Madagascar - Antananarivo Water and Sanitation Project

Document of The World Bank FOR OFFICIAL USE ONLY Report No. 8879 PROJECT PERFORMANCE AUDIT REPORT MADAGASCAR ANTANANARIVO WATER AND SANITATION PROJECT (CREDIT 1002-MAG) JUNE 29, 1990 Operations Evaluation Department This document has a restricted distribution and may be used by reciplents only In the performance of their official duties. Its contents may not otherwise be disclosed without World Bank authoriatIon. ACRONYMS AND ABBREVIATIONS DUO - Urbanism and Housing Directorate, Ministry of Public Works SEN - Eau et Electricit6 de Madagascar Government - Government of the Democratic Republic of Madagascar JIRAMA - Malagasy Electricity and Water Company MOF - Ministry of Finance MOPW - Ministry of Public Works VAT - Value Added Tax COUNTRY EXCHANGE RATES Malagasy Franc (FMG) 1US$ Appraisal Estimate (7/78) US$1.00 = 210.00 Actualss Average 1980 - 211.28 Average 1981 - 271.73 Average 1982 - 349.74 Average 1983 - 430.45 Average 1984 = 576.64 Average 1985 - 662.48 Average 1986 - 676.34 FOR OFFICIAL USE ONLY THE WORLD ANK Washagton. D . 20433 U.SA Ote of DucOr-Genstal opesmao Iwtuats June 29, 1990 MEMORANDUM TO THE EXECUTIVE DIRECTORS AND THE PRESIDENT SUBJECT: Project Performance Audit Report on Madagascar Antananarivo Water and Sanitation Project (Credit 1002-MAG) Attached, for information, is a copy of a report entitled 'Project Performance Audit Report on Madagascar Antananarivo Water and .Sanitation Project (Credit 1002-MAG)w prepared by the Operations Evaluation Department. Attachment This documaet has a metuicted distribution and may be used by eciplets only in the performance of theit official duties. Its contents may no otherwim he dilod withont World Rank anthmifnm POR OFFICIAL USE ONLY PROJECT PERFORMANCE AUDIT REPORT MADAGASCAR ANTANANARIVO WATER AND SANITATION PROJECT (CREDIT 1002-MAG) TABLE OF CONTENTS Pase No. Preface ........................................................ Basic Data Sheets ........ ......................... .......... Evaluation Summary ............................................. v Background ..................................................... 1 Project Objectives ............................................. 2 Implementation ................................................. 4 Project Costs and Credit Disbursement .......................... 5 Maturity, Grace Period and Exchange Risk ....................... 7 External Audit ................................................. 8 Institutional Development ...................................... 9 Financial Performance ................................... 10 The Threat of Suspension of Disbursements ...................... 13 Procurement ..............* * ***................................. 14 Conclusions and Lessons ........................................ 15 This document has a restricted distribution and may be used by recipients only in the performance I t sh M- a lt An nha me * ....*: 1#e asha.. --." A2.' haI.A .. U..*Br-.IA 0-16mek .-s - i - PROJECT PERFORMANCE AUDIT REPORT MADAGASCAR ANTANANARIVO WATER AND SANITAT1ON PROJECT (CREDIT 1002-MAG) PREFACE 1. This is a Project Performance Audit Report (PPAR) on the Antananarivo Water and Sanitation Project, involving an IDA credit in the amount of US$20.5 million to the Democratic Republic of Madagascar. The credit was approved on April 1, 1980, and became effective on December 17, 1980. US$7.0 million of the credit was cancelled. The Closing Date of June 30, 1985 was extended to June 30, 1986. Final disbursement was made on November 4, 1986. 2. The PPAR is based on the Project Completion Report (PCR) of the project prepared jointly bl the Africa Regional Office and the Borrower (Part II) and issued in 1989, the Staff Appraisal, the Credit and Project Agreements, supervision reports, correspondence between the Bank and the Borrower, and internal Bank memoranda. An OED mission visited Madagascar in November 1989, and discussed the effectiveness of the Bank's assistance with the Borrower. Their kind cooperation and valuable assistance in the preparation of this report is gratefully acknowledged. 3. The PCR provides a comprehensive account of the project experience. The PPAR elaborates on selected aspects of project implementation, in particular the institutional and financial features, and the performance of the Bank. 4. The draft PPAR was sent to the Borrower for coaments but none were received. 11 Project Completion Report, Madagascar - Water and Sanitation Project (Credit 1002-MAG), Report No. 8091, dated September 28, 1989. - iii - PROJECT PERFORMANCE AUDIT REPORT MADAGASCAR ANTANANARIVO WATER AND SANITATION PROJECT (CREDIT 1002-MAG) BASIC DATA SHEET ------------------------------------------------------------------- KEY PROJECT DATA Appraisal Actual Actual as I Item Expectations Estimates of Apyr. Est. Total Project Cost (US$ million) 33.7 15.9 -53 Credit Amount (0) 20.5 13.5 -34 Date Physical Components Completed 12131184 06130/86 Economic Rate of Return (2) for water 7.4 1.2 -6.2 points Institutional Performance ----------------------- -------------------------------------------- CUMULATIVE ESTIMATED AND ACTUAL DISBURSEMENTS (US$ million) FY ended June 30, 1981 1982 1983 1984 1985 1986 1987 (1) Appraisal Estimate 4.5 9.5 15.5 20.0 20.5 (2) Actual 0 0.8 2.4 4.6 5.6 11.98 13.5 (3) (2) as I of (1) 0 8 15 23 27 n.a. n.a. Date of Final Disbursements 6/30/86 ------------------------------------------------------------------- PROJECT DATES Original Plan Revised Actual First Mention in Files 9/14176 Negotiations 12/10/79 Board Approval 04/01/80 Loan Agreement Date 05/08/80 Effectiveness Date 10/08/80 11/30/80;1/30/81 12/17/80 Amendment Date 05121/85 Closing Date 06/30/85 06/30/86 06/30/86 - iv - STAFF INPUT (Staff-Weeks) Bank FY 1978 1979 1980 1981 1982 1983 1984 1985 1986 1989 Preappraisal 1.0 17.7 11.9 2.8 Appraisal 42.1 Negotiations 1.1 12.9 Supervision 3.0 4.7 5.2 11.8 12.0 7.9 5.8 Other Totals 1.0 59.8 13.0 18.7 4.7 5.2 11.8 12.0 7.9 5.8 ----------- -------------------------------------------------------- MISSION DATA No. of No. of Date of Month/Year Days Persons Manweeks Report Identification 4/77 12 1 1.7 5/25177 Preparation 1 12/77 9 3 3.9 12130177 Preappraisal 4/78 5 1 0.7 5116178 Appraisal 7/78 23 2 6.6 8/10/78 Post-Appraisal 2/79 8 1 1.1 3/26/79 Supervision I 4/80 5 1 0.7 5128/80 Supervision II 5/81 7 2 2.0 8/18/81 Supervision III 2/82 14 1 2.0 4113/82 Supervision IV 11/82 10 2 1.4 1/28/83 Supervision V 4/83 8 2 1.1 6127183 Supervision VI 2/84 11 1 1.6 3/30/84 Supervision VII 12184 6 2 0.9 2/11/84 Supervision VIII 5/85 8 2 1.1 6/20/85 Supervision IX 1/86 7 1 1.0 2/26/86 Supervision X 11/86 7 1 1.0 2/11/87 Completion 2/89 5 1 0.7 5/15/89 Totals (averages) 9.1 1.5 1.7 ------------------------------------------------------------------- OTHER PROJECT DATA Borrowert Democratic Republic of Madagascar Executing Agencies: JIRAMA, FIVONDRONANA, Ministry of Public Works Fiscal Year of Borrower: January 1 - December 31 Follow-up Projects: Urban Project, Credit 1497-MAG Tana Plain (under preparation) . V - PROJECT PERFORMANCE AUDIT REPORT MADAGASCAR ANTANANARIVO WATER AND SANITATION PROJECT (CREDIT 1002-MAG) EVALUATION SUMMARY Introduction 1. The Antananarivo Water and Sanitation Project was IDA's first water and sanitation operation in the water and sanitation sector. Madagascar is a poor country where urban migration is occurring at a faster pace than popu- lation growth because rural inhabitants keep looking for better employment prospects. This unabated demographic trend put an increased burden on limi- ted and aging urban infrastructures. Following up on a UNDP/WHO sector study and master plan, the Government of Madagascar decided in the mid-1970s to invest in crucial water and sewerage facilities in order to provide better standards of living and of health in the capital. The Bank Group supported this strategy, which met its own poverty alleviation emphasis. Through its IDA window, a credit was granted to finance three complementary components of infrastructure: water, sewerage, and solid waste. An added justification for choosing Antananarivo was the need to address through a drainage compo- nent the recurrent flooding problem, which had become a political as well as an economic priority. The Project and Its Objectives 2. The objectives of the project were a priori conventional and were in line with the Government's strategy of improving public services in the capital and largest city. Four objectives were stated: providing infra- structure facilities, strengthening implementing entities, cost recovery, and preparing long-term expansion plans (PCR, para. 2.08 and PPAR, para. 1.05). The project approved by the Board comprised the construction of primary and secondary infrastructures for the provision of water, the collection of used water, the drainage of flash floods, and the transport and treatment of solid waste. The project was implemented by the JIRAMA, the national water and electricity company, and by the FIVONDRONANA, the municipality of Antananarivo. The total project cost was initially estimated at US$33.7 million, of which 61% (US$20.5 million) was to be financed by the IDA credit. 3. The project represented the implementation of the second phase of the Antananarivo master plan (whose first stage consisted of limited improve- ments executed with local resources) with all design and supervision to be carried out by local agencies, assisted by foreign consultants in special areas. The project consisted of conventional water supply investments (302 of the estimated cost), drainage investments (40%), sewerage works (13%), solid wastes equipments (112), training benefiting JIRAMA (52) and the Fivondronana (1%). Taxes (customs duties and a 102 VAT) were 9.5% of project - vi - cost. Foreign costs were 412 which implies that IDA was to finance 33Z of local costs. Implementation Experience 4. This evaluation focuses on two operational issues, financial viability and institutional strengthening, which are key to th- success of any development project. The rationale for this highlighting is threefold. First, the extent of organizational, administrative and personnel changes to be made before the project could be started or be sustained made them pre- requisites for its success. Second, this first project had no follow-up until the Tana Plain project (currently under final processing) despite the dire need for basic infrastructure (i.e. debt servicing was absorbing too much cash flows to allow reasonable self-financing of follow-up projects). This lag can be attributed in large part to the slow institutional build-up in the sector. Third, the initial project could not be duplicated because its financing plan had no real sustainability. Among a number of other issues, adequately covered in the PCR, are the uneven performance of consul- tants which affected project implementation, the faulty procurement due to limited international contracting experience, and the delayed tariff increases which affected the project's covenanted returns and cost recovery. 5. Although the diagnosis was clear before appraisal about the weak- nesses of the implementing agencies, its full implications were not carried over in the risk assessment part prior to committing funds to the project. Given that the two entities in charge of undertaking the sewerage component had either no or limited previous investment experience with both IDA and ICB procurement, implementation problems should have been anticipated. What should have been addressed as a problem due to lack of experience, which has no sensitivity connotation, was instead handled as a problem of turf among institutions, by nature delicate to resolve. Neither the Ministry of Public Works (MOPW) nor the Fivondronana has ever been involved in major sewerage investment, whether feasibility studies, procurement, or supervising con- struction. Results and Sustainability 6. In large part because of the above impediments, the project was delayed, had one component cancelled (sewerage) and another in need of reha- bilitation before the credit was closed (solid waste) (PPAR, paras. 1.11 and 1.37). Regarding the components which were fully implemented, the Audit accepts the PCR claim that the sustainability of their economic benefits is assured based on the recalculated ERR (PPAR, para. 1.14). Yet the physical sustainability of the project is not completely assured. A case in point is the largest component, the extension of the water services, which is now overextended because the demand is forcing overcapacity utilization which is bound to take a toll; comments received from the plant supervisor were specific about the repercussions on maintenance of 24-hour a day pumping; no new investment has taken place since that financed under Credit 1002-MAG. The solid waste component had a short life because the entire garbage truck fleet had to be replaced due to lack of spare parts. The sewerage component was not implemented (PPAR, para. 1.15). At the institutional level, - vii - furthermore, the viability of the agency in charge of the only surviving component, water treatment and distribution, is not assured given the recurrent nature of its financial predicament. Significantly, JIRAMA has not repaid any of its share of Credit 1002-MAG to the Government, whose debt servicing is scheduled to start in April 1990 (PPAR, para. 1.32). Lessons 7. It is generally the case with infrastructure projects that, once the procurement hurdle has been passed, the physical components are satisfac- torily implemented, albeit often with time and cost overruns. It is no exception with this project whose only tangible result is the extension of JIRAMA's production and distribution capacities, which delivers essential water to a growing population. The other components had disappointing out- comes largely because the implementing agencies had shortcomings. A lesson drawn from this project's experience deals with the need to hedge against institutional risks. If the institutional risk is substantial, priority should be given to offset likely problems, among others, of (i) coordination between implementing agencies or of (ii) lack of experience by ensuring that the physical content of the project is as detailed-prepared as possible before credit approval, using appropriate technical assistance. 8. As originally formulated, the project's physical objectives (PPAR, para. 1.07) were manageable provided that some measures of institutional strengthening would be implemented to support the ambitious sector objectives (PPAR, para. 1.5). The strategy to implement them was, however, deficient at the institutional level because bot- agencies had major handicaps. The water company had, and still has, str; - ural financial problems, and the Fivondronana was, and still is, administratively and financially weak. The humbling lesson from this project is that stating objectives and targets for a development project is not sufficient if the means selected to achieve them are inadequate, as was the case here. This made the whole strategy flawed to start with. The correct procedure should be to formulate a balanced strategy first, then to tailor it - by cutting components if necessary - according to the absorptive and delivery capacities of the implementing agencies upon which the project outcomes lay. 9. Multi-components projects are inherently difficult to implement satisfactorily because they cannot be deemed successful if only one component works. Overloading a single and first operation with too m,ry components in an untested environment was a major bet on the absorptive and implementation capacities of largely bureaucratic organizations. It was lost early on dur- ing implementation as the cancellation of one of the two largest components and the financial restructuring of one of the two entities, both in 1985, eventually illustrated. The obvious lesson from this project is that IDA should refrain from the temptation to try and solve all problems at once through complex projects in economies whose structures have not reached the needed maturity to handle parallel components. 10. The project experience demonstrates that success owes much to the strong technical competence developed and maintained over the years. It also underscores the importance of financial autonomy and discipline. Efforts at - viii - financial restructuring initiated in 1985 were short-lived and did not have a lasting effect (PPAR, para. 1.30). 11. The discrepancy between IDA credit terms (e.g., 0.75Z and 50 year maturity) and that for subloans is irreconcilable and inconsistent with the effective cash flow generation ability. If the ERR of the water component was estimated at 7.2Z and the FRR at only 5.1Z, it is difficult to envisage how JIRAMA would have been able to remunerate a loan at 8% plus a substantial exchange risk. A fortiori, when the ex post ERR was reduced to 1.22. A lesson to be drawn from this situation suggests the need to test whether subborrowere can afford IDA's on-lending terms which stipulate grace period as short as three years and maturity as low as 20 years. Furthermore, to peg the interest rate charged to IDA beneficiaries to the IBRD loan rate plus one or 1.5 points should not be automatic, but must be justified in every case on the basis of risk-adjusted cash flows projections. This project demonstrated the inability of an-otherwise technically competent-agency to repay 4.52 p.a. of its principal debt, let alone an 8.52 annual interest. in this respect, financing local costs with IDA funds is expensive because of the underlying exchange risk which is all the more potent that the risk is stretched over a long maturity. 12. Conditionality had mixed results in its application. There was no effectiveness conditionality despite the fact that some were clearly outlined at pre-appraisal. Cancellation with little warning was used to erase a com- ponent which had still some imple-entation potential and has had no replace- ment from any donor in ten years. Indeed, the fact that waste water disposal investments were not undertaken in parallel with the water extension means that Antananarivo is still declared a zone with malaria and other waterborne parasitic diseases. To implement a financial restructuring, which should have actually been done without pressure because it was to lift JIRAMA Zrom insolvency, suspension of disbursements was threatened for an excessively long period. The mala lesson is that suspension threat is not an effective institution-building tool if not executed for a long period (over 18 months in the case of Credit 1002-MAG). Quicker results were obtained by threatening to postpone processing follow-up projects or by making cross- conditionalities in other loans. 13. Overall, the external audit of the implementing agencies' accounts was a tool underused by IDA staff given that structural financial problems were evident since appraisal. The audit reports offered an yearly opportun- ity to disclose the true financial situation of JIRAMA in order to accelerate decision about its needed financial restructuring. Too much time was spent arguing about the auditor's repeated qualification of JIRAMA's accounts and not enough to the fact that its cash flows situation was overstated. To this day, audited accounts are presented according to the local Accounting Plan; the forthcoming changes should bring in better light the financial flows. Given the magnitude of the arrears, however, nothing prevented IDA from requesting the external audit to present the financial situation more on a cash basis than it had been done. 14. JIRAMA's financial predicaments are the accumulation of receivables on the assets side of its balance sheet and of an exchange risk on the - ix - liabilities side. Payment arrears from clients can be reduced if not elimi- nated. Cutting supply is a lever which should be used to educate consumers in responsible behavior; such a penalty has been incorporated in the "program contractsO signed with municipalities in 1989 (PPAR, para. 1.29). The exchange risk is outside JIRAMA's control, except for the temptation to borrow abroad for local cost investment which should be refrained because it is not economical. Furthermore, being a domestic producer, JIRAMA has no means to hedge at a reasonable cost against foreign currency risk. By raising its debt-equity ratio into uncreditworthiness range, the exchange risk impedes the future borrowing ability, he-ice puts the financing of necessary investment in jeopardy. As long as the exchange risk will be borne effectively by JIRAMA, despite legal indications that it is covered by the Government, JIRAMA's finances will become weaker every year since the exchange risk increases faster than sales revenues (PPAR, para. 1.33). The lesson to be drawn from this case is that a utility which has neither the means nor the experience to hedge against the exchange risk should not bear it. However, a fixed portion of this additional cost of borrowing in foreign currencies should be passed on as an additional interest rate to the final borrover. PROJECT PERFORMANCE AUDIT REPORT MADACASCAR ANTANANARIVO WATER AND SANITATION PROJECT (CREDIT 1002-MAG) Ba. .ground 1.01 Antananarivo, the capital of Madagascar, had a population of 635,000 in 1975 three years before appraisal. Total population of Madagascar (8 million in 1977) was then growing at a rapid pace of 2.5% p.a. expected to accelerate to 2.92 until 2000, but even more rapidly in urban centers (around 6%). In 1975, of the 1.5 million of urban population 712 were sup- plied with treated water1!, but only 332 of which through house connections, and only 3% of the same population was connected to public sewer systems. 1.02 Water supplied to Antananarivo is drawn from a small lake south of it and fed by pumping upstream from the city, thus avoiding urban pollution. To meet the demand in the 1980s and 1990s, the production plant on the lake needed expansion in treatment and pumping capacities; storage in reservoirs scattered in the city hills needed to be increased; high pressure booster stations were needed in high and remote locations; and old portions of cast- iron pipes needed to be replaced. The quality of the water service was good due to the technical know-how inherited from Electricitd et Eau de Madagascar (EEM); JIRAMA (Jiro sy Rano Malagasy, which means Electricity and Water of Madagascar) was established in 1976 and took over the assets of EEM, the former private utility set up by French Lyonnaise des Eaux. In contrast, the public sewer/drainage system was old and covered only 10% of the population. The hills are served through pipes while the plain is by a rural-type system of canals, often with low slope, badly maintained and clogged by illegal dumping of garbage. Because the hill pipes are able to discharge much greater quantities than the plain system can carry, the capital's low areas are flooded during the rainy season. These annual floods, which affect the low-income urban population most, added a political dimension to the urgency of correcting the situation. Collection of solid wastes covered 90% of the needs of the city, but the bottlenecks were the inadequate capacity of the containers throughout the city and the insufficient transportation fleet to the land-fill 9 km from the city. 1.03 Several institutions played a role in the sector: the Ministry of Public Works (MOPW) which has overall responsibility for the water and sani- tation sector and specific responsibility for investing in public sewers; JIRAMA, the water and power company for the whole country, ad the Fivondronana, the municipal government of Antananarivo, which operates the public sewer system. Most had overlapping areas of responsibility which complicated the decision-making process. To further confuse matters, the Antananarivo municipality had a tradition of independence by virtue of being in the political opposition. This de facto autonomy of decision proved to 1/ With quality control done by the Pasteur Institute. - 2 - be a factor in the failure of the sewerage component in the IDA project. While EEH had a contract with the Municipality running until 1979, the latter was not compensated when JIRAMA took over in 1976 and this led the Municipal- ity to refuse paying water charges. The last major investments in the water sector were undertaken between 1965 and 1969 by EEM, the private utility preceding JIRAMA, and during the 1970s most of the investment effort was directed toward the power sector, including the Andelakela dam financed with two IDA credits totaling US$43 million.2 1.04 To face up to the challenge of the rapid urbanization of the capital from the pre-Independence low level, the Government sought financial and technical assistance from bilateral and multilateral agencies to remedy the sectoral deficiencies. IDA's response was the financing of the Antananarivo Water and Sanitation Project (Credit 1002-MAG) whose outcome is evaluated in this report. This was the first project undertaken by the Bank Group in the water and sanitation sector and it followed US$225.13 million of credit granted between 1966 and 1979 mainly to infrastructure such as roads (37%); railways (8%) and ports (5%); agriculture (22%); and power (19%). Project Obiectives 1.05 The project's objectives were multiple, but, given the pressing needs of the capital, could be considered timely and appropriate. They were in support of the Government's strategy of improving water supply and sanita- tion services country-wide, starting with the primary cities where the lack of water and sanitation infrastructures had the most acute effects. Four objectives were stated: one physical and three institutional. They were: (1) to improve and develop the existing water supply, sewerage, drainage and solid waste facilities in the Antananarivo area; (2) to strengthen the insti- tutions, especially in the sanitation subsector, in order to attain proper standards of operation and maintenance; (3) to assist the Government in developing appropriate cost recovery policies, and (4) to assist the Government in preparing a more precise long-term development program for urban water supply from which a sector lending strategy could be devised (SAR, para. 3.02). 1.06 The project had four physical components and two training com- ponents. The former comprised a water supply, a storm drainage, a sewerage, and a solid waste component. Training was to be provided to staff of the two implementing agencies. The water supply component included construction of minor additions to production works, about 17 km of transmission lines, 1.8 km of pipelines to supply suburban communities, extension and reinforcement of the water distribution system (reservoirs, pipes, booster station) and supply of heavy equipment for maintenance. Overall the component aimed at 21 Credit 817-0 for US$33 million granted in 1978 and Credit 817-1 for US$10 million granted in 1980. Although both have been closed, in 1983 and 1986, respectively, none has been subjected to a Project Completion Report. meeting the water demand of the Antananarivo Fivondronana and of all suburban communities up to 1986. Except for public standposts, all above investments were to be undertaken by JIRAMA. 1.07 The three other physical components were dealing with the Fivondronana. The drainage component included constructing collectors, river dikes, a pumping station, ten storage basins, a 3-km canal, dredging 10 km of existing canal and rerouting of another. The objective was to reduce the frequency of floods in the low areas of the town during the rainy season. The sewage component was modest in size and included constructing sewerage collectors and reticulation, a pumping station, pumping discharge mains and pilot stabilization ponds. The objective was to complete the facilities in the densely populated quarters of the city center and to reduce the pollution of the canal which crosses the heavily populated and poor areas of the town. The solid waste components consisted of the construction of 560 collective dumping containers, the supply of 19 collection trucks and of heavy equipment to maintain the existing dumping site, and the improvement of an experimental composting facility. The objective was to meet the needs of the community up to 1984. The project also included training for JIRAMA's water supply staff and the sanitation staff of the Fivondronana plus the financing of 30Z of the extension of the JIRAMA training centqr. 1.08 From 1971 to 1975, UNDP financed and WHO managed a master plan and feasibility studies for the water and sewerage sector in Antananarivo. In December 1976, the Government formally requested that the Bank include the project in its lending program. Identification took place in February 1977, and three preparation missions visited the capital between April 1977 and July 1978. Then the project cost was estimated US$29.0 nillion and IDA was to provide US$15 million. Further processing was delayed because of the per- ceived tough conditions (PCR, para. 2.05). Three conditions of negotiation were sets financing by the Government of JIRAMA's water deficit for 1978, reduction of Government arrears vis-&-vis JIRAMA, and the electricity tariffs to be raised so as to yield 8% on revalued net fixed assets in operation. Two conditions for Board presentation were stipulated: to increase water tariffs by 50%, and to appoint six key staff in the Sanitation Division of the Fivondronana and the financial manager for JIRAMA.3 1.09 Although none of the conditions were met, negotiations were scheduled for December 1979 with the claim that 'no major difficulties (were] anticipated' (Memo from loan officer, December 7, 1979). During negotia- tions, the credit amou-t was increased to US$20.5 million from US$17.5 million. Delays and new estimates of international and domestic inflation had increased project costs to US$33.7 million. Water tariffs increase was downgraded to an effectiveness condition. They were raised by 12% in January 1980 despite the understanding that it would be at least 7.5 percentage points above the revenues generated by the automatic adjustment formula (telex to Government, December 28, 1979). The increases of less than 25% (counting one made in October 1979) were found acceptable by the Bank despite 3/ Back-to-office memorandum dated July 6, 1979. - 4 - a negative rate of return expected for 1980 (Memo to Mr. Stern, February 12, 1980). Additional outcomes of the negotiations were (1) the agreed cessation by JIRAMA of any industrial activity (manufacturing of water meter and home appliances) after the end of 1980, and (2) the extension from 30 to 35 years of the maturity of funds to be onlent to the Fivondronana (PCR, para. 2.06). Board approval was granted on April 1, 1980 and, despite the delays, the closing date was not revised from initial appraisal which implied completion in less than four years (memo dated February 14, 1980). The credit became effective on December 17, 1980 although the only conditions were the signing of two subloan agreements. Implementation 1.10 Implementation of the project was highly contrasted, which should suggest useful lessons. It was trouble free for the water supply component, but was plagued with problems for the sanitation components. The implementa- tion schedule set in the appraisal report should have been considered highly optimistic given the lack of previous experience in either component. Seven months after effectiveness, the sanitation components were more than one year late (letter to MOPW, July 28, 1981). Even for the water d.partment of JIRAMA it was the first major investment and international contracting exper- ience. Although the Fivondronana had no experience ot all, it was let be in charge of the solid waste component. The drainage and sewerage components were entrusted to the MOPW. Coordination was deemed necessary by the Government and a staff of the General Directorate of Planning was appointed Project Director. This layer of administration proved disappointing and did not prevent key components from failing to be implemented (PCR, para. 3.01). 1.11 The longest delay occurred on the drainage and sewerage components because of the insistence on using unproven local consultants. It had been agreed that the final design and the preparation of bidding documents would be carried out by MOPW's own staff. Later on, MOPW requested IDA to finance a contract with a local consulting firm which was eventually dismissed for incompetence. To replace it, MOPW selected JIRAMA in January 1982; implemen- tation was then 2 1/2 years behind schedule. In turn, JIRAMA selected the foreign firm which had carried out the UNDP study in the early 1970s. Although the choice was logical, the firm was only a subcontractor with no direct access to MOPW which insisted on finalizing bid documents without prior experience in this sector. This led to a procurement problem (PPAR, para. 1.37) which resulted in the eventual cancellation of the drainage and sewerage components. In contrast, despite the initial delay due to effec- tiveness, the water component was completed within target, a remarkable achievement given that it was JIRAMA's first such project. 1.12 The review of the project-related correspondence and progress reports shows that institutional problems were apparent across the board from the first supervision mission. JIRAMA's financial situation was called seri- ous due to public sector arrears and the financial advisor was not empowered to correct the deteriorating situation. The six staff recruited by the Fivondronana as Board condition were not working for the project and 75 positions remained to be filled. Most of the covenants were not complied with: the water surtax to recover sewerage works was not computed between Fivondronana, the beneficiary, and JIRAMA, its collector; no sub-borrowers had submitted financial reporting; JIRAHA had not hired a much needed finan- cial director and not divested from industrial activities. The subsequent nine supervision missions would deal with the same problems making either little or no progress on these issues while handling a major financial crisis in JIREMA (PPAR, paras. 1.28-1.33). 1.13 In April 1984, some regions of Madagascar were severely damaged by a cyclone. In order to contract on short notice for repair works, US$2.9 million were reallocated from Credit 1002-MAG. Mention in the PCR of the effects of these investments is limited, in part because there was little functional relationship, aside from the use of this project as a funding source, between the Cyclone and Water and Sanitation projects (PPAR, para. 1.16). 1.14 Achievements were limited compared to the original objectives. The project was delayed, had one component cancelled (sewerage) and another in need of rehabilitation before the credit was closed (solid waste). The physical sustainability of the project is not likely to be achieved. The largest component, the extension of the water services, is now in jeopardy because the demand is forcing over-capacity utilization which is bound to take a toll; no new investment has taken place since that financed under Credit 1002-MAG. The solid waste component had a short life because most of the garbage truck fleet needed to be replaced due to lack of spare parts. The sewerage component was not implemented and thus its survival could not be tested (PPAR, para. 1.15). Regarding the components which were fully implemented, the Audit accepts the PCR claim that the sustainability of their economic benefits is assured albeit based on a recalculated ERR of 1.22 (water component). At the institutional level, however, the viability of the implementing agency for the longest-survived component, water treatment and distribution, is not assured given the recurrent nature of its financial predicaments. Significantly, JIRAMA has not repaid any of its subloan from Credit 1002-MAG to the Government, whose debt servicing is scheduled to start in April 1990 (PPAR, para. 1.32). This is partly due to the heavy exchange risk burden that JIRAMA must de facto assume (PPAR, para. 1.33). Project Costs and Credit Disbursement 1.15 The PCR compares the investment cost estimated at appraisal and the actual cost of the project (PCR, para. 3.11) and concludes correctly that comparison is difficult due to the changes affecting the drainage and sewer- age components. The civil works for water supply had a cost overrun in dollar terms of 582 which the PCR blames on the delays and local inflation (PCR, para. 3.11). The review of the report prepared by the borrower suggests a different explanation although cost tables in (1) the SAR, (2) the PCR, and (3) the borrower-prepared PCR cannot be readily reconciled. Local cost expenditures showed savings of 51, but foreign currency ones (all in French Francs) had a 262 overrun. There was a 132 saving on equipments planned at appraisal, but new equipment (telemonitoring of hill reservoirs and dredging of the lake feeding the plant) added US$580,000 to costs. How- - 6 - ever, the significant overrun was on civil work (5502 in FMG according to the borrower; 582 in US$ according to the PCR), partly due to additional work performed (intake and reservoirs). 1.16 The PCR does not identify which of the components got the bulk of the US$1.75 million spent on civil works for drainage, sewerage and solid wastes although urgent works were done in drainage and sewerage about which this Audit could not find any detail (PCR, para. 3.11). A fleet of garbage collection trucks was, however, purchased through ICB after a 2-year delay. The PCR does not comment on the works performed under Cyclone Rehabilitation amendment although they amounted to US$2.67 million or 17% of final project cost (PCR, para. 3.11). Except for two items clearly falling under cyclone rehabilitation (such as the rehabilitation of power networks in six towns, the repair of various buildings), a substantial share of the credit amendment contributed indirectly to increasing disbursements under Credit 1002-MAG. Table P.14 of the PCR shows that 27% went for a consultant study justified in part to prepare the first urban project and the rest benefited JIRAMA with repair of fixed assets and also the replenishment of its vehicle spare parts stocks. 1.17 Two months after amending the Credit Agreement, US$7 million earmarked for the drainage and sewerage components were cancelled (June 30, 1985) and the closing date was postponed from June 30, 1985 to June 30, 1986 to process disbursement requests on the water supply component. Excluding the Cyclone reallocation, the original credit amount was used at only 53% which was a disappointing result for a first project in the sector. When the cancellation was decided, 27? of the credit had been disbursed. The cancel- lation seemed to have surprised the authorities; the project unit within the Ministry of Planning claimed not having been notified. The Audit found, however, the correspondence clear as to who was notified and when.4 Regard- less of the actual sequence of events, the fact is that the responsible ministries and implementing agencies kept on submitting disbursement requests even though the credit balance had been fully disbursed. The lack of commun- ication resulted in the credit being overcommitted and one JIRAMA contractor who could not be paid under Credit 1002-MAG and had to wait three years to be compensated out of the following IDA credit (telex to Fivondronana of October 29, 1985 and letter to JIRAMA of February 20, 1986). Seven months after the cancellation, the Ministry of Industry was requesting financing of additional works on the remainder of the credit. The supervision mission of February 1986 noted the problem of coordination because the Project Unit did not inform the concerned agencies. 4/ The cancellation was proposed in the same telex (dated May 23, 1985) announcing the Cyclone amendment and the closing date extension. US$10 million was initially suggested, but the figure was eventually reduced to US$7 million after allowing for urgent rehabilitation works of san iation facilities and adding some equipment to the successful water component (memo dated June 17, 1985). The official telex to the Government announcing the IDA decision was dated July 2, 1985, and was acknowledged the next day by JIRAMA. - 7 - Maturity, Grace Period and Exchange Risk 1.18 The IDA credit was eventually onlent by the Government to the two agencies at substantially different terms, more favorable to the Fivondronana than to JIRAMA. On the Government's request, IDA agreed at negotiation to extend from 30 to 35 years, including 5 years of grace, the repayment period for funds onlent to the Fivondronana and to lower the interest rate to 42. After appraisal, the Ministry of Finance decided that JIRAMA would also get funds at 42 p.a. over 30 years with 5 years of grace instead of the 8% over 25 years recommended by IDA. Furthermore, the exchange risk was considered an *internal question*, meaning to be covered by the Government (letter from Ministry of Finance dated September 5, 1978). Given that the SAR assumed a 5.152 internal rate of return (FRR) for the water component (SAR, Annex 4.2), onlending at 4% was justified provided, however, that JIRAMA did not absorb the exchange risk. Yet, some confusion still remains about these terms. The SAR noted that the exchan*,e risk would be borne by the Government (para. 6.01) and that funds would be onlent to JIRAMA for 25 years at 8%, not 42 (PPAR, para. 1.19). Despite a Government commitment renewed in 1984 (memo following a meeting with the Minister of Industry, April 17, 1984), the exchange risk is still fully supported by JIRAMA. The 1988 external audit noted that the subloan agreement for the 1002-MAG does not stipulate repay- ment on the basis of exchange rates at repayment time.5 Although JIRAMA's financial situation would be improved considerably without the exchange risk burden, the external auditor and this Audit consider JIRAMA liable for the exchange risk short of an automatic coverage by the Government that would be reflected clearly in the balance sheets. It is regrettable that not enough leverage has been applied to have the Borrower honor its exchange risk coverage commitment. 1.19 At the end of 1988, the last year for which an audit is available, the accumulated exchange risk in JIRAMA's balance sheet amounted to FMG144 billion (US$94.3 million equivalent) on FMG39.6 billion of outstanding debt at historical rates, thus resulting into a almost fourfold increase in obli- gations. Among the loan portfolio, Credit 1002-MAG showed a 63% increase to FMG5.32 billion. Significantly, JIRMA. has not repaid its subloan to the Government and had accumulated arrears totalling FMG2.06 billion at the end of 1988. This Audit also found that, contrary to information reported above, Credit 1002-MAG is shown in the balance sheet as to be repaid over 20 years after the grace period, not the 25 years decided by the Government in 1978. In the same token, the interest charged to JIRAMA may not be as low as 4% although this information could not be verified.6 Given that the recalcu- lated ERR for the water component is only 1.22 which translates into an even 5/ This statement implies that JIRAMA does not bear the exchange risk on Credit 1002-MAG (external audit report for 1988, p. 8). 6/ JIRAMA's financial director indicated to this Audit that the rate was 82 which is consistent with the 82 charged on Credit 1526 and the 9.5% charged on Credit 1787 as reported in the 1988 audit. lower FRR, JIRAMA's default can be explained by its inability to generate enough cash flows to remunerate IDA subloans at 8% plus exchange risk. No comparable data was available in the file in the case of the Fivondronana. If it is charged the same 82 plus exchange risk, the resulting percentage is a negative return because none of the components which would have benefited the Fivondronana have been implemented. External Audit 1.20 The audit coverage under the project was uneven. The Government refused to produce separate accounts for the Fivondronana's Sanitation Division. It prepared, however, pro forma accounts which are acceptable for statistical analysis, but not for accounting and hence which cannot be audited. IDA insisted that accountants be hired by the Municipality to improve financial reporting, but none was provided for three years. 1980- 82 audits were available in 1984, but the external auditor had to be paid on the training component of Credit 1002-MAG. Under the first urban project, the external audit of all the Municipality's accounts has been implemented. 1.21 JIRAMA's audits were initially late by wide margins which reduced their usefulness (the 1979 accounts were available in June 1981; the 1982s in February 1984; the 1983s in January 1985). The delay has been reduced in recent years and this Audit was able to rely on audited accounts for 1988, ready in September 1989. As stipulated in the Project Agreement, the produc- tion of audits should be reduced to five months after the end of the year to monitor effectively JIRAMA's financial problems. An argument on the choice of external auditor developed in 1982 which reflects poorly on IDA staff's interpretation of audit rationale. Arguing a possible conflict of interest within the auditing firm between its audit and organization arms, IDA insisted on having JIRAMA change of auditor although a different team worked on each assignment (letter to JIRAMA, December 22, 1982). This evaluation found, however, two other reasons for this drastic recommendation. First, the external auditor refused to certify the accounts because the transfer of EEM assets to JIRAMA in 1976 had not been settled with the Antananarivo Municipality (settlement was being finalized in November 1989). Second, IDA considered that the failure to get its accounts certified harmed JIRAMA's credit standing. The argument would have been valid only if the underlying finances were healthy and if the audit report misrepresented their states. Yet it was never the case because JIRAMA's financial situation was indeed in disarray. 1.22 The point missed by IDA, however, is that, while the external auditor had good reasons to adopt a firm position, the audit report was not fully disclosing the extent of the financial problems either. A case in point is the audit for 1988 which, ironically, is the first one not to be qualified. The audit report disclosed in its Sources and Applications of Funds table a surplus in both 1987 and 1988 which was not really available. The Income Statement shows, however, that, before interest payments and provisions for depreciation, there were losses which implied that debt- servicing and provisioning generated more losses. As a result, self- financing of FMG5.16 billion claimed in 1987 and 4.06 billion in 1988 were - 9 - not real in cash terms. Furthermore, only 411 of the FMG9,503 million (US$6.2 million equivalent) of debt falling due was repaid in 1988. The balance was reallocated under the label *State's current account" which gave the impression that it was quasi-equity when the Government has yet to provide its contribution to debt-servicing. Institutional Development 1.23 The project hinged on important institutional building efforts which were crucial for its sustainability. As originally formulated, the project's physical objectives were manageable despite their multiplicity provided that some measures of institutional strengthening would have been implemented to support the ambitious sector objectives. The strategy to implement them was, however, deficient at the institutional level because both agencies had major handicaps. The water company had, and still has, structural financial problems, and the Fivondronana was, and still is, administratively and financially weak. 1.24 IDA's insistence on appointing six accounting staff at the Fivondronana7 was not justified in view of the real priority which should have been instead to provide the Fivondronana with financial management capabilities. They were needed because the Fivondronana was considered autonomous by the Government when JIRAMA requested the clearing of large arrears. By 1981, the six staff recruited for the project as Board condition were not working anymore for the project and 75 positions remained to be filled in the Sanitation Division (supervision report, August 18, 1981). The help offered repeatedly by JIRAMA was not taken up. Little institutional progress was achieved under the project, and it is only with the first urban project (approved in 1984) that the Municipality's management problems were addressed. 1.25 In contrast, IDA's insistence on appointing a financial director at JIRAMA was justified, but could not be implemented for several years. Although at one time during project preparation the appointment of a finan- cial director was to be a condition of Board presentation (Back-to-Office report, July 6, 1979), JIRAMA was without a financial director until its reorganization in 1987 mainly because the agency's top management did not share the rationale underlying IDA's recommendation. In the period 1980- 82, an expatriate advisor provided technical support to JIRAMA'S management. Yet, as an advisor, he was not empowered with the necessary authority (he was reporting to one of the technical directors) and, therefore, he had little influence on events. In 1982, IDA notified its willingness to finance the cost of the financial manager position, but this failed to induce action (supervision report, April 13, 1982). For several years, JIRAMA did not meet IDA's request to prepare a 5-year financing plan. Although measures have since been taken, this Audit ascertained that JIRAMA's financial management is not fully apprised of the depth of its financial problems, nor is the 71 Prior to negotiations, a mission recommended that it be a condition for Board presentation (Back-to-Office report, July 6, 1979). - 10 - external auditor (PPAR, paras. 1.22 and 1.29). JIRAMA's resistance to financial management expertise is hard to explain in view of its undertaking the visible Andelakela hydro project and being in contact with numerous lenders throughout the 1970s. It certainly explains the striking discrepancy between its well-run technical operations and the structurally-weak state of its finances. 1.26 The lack of success in convincing JIRAMA to divest from its industrial activities illustrates IDA's lack of consistency on institutional strengthening issues, due mainly to a failure to enforce its sound recommen- dations. At appraisal, it was recognized that JIRAMA's subsidiary, ANJARA, was a drain on the parent company's finances regardless of the merits of the venture and divesting was to be completed at the end of 1980. In January 1983, an agreement was reached to have the Fonds National d'Investissements (FNI) take a majority stake in ANJARA; a deadline of end 1984 was set in May 1984 for transferring the industrial activity. It did not take place because FNI lacked financing to consummate the deal. According to the 1988 audit of JIRAMA's accounts ANJARA and the light bulb plant had combined losses of FMG202 billion which were offset by the Treasury as part of the agreement under the Energy-I project (Credit 1787-MAG). The main problem remained, as identified ten years before, that non-utility activities are still managed by JIRAMA and consolidated in its balance sheet which is not desirable. 1.27 Pressure to reorganize JIRAMA has been applied by IDA since the beginning of the project with two objectives: to reduce both personnel size and operating costs. Designed originally along geographical departments to cover the whole country and functional responsibilities which mixed the two different business activities, JIRAMA had become a large employer with 3,900 staff at appraisal; by 1981, it increased to over 4,800 (PCR, para. 4.01). In 1982 IDA requested JIRAMA to submit a plan to reduce costs, including payroll through personnel cut. It took five years and two consultants (EdF, then CEGOS) to carry out the reorganization study. While pushing JIRAMA to reorganize, IDA kept an arm-length position on the issue (PCR, para. 5.02) and was skeptical of the outcome as suggested by the fact that both CEGOS and JIRAMA felt compelled to send letters to advocate the proposed reorganiza- tion. The new organization reduced the hierarchy layers from to four to two and promoted increased decentralization along three departments: water, electricity, coordination and control. Financial Performance 1.28 Over the period 1980-86, JIRAMA was able to comply with the covenanted rates of return only once and yet mainly due to a short-lived financial restructuring. At appraisal, JIRAMA's stated objectives were to self-finance about 64% of investments over the period 1980-86 and to increase tariffs to achieve rates of return of 2.4Z in 1981 and 1982 and 6% there- after. The PCR claims that these objectives were partially achieved (PCR, para. 4.05). This Audit does not find this to be a correct statement unless results are taken in isolation of each other to prove covenants one at a time. The only way JIRAMA could have financed 49% of its investments during the period 1980-86 is by generating FMG6.04 billion of free cash flows, that - 11 - is, net of debt service. According to the PCR (Table P. 21), net cash flows amounted to FMG8.34 billion. This is, however, a substantial overstatement because the debt service deducted from the net cash-flows was under- estimated; for example, instead of the FMG5 million shown for 1986 FMG347 million were due on Credit 1002-MAG alone (see PPAR, para. 1.32). Not only the covenanted debt-servicing coverage ratio was not met (1.35 in 1987 and .69 in 1988)8, but the utility's earning power was actually lower than dis- closed. Revenues should be adjusted for the large increases in receivables or arrears, which amount to foregone income, and the surplus on water opera- tions could not be used because of the losses on the electricity side. On a consolidated basis, losses amounted to FMG20.7 billion, to be added to FMG20.1 billion of outstanding receivables at the end of 1986. In other words, while it is statistically correct that the water department made a positive contribution to earnings, it was not available on a cash basis to finance investments in the sector. Similarly the PCR claims rates of return of 7.72 in 1985 and 3.21 in 1986, but notes that surpluses were "not avail- able' before (Table P. 20). These results could not be duplicated because the PCR does not allocate fixed assets between water, power and industrial activities. Various supervision reports quoted the missing of targets in 1980 and 1981 and -1% in 1982. Oddly enough, from 1985 to 1986 all super- vision reports repeated the 4Z achieved in 1983, but gave no figure for 1984 noe 1985. The 1988 report by the external auditor estimated the rate of return for water operations at 3.5Z in 1987 and 2.4% in 1988, levels well below covenanted targets. 1.29 The main drain on JIRAMA's finances had been identified early on as consumers' unpaid bills accumulated, mostly in the public sector broadly defined (ministries, agencies, universities, and municipalities). IDA has kept pressure on the authorities to reduce these accounts receivable since negotiations of this project in 1979. Six months after credit effectiveness, the Government was asked to reduce arrears to less than three months of billing. Practically insolvent because of mounting receivables, JIRAMA stopped paying its suppliers, a pattern which generally underscores the structural nature of a liquidity problem. The Government, without funds to clear its own arrears, indicated in 1982 that it lacked control over the municipalities, hence could not force them to clear their arrears (super- vision report, April 13, 1982). Indeed, MOF proposed a schedule for paying FMG2 billion due by ministries, but refused to cover arrears from the 8auto- nomous agencies' (memo, May 31, 1983). In 1984, MOP issued firm instructions to the Government's various administration to pay JIRAMA; FMG1 billion had been paid in 1983 to clear pre-1982 arrears (which totaled FMG9.8 billion), the rest to be paid during 1984. There was, however, some apprehension that current bills for 1984 could not be paid in addition to clearing arrears (memo after meeting with delegation, April 17, 1984). Shortly after, the Government's commitment was revised to include FMG3.2 billion in 1984 and 8/ Using the more favorable interpretation of Section 4.04 of the Project Agreement. When excluding non-operating losses, the figures were .29 in 1987 and negative in 1988. For the water activity alone, they were 1.55 and .69, respectively. - 12 - the rest (FMG4.8 billion) to be paid in 1985. The target was to lower receivables to three months of billing which was ambitious since it had never been achieved before. Actually the Government cleared FMG3.9 billion of arrears in 1984, but as much was added to the stock from unpaid billing the same year. Although the latest supervision missions did not focus on this issue, the problem still remains as evidenced by statements in the 1988 audit of accounts. State arrears have been cleared up to the end of 1986, but the years 1987-89 are still shown as receivables. Furthermore, arrears from universities (FMG5 billion as of September 1989) have not been addressed despite their unlikely recovery and the absence of provisioning. However, procedures have been improved by the signing of 'contracts" between public consumers such as the Fivondronana and JIRAMA obligating the former to pay or risk supply cutoffs, and the latter to generalize metered billing to avoid disputes. 1.30 The PCR credits the 1986 Energy-I project for improving the financial situation. This Audit concluded that the otherwise-awkward threat of disbursement suspension was a key factor in the better results recorded in the water activity in 1985 (PPAR, paras. 1.35-1.36). Profits trebled and cash flows doubled; for the first time, receivables declined, the current ratio turned positive and the covenanted rate of return was met. The improvements did not last, however, because the actions were overdue stop- gap measures.9 JIRAMA's mandate is still that of a technical utility which has not been given the effective means to become financially autonomous. 1.31 The table below shows that the benefits of tariff increases can be entirely erased by receivables if they get out of control. Between 1980 and 1986, there were six tariff increases amounting to an aggregate 134%. As receivables rose from 120 days in 1980 to 298 in 1986, they offset entirely the benefits derived from tariff increases. Indeed, when sales are reduced by the increase in receivables, their growth rate was only 16% for the whole period; moreover, all of it was accounted for by increase in sales quantities (18?). In contrast, as receivables declined to 197 days in 1987 and 1988, the 31% of tariff increases then enacted by JIRAMA translated into actual revenues. On a compounded basis over the last eight years, tariffs were effectively raised by 109? instead of the 206? expected from nominal increases requested by IDA covenants. It is as if one out of two increases had been wasted. 9/ A recent aide-memoire overstated the strength of JIRAMA's - ancial situation, claiming inter alia a debt-service coverage ratio of 250% which seems dubious given that JIRAMA is defaulting on the majority of its loans. - 13 - 1980 1986 1988 Vo.ume sales 100 118 126 Sales revenues 100 237 324 Sales adjusted for receivables 100 116 264 Tariff increases (nominal) 100 234 306 Tariffs adjusted for receivables 100 98 209 1.32 JIRAMA is servicing its debts very selectively. In 1988, 412 of all principal due were repaid. The rest was moved to waiting category called OCompte-courant de l'Etat" which implied that the Government would cover the obligations whereas no commitment had yet been obtained. Out of 22 loans currently in their repayment phase, only ten are being Pqrviced. These ten loans are concentrated over six lenderst one is local (BI), two are French (Caisse Centrale de Coopdration Economique and BNP) and one Swedish (Svenska). None of the IDA Credit 1002-MAG due since 1985 has been repaid yet to the Government; by the end of 1988, these arrears amounted to FMG2.06 billion, of which 63% were accumulated exchange risk due to the difference between rates at time of disbursement and current rates. Credit 817-MAG was in the same non-accrual status.10 1.33 The exchange risk is the key factor in the difficulties facing JIRAMA in managing its liabilities. By the end of 1988, the accumulated risk was FMG158.6 billion on 7MG44.4 billion of original borrowings; during 1988 alone the exchange risk added FG14.92 billion or 92 in obligationst that was also 11 times disbursements of new loans or 6 times water and electricity revenues increase in 1988. The Government has accepted to cover the exchange risk on electricity loans up to 1986 only, by consolidating it into a portion of the FHG17.4-billion capital completed in 1988. Because of the remaining uncovered risk, the debt-equity ratio has deteriorated from 1.98 in 1987 to 2.45 in 1988, a range which does not qualify JIRAMA as creditworthy. Given that JIRAMA has no means to protect itself from foreign liability risk, a clear ex ante policy dealing with the financing of the exchange risk would go a long way toward removing a major uncertainty in JIRAMA's finances and to improve its creditworthiness. The Threat of Suspension of Disbursements 1.34 Borrower-lender relationships were marred with various displays of pressure on both sides. Besides the threat of suspension detailed below, the list is varied. The leverage of new project processing (e.g., first urban project) was used in May 1984 to make progress in the financial restructuring of JIRAMA recommended five years before. There was a letter from the President of the Republic to the Bank's President lobbying about a smaller tariff increase (April 3, 1984). A letter from the Planning Directorate blamed Bank missions with too many changes on the institutional framework 10/ JIRAMA's financial director indicated to this Audit that it had been paying interest on IDA credits for one year. - 14 - for the slow drainage component (July 27, 1984). Correspondence over a procurement case was at times abrupt (PPAR, para. 1.37). Although not coomented upon in the PCR, the suspension threat was the main leverage used from January 1983 to May 1984. 1.35 After two years of writing aide-memoires and letters about the deteriorating financial situation of JIRAMA without getting neither reply nor action, IDA staff recommended to Management to give the authorities 60 days to answer or to stop preparing new project with JIRAMA (supervision report, January 28, 1983 and letter to JIRAMA, February 8, 1983). Without reply four months later, the pressure was increased. If, before the end of June 1983, the Government had not taken the following actions: (1) to increase tariffs within three months; (2) to pay arrears within one year; and (3) to undertake a cost-reduction program within three years, disbursements of Credit 1002- MAG were to be suspended (supervision report, May 31, 1983). IDA management waited, however, until after the Annual Meetings to communicate this intent to the Malagese authorities (letter to Minister of Finance, October 3, 1983). After another three months of waiting the Government announced a series of positive actions. IDA deemed them sufficient to lift the threat of suspen- sion provided that water tariffs would be increased by 50% either immediately or gradually (memo to Mr. Stern, February 14, 1984 and telex from VP to Minister of Finance, February 21, 1984). Actually the Government decided later to raise tariffs by only 202 while requesting that the deadline for suspension be postponed until after the visit by the Bank's President (telex from resident representative, March 7, 1984) and again after a visit by a delegation in Washington (telex from chief programs division, March 20, 1984). This argument prevailed because the SVP cabled *not to suspend' as the Bank accepted the Government's intent to find an acceptable solution to JIRAMA's problems (telex from resident representative, March 26, 1984). Although the Bank's President did not get into detail of the tariff issue, Madagascar President wrote him to accept only a 2C% increase in 1984 (letter of April 3, 1984). Indeed, this was the only increase enacted in 1984. 1.36 This extraordinary intervention had the effect of diluting the use of the suspension threat and for new levers IDA turned to both the scheduling of appraisal of a follow-up project (letter to Minister of Industry, April 19, 1984) and to the processing of the urban project (telex to Minister of Finance, May 15, 1984). This worked within ten days (telex from Minister of Public Works, May 23, 1984 and letter from Minister of Industry, May 28, 1984). The urban project was approved by the Board in June 1984, but the financial restructuring of JIRAMA had to wait early 1985 to be effective. Yet it was good for only one year and further improvements were requested under the Energy-I project. This Audit failed, however, to see any durable improvements in the distress which characterizes JIRAMA's financial situation. Procurement 1.37 The procurement process was affected by a questionable bidding over the sewerage component which led IDA to take a firm stance leading eventually to the cancellation of the corresponding credit amount. This Audit is not - 15 - questioning the rigid application of procurement guidelines, but submits that, given the context (e.g.. lack of previous experience, IDA's insistence on not delaying further the sanitation component), a more constructive approach could have been tried to salvage a long effort. For example, this could have taken the form of a mission in October 1984 to help authorities to restart the bidding process.11 Considering that, to this day, the sewer- age investment planned in the late 1970s has not taken place, the Audit concludes that it is a heavy development penalty paid for a correctable misprocurement. The Fivondronana had difficulty obtaining spare parts for the trucks. Non-payment of the Transactions Tax by the Fivandronana, and the subsequent withholding of spare parts by the supplier, created a situation of mutual non-cooperation which prevented progress from being made. Soon the fleet was inoperative. Conclusions and Lessons 1.38 The main and remaining tangible result from this project is the extension of the JIRAMA production and distribution capacities which delivers essential water to a growing population. It Is, however, generally the case with infrastructure projects that, once the procurement hurdle has been passed, the physical components are satisfactorily Implemented, albeit often with time and cost overruns. A lesson drawn from this project's experience deals with the need to hedge against institutional risks. If the Institu- tional risk is substantial, priority should be given to offset problems, among others, of coordination between implementing agencies or of experience by ensuring that the physical content of the project is as detailed-prepared as possible before credit approval, using feasibility studies and procurement technical assistance. 1.39 As originally formulated, the project's physical objectives (PPAR, para. 1.07) were manageable provided that some measures of Institutional strengthening would be implemented to support the ambitious sector objectives (PPAR, para. 1.05). The strategy to Implement them was, however, deficient at the institutional level because both intermediaries had major handicaps. The water company had, and still has, structural financial problems, and the Fivondronana was, and still is, administratively and financially weak. The humbling lesson from this project Is that stating objectives and targets for a development project is not sufficient if the mean selected to achieve them are Inadequate for the tasks which akes the whole strategy flawed to start with. The correct procedure should be to tonlate a balanced strategy first, then to tailor it - by cutting components if necessary - according to 11/ The calendar as evidenced by the file is as follows: September 13, 19841 telex to the Minister of Public Works requesting to postpone the bidding launch; October 5, 1984: telex to same requesting to stop the bidding launched on September 15; October 17, 1984: telex from Minister of Publics Works announcing that bidding was stopped; February 11, 1985: project rating increased from 3-3 to 2-1 because of the decision to cancel the sanitation component; June 17, 1985: internal memorandum proposing to cancel $7 million on June 30. - 16 - the absorptive and delivery capacities of the implementing agencies upon which the project outcomes lay. This appears to have been the rationale used for cutting the drainage and sewerage components after the procurement process had gone astray (para. 1.37). 1.40 Multi-components projects are inherently difficult to implement satisfactorily because they cannot be deemed successful if only one component works. Overloading a single and first operation with too many components in an untested environment was a major bet on the absorptive and implementation capacities of largely bureaucratic organizations. It was lost early on dur- ing implementation as the cancellation of one of the two largest components and the financial restructuring of one of the two entities, both in 1985, eventually illustrated. The obvious lesson from this project is that IDA should refrain from the temptation to try and solve all problems at once through complex projects in economies whose structures have not reached the needed maturity to handle parallel components. 1.41 This project demonstrated that success relied in good part on a strong technical tradition which, in this case, started when the utility was in private French hands and which has been maintained by its high level management issued from the former organization. It also demonstrated that technical tradition is not a sufficient condition if not supported by a fUnancial tradition that aims eventually at as broad as possible autonomy. The attempt at financial restructuring made in the case of JIRAMA in 1985 was too short-lived to have had a durable effect. 1.42 The discrepancy between IDA credit terms (e.g., 0.75% and 50 year maturity) and that for subloans is irreconcilable and inconsistent with the effective cash flow generation ability. If the ERR of the water component was estimated at 7.2% and the FRR at 5.1%, it is difficult to envisage how JIRAMA would have been able to remunerate a loan at 8% plus a substantial exchange risk. A fortiori, when the ex post ERR was reduced to 1.22. A lesson to be drawn from this situation suggests the need to test whether sub- borrowere can afford IDA's onlending terms which stipulate grace period as short as three years and maturity as low as 20 years. Furthermore, to peg the interest rate charged to IDA beneficiaries to the IBRD loan rate plus one or 1.5 points should not be automatic, but must be justified in every case on the basis of risk- adjusted cash flows projections. This project demon- strated the inability of an-otherwise technically competent-agency to repay 4.52 p.a. of its principal debt, let alone an 8.5? annual interest. In this respect, financing local costs with IDA funds is expensive because of the underlying exchange risk which is all the more potent that the risk is stretched over a long maturity.12 1.43 Conditionality had mixed results in its application. There was no effectiveness conditionality despite the fact that some were clearly outlined at pre-appraisal. Cancellation with little warning was used to erase a con- 121 Project staff state that this case illustrates when routine application of formal Bank policy turns out to be inappropriate. * 17 - ponent which had still some implementation potential and has had no replace- ment from any donor in ten years. Indeed, the fact that waste water disposal investments were not undertaken in parallel with the water extension means that Antananarivo is still declared a malaria zone among other waterborne parasitic diseases. Suspension of disbursements was threatened for an exces- sively long period to implement a financial restructuring which should have been done without pressure because it was to lift JIRAMA from insolvency. The main lesson is that suspension threat is not an effective institution- building tool if not executed for a long period (over 18 months in the case of Credit 1002-MAG). Quicker results were obtained by threatening to post- pone processing follow-up projects or by making cross-conditionalities in other loans. 1.44 Overall, the external audit of the implementing agencies' accounts was a tool underused by IDA staff given that structural financial problems were evident since appraisal. The audit reports offered an yearly opportun- ity to disclose the true financial situation of JIRAMA in order to accelerate decision about its needed financial restructuring. Too much time was spent arguing about the auditor's repeated qualification of JIRAMA's accounts and not enough to the fact that its cash flows situation was overstated. To this day, audited accounts are presented according to the local Accounting Plan; the forthcoming changes should bring in better light the financial flows. Given the magnitude of the billing arrears, however, nothing prevented requesting the external audit to present the financial situation more on a cash basis than it had been done. 1.45 JIRAMA's financial predicaments are the accumulation of receivables on the assets side of its balance sheet and of exchange risk on the liabilities side. Payment arrears from clients can be reduced if not elimi- nated; cutting supply is a lever which should be used to educate consumers in responsible behaviors. In this case, with 95Z of arrears due from Government and the municipalities, cutting supplies to Government services proved to be unpalatable. The exchange risk is outside JIRAMA's control, except for the temptation to borrow abroad for local cost investment which should be refrained because it is not economical. Furthermore, being a domestic producer, JIRAMA has no means to hedge at a reasonable cost against foreign currency risk. By raising its debt-equity ratio into uncreditworthi- ness range, the exchange risk impedes the future borrowing ability, hence puts the financing of necessary investment in jeopardy. As long as the exchange risk will be borne effectively by JIRAMA, despite legal indications that it is covered by the Government, JIRAMA's finances will become weaker every year since the exchange risk increases faster than sales revenues. The lesson to be drawn from this case is that a utility which has neither the means nor the experience to hedge against the exchange risk should not bear it. However, a fixed portion of this additional cost of borrowing in foreign currencies should be passed on as an additional interest rate to the final borrower.