydraulic structures. The following decisions, and improvements of the institutional and environmental context, have already been made by the GoM: * VAT has been cancelled for products used exclusively in agriculture; * Studies have translated into laws and regulations relevant to the irrigated sector in the context of RSDP; * Deliberations on the institutional context of post-emergency interventions have been conducted. (b) Cofinanciers: There were no co-financiers for this project. (c) Other partners (NGOs/private sector): No comments from other partners were received. 10. Additional Information - 15 -  Annex 1. Key Performance IndicatorslLog Frame Matrix Outcome I Impact Indicators: Total number of ha of inigation area 25,295ha 25,295ha rehabilitated: volumes of incremental crop production Rice: 14,800 Rice: 41,066 obtained (tons): Potatoes: 6,720 Fish: 116 no data provided by proect Triticale: 1025 increase in farmer income: US$195-240/ha US$203/ha Number of beneficiary water users reached: 148 148 number of water user groups established: 121 121 number of schemes formally transferred to water user groups: 28 14 Inigation area fully managed by water user 148 148 groups: contribution of water users to cost of 59-75% 59-75% operation and maintenance: cropping intensity 1.3 1.3 Output Indicators: End of project - 16-  Annex 2. Project Costs and Financing Project Cost by Component (in US$ million equivalent) Aaisal Actalatest Prcta Of Estniata Esthnate Appraisal Proec Cost By Componnt UaS$ million US$ million Cyclone rehabilitation and corrective rehabilitation work 4.10 4.50 109 Environmental assessment and protection 1.50 0.44 30 Rehabilitation of additional schemes 8.20 6.02 73 Support for services of Ministry of Agriculture 5.20 6.20 119 Training 0.70 0.00 57 Monitoring and Evaluation 0.20 2.31 140 Operations and maintenance by WUAs 1.30 0.64 47 Insecticides 1.91 Total Baseline Cost 21.20 22.02 Physical Contingencies 2.10 Price Contingencies 2.40 Total Project Costs 25.70 22.02 Total Financing Required 25.70 22.02 Project Costs by Procurement Arrangements (Aepraisal Estimate) (US$ million equivalent) Procurement Melf",d ExporWitwo Caftgory im NCO N.B.F. TotaI CGst 1. Works 13.20 1.20 0.00 0.00 14.40 (11.80) (1.00) (0.00) (0.00) (12.80) 2. Goods 0.97 0.46 0.30 0.00 1.73 (0.66) (0.33) (0.20) (0.00) (1.19) 3. Services 0.00 0.00 3.36 0.00 3.36 Consultants (0.00) (0.00) (3.18) (0.00) (3.18) 4. Miscellaneous 0.00 0.00 1.04 0.00 1.04 Training & Workshops (0.00) (0.00) (1.04) (0.00) (1.04) 5. Miscellaneous: 0.00 0.15 2.42 0.00 2.57 Allowances & Rec. costs (0.00) (0.13) (2.11) (0.00) (2.24) 6. Miscellaneous 0.00 0.00 0.00 0.00 0.00 (0.00) (0.00) (0.00) (0.00) (0.00) Total 14.17 1.81 7.12 0.00 23.10 (12.46) (1.46) (6.53) (0.00) (20.45) Project Costs br Procureme rt Arngements (Actual/Latest Estimate) (US$ million equivalent) Proeurenet Metowd EVxINIUre CalgWorY 1ic NCB Other2 N.B.F. TotalCost 1. Works 0.00 8.01 0.53 0.00 8.54 ________________________ (0.00) (0.00) (0.00) (0.00) (0.00) - 17 -  2. Goods 0.70 0.56 0.03 0.00 1.29 (0.00) (0.00) (0.00) (0.00) (0.00) 3. Services 1.21 1.20 0.74 0.00 3.15 Consultants (0.00) (0.00) (0.00) (0.00) (0.00) 4. Miscellaneous 0.00 0.00 0.95 0.00 0.95 Training & Workshops (0.00) (0.00) (0.00) (0.00) (0.00) 5. Miscellaneous: 0.00 0.00 4.52 0.00 4.52 Allowances & Rec. costs (0.00) (0.00) (0.00) (0.00) (0.00) 6. Miscellaneous 0.62 1.25 0.11 0.00 1.98 (0.00) (0.00) (0.00