Burundi - Urban Development Project

Document of The Wcrld Bank FOR OFFICIAL USE ONLY Report No. 7898 PROJECT PERFORMANCE AUDIT REPORT BURUNDI FIRST URBAN DEVELOPMENT PROJECT (CREDIT 1049-BU) JUNE 30, 1989 Operations Evaluation Department 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. FOR OFFICIAL USE ONLY THE WORLD BANK Washington. D.C. 20433 U.S.A. 00ace n Drector-CeneraI Operatums Ev stuatiom June 30. 1989 MEMORANDUM TO THE EXECUTIVE DIRECTORS AND THE PRESIDENT SUBJECT: Project Performance Audit Report on Burundi - First Urban Development Project (Credit 1049-BU) Attached, for information, is a copy of a report entitled "Project Performance Audit Report on Burundi - First Urban Development Project (Credit 1049-BU)" prepared by the Operations Evaluation Department. Attachment This document has a restricted distribution and may be used by recipients only in the performance of heir official duties. Its contents may not otherwise be disclosed without World Bank authorization. ,i.s ALL PROJECT PERFORMANCE AUDIT REPORT BURUNDI FIRST URBAN DEVELOPMENT PROJECT (CREDIT 1049-BU) TABLE OF CONTENTS Page No. Preface .......................................................... i Basic Data Sheet ................................................. ii Evaluation Summary................................................ iv PROJECT PERFORMANCE AUDIT MEMORANDUM I. SECTOR BACKGROUND ...................... ............... 1 A. Urban Development in Burundi ......................... 1 B. Institutional Framework .............................. 2 C. National Policy for Urbanization ..................... 3 II. THE PROJECT ............................................ 4 A. Project Background ................................... 4 B. Project Objectives ................................... 5 C. Project Description................................... 6 III. PROJF.M IMPLEMENTATION AND ACHIEVEMENTS ................. 7 A. Time Delays and Cost Overruns ........................ 7 B. Project Changes ...................................... 10 C. Project Achievements ................................. 11 IV. POINTS OF INTEREST ..................................... 17 A. Physical Impact ..................................... 17 B. Institutional Impact ................................. 18 C. Policy Impact ........................................ 19 D. Poverty Impact ....................................... 20 E. Other Points of Ir*erest ............................. 22 V. THE ROLE OF THE BANK .................................... 24 VI. LESSONS LEARNED ......................................... 25 A. Lessons for Physical Works ........................... 25 B. Lessons for Municipal Strengthening Programs ......... 26 C. Lessons for Urban Development ........................ 27 D. Lessons for Economic Development Programs ............ 27 E. Lessons for Project Design ............................ 28 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. TABLE OF CONTENTS (continued) Page No. VII. CONCLUSIONS,............................ .............. 28 ANNEXES 1. Comparison of Project Costs............... ................ 29 2. Photographs .................................................. 30 3. Comments from the Borrower.................................... 38 PROJECT COMPLETION REPORT I. Introduction........... ............................................. 41 II. Project Preparation and Appraisal........................ 42 III, Implementation........................................... 44 IV. Institutional, Financial and Economic Performance........ 54 V. The Role of the Bank..................................... 58 VI. Post Evaluation.......................................... 59 VII. Conclusions.................................................. ........... 60 ANNEXES PROJECT PERFORMANCE AUDIT REPORT BURUNDI FIRST URBAN DEVELOPMENT PROJECT (CREDIT 1049-BU) PREFACE 1. This is a Project Performance Audit Report (PPAR) on the Burundi rirst Urban Development Program. Approved on June 24, 1980, the project was supported by Credit 1049-BU for US$15.0 million which was totally disbursed. 2. The PPAR consists of: (i) an Evaluation Summary and a Project Performance Audit Memorandum (PPAM), written by the Operations Evaluation Department; and (ii) a Project Completion Report (PCR), drafted by the staff of the Infrastructure Operations Division of the South-Central and Indian Ocean Department (AF3IN). 3. The PCR provides a short account of project experience, including the achievements and failures, without elaborating on the reasons for these achievements and failures. To give a wider perspective, OED has attempted to provide additional information about the history of the project, its eventual outcome, the issues raised, the problems of implementation and possible alternative approaches. The memorandum is based on the PCR, on the completion report prepared by the project unit in Bujumbura, on the SAR, on project files, legal documents and other relevant material. It also drew from interviews with Bank staff members, with representatives of the Burundian government, and tbe urban project team as well as with consultants who worked on the project. 4. The Audit agrees in most respects with the PCR. It also stresses the key role played by Bank staff in influencing the project's outcome. 5. Following standard OED procedures, copies of the draft PPAR were sent to the Borrower. The comments received from the Ministry of Public Works and Urban Development are reproduced as Annex 3 to the PPAR. - 11 - PROJECT PERFORMANCE AUDIT MEMORANDUM BURUNDI FIRST URBAN DEVELOPMENT PROJECT (CREDIT 1049-BU) BASIC DATA SHEET KEY ?OJECT DATA IteS Appraisal Actual Estimate Total Prol-ct Cost (USS million) 16.7 16.8 Cost Overrua (1) 0.0 Credit Asmuat (US$illion) 1.0 15.0 Disbursed 15.0 15.0 Caacelled 0.0 0.0 Date of Physical Coaspletion 12/31/84 12/31/87 Time Overrun (3 ... 72.0 Econasic Rate of Return W1) 11.0 13.3 CUMLATIVE D1IURSEHENTS I US Thousands Fiscal Year Appraisal Estimate Actual ActuallEstisatatl 1981 800.0 1116.4 139.6 1982 3200.0 2802.8 90.1 1983 8000.0 5433.7 68.0 1984 1200.0 7460.4 59.2 1985 15000.0 9063.8 60.4 1986 - 11128.5 74.2 1987 - I*388.4 95.9 1988 - 15000.0 100.0 PROJECT DATE,- Ites Appraisal Actual Estimate First Ilention in Files 03110177 Negotiations 04/30/0 03115-16180 Board Approval 0626/8r 06/24/80 Credit Agreement Date 07103/80 Effectiveness Oate 10/31/80 Closing Date 12131/85 12/31/87 - iii - MISSION DAT 9o. of Na. of Staff- Date of NonthiYr Days Persons Weuks ieport Idestification 03/78 10 2 4.0 04/04/78 Preparation I 07/78 8 T 4.8 08108/78 Preparation 11 11178 5 1 1.0 Preparation III 017 8 1 1.6 02/28179 Preparation IV 04/79 7 3 4.2 06114/79 Preparation V 0179 10 6 12.0 07/26179 Preappraisal 10179 12 7 14.8 10130/79 Apressal 12179 15 7 1.0 06/02/80 Post-Appraisal 02180 11 * 6.6 Supervision I 10/90 10 1 2.0 11/24/80 Supervision 11 02/81 11 2 4.4 03/24181 Supervision 11 07/81 9 3 5.4 08112/81 Supervision IV 12181 6 I 1.2 01/29182 Supervision V 06/82 12 3 7.2 07/15/82 Supervision.VI 02183 8 2 3.2 04/11183 SPevisim VII 06/83 4 1 0.8 07115/83 Supervision VIII 03184 9 3 5.4 06111/84 Supervision II 07/84 7 2 2.8 08148/84 Supervision I 03/85 8 1 1.6 05107/85 Supersion II 11/5 5 I 1.0 12118/85 Supervision III 06/86 5 1 1.0 07130/86 Superwvision IIII 05/87 5 3 3.0 06/18;87 Completion 03/8 5 1 1.0 112.0 OTHER PROJECT DATA Borrower Republic of Burundi Executing Agencies -Ninistry of Public Norki (Project Unit) -Nuncipality of Oujusbura Fisral Year of Borrower Jauary I - Deciber 3t Follow-on Project Second Urban Project EICHAIIE RATES Currency Unit : Burundian Franc ( FBa ) Anual Averages YEA 1 1980 through 1982 1983 1984 1995 1986 1917 US 143 90.0 93.0 119.7 120.7 114.0 123' 5 - iv - PROJECT PERFORMANCE AUDIT REPORT BURUNDI FIRST URBAN DEVELOPMENT PROJECT (CREDIT 1049-BU) EVALUATION SUMMARY Introduction 1. While Burundi's urban population accounted for less than 10% of the total in 1985, it has been growing very rapidly. The capital, Bujum- bura, reports a current growth rate of 8, up from 6Z at the time of Appraisal. In view of the rural sector's limited capacity to absorb future populations, Government policy now calls for increased urban development, supports decentralization and municipal strengthening and encourages urban economic activities and the informal sector in particular. GOB interest in urban development was sparked during the 1970s when it fostered 15 special programs for housing and infrastructure in Bujumbura, all financed with foreign assistance. Bank lending programs had concentrated on rural devel- opment until the late seventies, when the GOB's growing support of urban projects opened the way for Bantc involvement in the sector. Objectives 2. The Government of Burundi requested a US$10 million loan for urban upgrading (some 8,000 to 10,000 sites) and new lot development (3,500-4,000) in September of 1978. The Bank agreed to the GOB's general goals, but proposed to increase the project cost to US$20 million, adding about 10,000 construction loans, schools, clinics, and informal-sector support. During the preparations which followed for another year and a half, the health component was dropped, while additional upgrading and maintenance works and an experimental energy conservation program were added. Project objectives were subsequently extended from the original shelter and poverty focus to include a wide range of goals explained by the project officer as "not a large-scale attempt to rehabilitate slums or eliminate poverty, but (as) a modest series of advances on a broad front." 3. Bank preparation missions also focused on municipal institutions, noting Bujumburas' "absolute low level" of income, plus the budget sur- pluses which accumulated due to its "inability to increase levels of opera- tions rapidly enough to spend the resources it takes in". Pinpointing the city's "inadequate operating capacity", the project proposed to bolster municipal revenues through the introduction of property and rent taxes and to assign the city new functions in the areas of maintenance, housing, and tax collection. Maintenance and public services costs were to be covered by the new taxes, while investment in upgrading and shelter were to be charged directly to the beneficiaries. - v - 4. Fifty percent of project cost was devoted to shelter components; with 7? for development of a program of about 800 serviced lots at Kwijabe, 19? for home construction and improvement loans, and 20Z for the upgrading of six neighbox.oods containing about 1',000 households and about three quarters of Bujumbura. Upgrading was to provide roadways, stormwater drainage, public lighting, and public standpipes. An idditional 13? of the projeut was designated to equip and to establish a municipal service de- partment, charged with street and service maintenance as well as with garbage collection. School construction, an artisan promotion program (which included the development of a low-energy stove), design, management and technical assistance accounted for the remainder of the project. 5. Project preparation emphasized the project's social and economic aspects more than technical and engineering questions. Financial aspects were not fully analyzed, in part because of the inexperience of the con- sulting firm charged with developing the feasibility study. One result appears to be the number of contracts executed by force account. Since local expenses and construction by direct administration were to be reim- bursed without cosplete documentation, over 60% of the project was subject only to spot reviews. Implementation Experience 6. While the physical works component started up on time, the con- struction projects by direct administration experienced cost and time over- runs, while progress on the cost recovery mechanisms floundered. When the property tax ran into difficulties in Burundi's newly formed Congress and it became apparent that urban upgrading could not be recovered through direct charges, as proposed, the project officer moved to cut back civil works until a viable cost recovery mechanism was in place. When the proj- ect's termination dace approached with no satisfactory solution in sight, the remaining construction and acquisitions, amounting to 17? of the over- all project cost, were eliminated. 7. In consequence, the upgrading component was cut by 85%, eliminat- ing five of the six neighborhoods originally slated for improvements. This has had some unfortunate side effects including serious flooding in surrounding neighborhoods, since the key primary stormwater collecters were eliminated from the program, and drains which were constructed under the project have no place to discharge. Greater participation by a sanitary or hydraulic engineer might have stemmed the problem. At the same time, in- vestment in the city maintenance department was also cut back by 40Z; plans for building and road maintenance teams were suspended. 8. In 1985, a new project officer proposed to extend the project and to make use of the funds which had been suspended earlier by increasing the community facilities component through the construction and repair of some twelve additional schools. Although the need for additional schools had not been mentioned in the appraisal report, the Municipality argued that the Bank had barred urban schools from its lending program for education and that schools were always needed. Even though two of the new schools - vi - are currently used as craft workshops, the expansion and improvement of basic educational facilities are perceived locally to have generated a significant benefit to Bujumbura's population. Results 9. The project, through its cost recovery component, led to the establishment of a highly effective property tax system. By the end of I' 1983 the Congress voted to pass two existing taxes, one on assessed values and other on rental incomes, to the city. Under the project some 15,000 properties were registered and assessed on the city's tax rolls. Municipal income increased by 130% as a result in 1984. 10. The home construction loans program was designed to assist new residents of the serviced lots project and was complemented by a home im- provement loans program for families from target neighborhoods. In both cases the original loan amounts were increased considerably to encourage demand. Recovery of initial capital investment has been poor but for dif- ferent reasons. While the owners of serviced lots are making payments on time, the Government mpde each borrower an additional, non-reimbursable, grant as part of the loan. Beneficiaries of home-improvement loans, on the other hand, are behind in repayments. 11. Promotion of artisans proved more difficult than foreseen at Appraisal. When demand for loans proved slow in forthcoming, the component shifted its emphasis from credits to training and established workshops for carpenters, metal-workers, seamstresses and tailors. As a result, impact was lower, and cost/beneficiary higher than exp3cted. The energy conserva- tion component fostered the development of an effective peat-burning stove, which saves on charcoal consumption, but it has not managed to make produc- tion costs competitive nor self-sustaining. 12. Project completion took three years more than foreseen in the Appraisal timetable. Nonetheless, the project completion time compares favorably with Bank urbar. project averages and with other Bank projects in East Africa. A second project, emphasizing housing, infrastructure upgrad- ing, artisan promotion and community facilities, is underway in five Burun- dian secondary cities. It is scheduled for completion in seven years. Sustainability 13. Positive physical impact can be seen in the 780 new serviced lots, in increased public services, such as garbage collection, road and drain maintenance and, above all, in the public water standpipes, augmented by 80Z. This must be measured against project shortcomings, such as the flooding mentioned earlier and problems with inadequate drainage and la- trine design. The city's limited maintenance capacity has made it difficult to keep up roads and drains in upgraded neighborhoods. The sustainability of benefits resulting from project infrastructural investments is, thus, uncertain. At the same time, Bujumbura's population has grown dramati- cally. Municipal services have kept up with the pre-project levels, while planning and development of new neighborhoods has fallen behind. - vii - Findings and Lessons Learned 14. The projeet's direct poverty impact through the provision of .:sasing and income generation was mixed. Beneficiaries of the serviced lot project tend to have 4igher incomes than anticipated, while the artisan promotion program has reached only a fraction of the city's craftsmen and has left the informal economy quite untouched. On the other hand, the project has probably had an important indirect poverty impact through the creation of two markets which account for the daily occupation of several thousand persons, and through job generation, direct and indirect, in areas subject to upgrading, where subsequent construction activity has been impressive. 15. The project had important policy and institutional impacts. Of the four institutions developed by the project only one, the tax collection bureau, has been absorbed by the municipality of Bujumbura. The municipal services department (SETEMU), the housing office (ECOSAT) and the project office's artisan support program (DUB) remain autonomous. They have, none- theless, found important roles to play in national, as well as, municipal development. Insofar as the project has reinforced GOB's decentralization and municipal strengthening policies, as well as its poverty orientation and support for informal urban economies, the various institutions will be important in the development of subsequent projects in secondary cities. 16. The project managed to complete a civil works program and, at the same time, to meet its revenue mobilization goals, a rare achievement. The successes on each side must be qualified. With reference to cost recov- ery, Bujumbura's new financial capacity is not matched by an increase in its operational capacities. And regarding civil works, the increased school construction did not necessarily address Bujumbura's priority urban needs. 17. Serviced lot projects were originally conceived as a means of controlling urban sprawl and stemming the disorderly spread of spontaneous settlements. But serviced lot programs tend to be implemented on too small a scale to have much impact on overall urban development. Furthermore, by virtue of their requirements, in particular for residency and steady income levels, serviced lots tend to be beyond the reach of most urban*migrants and thus never play an effective role as *reception areas'. Long-term rational planning for urban development must be based on mechanisms other than serviced lots. 18. This project offers several examples of how local materials, ap- propriate technologies and low-energy systems may not be compatible with economic development, or with employment generation programs because, in the end, they cost more than standard technologies and are not affordable by low-income, informal sector families. 19. The advantages of contracting projects over construction by force account became quite clear in this project. Construction by force account took longer, appears to have been more costly and difficult to document, and, at the same time, it defeated the purpose of a simultaneous bank proj- ect designed to reinforce the private sector construction industry. - viii - 20. Engineers from relevant fields should participate in appraisal and supervision of projects with a high percentage of physical works. Timely consultations with experts in soils engineering, hydraulics, sanitation and civil construction might have saved the project from some of its negative results. 21. The project opted for a series of half-doses; a limited amount of garbage collection, road maintenance and drainage clean out. The project results raise the very pertinent question whether civil works and mainte- nance projects should be designed to give complete coverage in one area, rather than partial coverage in several. 22. The larger and more complicated a project, the longer it will take to execute. As a corollary, projects which hinge on the introduction of new taxes, new laws or new cost recovery practices, will generally face resistance. An ample time frame should be applied. 23. In this case, it appears that the Bank may have proposed and sup- ported a project larger than it felt the GOB could readily handle in a four year disbursement period. Subsequent problems in implementation and ef- forts to reduce project scale once execution began suggest that both time frames and existing local institutional capabilities should be more closely considered at the outset. PROJECT PERFORMANCE AUDIT MEMORANDUM BURUNDI FIRST URBAN DEVELOPMENT PROJECT (CREDIT 1049-BU) I. SECTOR BACKGROUND A. Urban Development in Buri.idi 1.01 Burundi is one of Africa's smallest yet most densely populated countries. With an estimated Gross National Product of US$240 per capita, Burundi ranks among the world's poorest countries. As food crop production has not been kept up with population growth, and export crop production has actually declined, recent economic growth has been concentrated in the urban sector, particularly in manufacturing and construction activity. 1.02 While its urban population accounted for less than 10? of the total (about 4.7 million), in 1985 Burundi's capital, Bujumbura, housed about 7.5Z of the nation's families and is presently growing at about 8Z per year (from 6? at the time of Appraisal). Government policy currently calls for increased urban development given the view that the rural sector cannot continue to absorb population increases in the future. The first urban project, which aimed to raise the standard of living in Bujumbura served as the first step in this national strategy (now being followed up through the second urban project in Burundi). 1.03 At time of project preparation about 105,000 persons, or 75? of Bujumbura's population, and virtually all of its poor, lived in six self- built and unregulated neighborhoods. While the private home construction sector (mainly informal contractors and day laborers) had kept pace with demand, the public sector lagged behind. Poorer neighborhoods were notori- ously lacking in public services (piped water, sanitation, garbage collec- tion, drainage) and in community facilities, notably schools. Higher dens- ities (up to 340 persons/ha in some places) and low quality services char- acterized the poorer settlements (SAR, para. 1.06). In the irregular set- tlements, conditions were deteriorating due to inadequate maintenance. According to the Appraisal Report, "personnel, equipment, and organization for garbage collection are inadequate to achieve acceptable sanitary condi- tions. Most drainage canals are blocked. Water standpipes are frequently out of service, difficult to use because of eroded foundations and hazard- ous to health because of standing water." (SAR, para. 1.13). 1.04 The situation in the country was delicate, as well as difficult, because of the ethnic situation. "Most of Burundi's urban population is crowded into 'indigenous quarters' geographically segregated from the town proper, and at least a third live in absolute poverty. Virtually no new services have been provided for the urban poor in at least a decade" (BTOR, - 2- April 1978). Similar to other newly independent African countries, Burun- di's urban development appeared to follow a pattern derived from its colo- nial past wherein the enclave of foreign, or wealthy residents of a city assumed the cost of services for the entire population. Logically enough, the Government, continuing the tradition of giving free services to the majority, chose to focus on housing and servicing the well-to-do who pay for the rest. As a result, Government financed high-standard housing for civil servants, while public services were offered free of charge to the rest of the city. 1.05 The history of urban services provision and cost recovery sheds light on Bujumbura's current problems. Local government authorities used to collect FBu 800 from each plot for general revenues, FBu 300 for water and FBu 200 for electricity (about US$6.00). But the utility fees were never passed on to the utility compan,es and, after a change in administra- tions, the Municipality abolished the general tax in 1978. Except for individual connections, all cost recovery ceased. 1.06 Land tenure reflects a similar ex-colonial pat ern. Clear title requires that land be developed according to high sta ard codes. Thus, about 80% of the city is occupied by residents on a provisional basis. At the same time, an estimated 30% of families rent lodging, both in the de- veloped and titled central neighborhoods and in peripheral settlements. The tax on rental incomes has subsequently proved the most lucrative source of municipal revenue. B. Institutional Framework 1.07 At the time of project preparation, housing, land titling, commun- ity facilities and most urban services (water supply, sanitation, electric- ity, road construction and maintenance of principal roads) came under the aegis of various national ministries and agencies: the Ministry of Public Works, Equipment and Housing, the M.nistry of Agriculture and REGIDESO, the national water and electricity authority. In addition, the National Hous- ing Office (ONL) and the Public Housing Corporation (SIP) were involved in high standard shelter construction, chiefly for civil servants. The National Bank for Savings and Investment (CAMOFI) and the National Bank for Economic Development (BNDE), the main financiers of public and private housing projects, focussed mainly on high-income homes. 1.08 Under the Authority of a Mayor appointed by the President, the Municipality of Bujumbura was responsible for security, garbage collection and the maintenance of markets, secondary roads and drainage ditches (SAR, para. 1.20). The Municipality concentrated on collecting what garbage it could with three dump trucks and relied on the sanitation department of REGIDESO to maintain stormwater drains, sewers and septic tanks, while the Ministry of Public Works' Department of Roads and Bridges was responsible for road repair and rebuilding. The PCR notes, however, that "The Munici- pality was unable to carry out its responsibilities in 1978 because of insufficient human and financial resources" (para. 1.03). The SAR differ- ing somewhat, underlined the lack of manpower or management capability, but - 3 - suggested that finances were not a serious problem: "Market taxes, col- lected daily, constitute 40-50Z of municipal revenues, most of the rest coming from property and commercial taxes and fees. The most striking fact about these revenues is their low absolute level--an estimated $450,000 in 1979--and the continued budget surpluses amounting to a quarter of revenues in 1977 and 1978. The main reasons for the surpluses, which a,.- placed in a frozen government savings account, is the city's inability to increase its level of operations rapidly enough to spend the resources it takes in" (SAR, para. 1.30). C. National Policy for Urbanization 1.09 A 1961 decree brought all vacant, unimproved and traditionally- occupied lands under public domain and opened the door for spontaneous occupation of peripheral urban land. But it was not until the mid 1970s that Government began focusing attention on housing and urban development and took steps to define a policy framework and key institutions to carry out planning for urban growth, increasing the supply of serviced land and improving existing neighborhoods (SAR, para. 1.21). Accordingly, the Government of Burundi undertook several projects aimed at expanding and upgrading urban services (including a master water plan supported by the ADF and a sewage and drainage project financed jointly by AFDB, BADEA and GOB). IDA supported various technical assistance programs, among them a study of urban housing policies which contributed to the reorganization of the Public Works Ministry's housing functions. Finally, with the assis- tance of urban plannezs financed through French bilateral aid, a new sketch plan for Bujumbura was developed as the first step in a new planning frame- work for the city. 1.10 Until 1987, the Government of Burundi maintained planning and migra+-ion policies which limited the entrance of artisans and the estab- lishment of informal sector activities in its cities (SAR 2, para. 1.24). These and other constraints, some fomented by the Government, some inherent in the nature of informal activities themselves, led to a rather dormant and predominantly rural informal sector, in contrast to that in other un- derdeveloped countries (SAR 2, para. 1.20), and despite an evident demand for certain services (SAR, para. 2.18). Project appraisal missions identi- fied a lack of technical and financial resources As the principal cause of low productivity among informal sector artisans in Bujumbura. 1.11 During the period 1983-1Q87 (the 4th Development Plan), av;areness of the need for greater administrative decentralization emerged, and the preparation of the 5th development plan (1988-1992) gave rise to a new appreciation of the role and importance of urban growth and the approval of guidelines for a national urban policy. The incorporation of an urban development unit in the Ministry of Public Works (now Ministry of Public Works and Urban Development) symbolizes the GOB's new policy approach. The Bank's second urban development project will support this policy by financ- ing activities in five secondary cities (SAR 2, para. 1.10). II. THE PROJECT A. Project Background 2.01 The Minister of Public Works at the time of project preparation had arrived recently from the UN-Xabitat office in New York and was per- sonally interested in low-cost, high-impact shelter programs. A pilot serviced lot project in the Bujumbura suburb of Musaga, developed with U.N. support, constituted a first attempt to establish a pianned "reception area" for urban immigrants, as an alternative to the existing squatter settlements. In all, the 1970s saw some fifteen different urban develop- ment projects or programs financed with foreign assistance in Bujumbura. Ranging from half a milliou to eleven million US dollars (first stage, sewerage master plan) they drew on resources from the UN, the African Development Bank, BADEA, and German, Chinese, U.S. and French bilateral sources. 2.02 With the exception of a 1966 credit to improve Bujumbura's water supply system, the Bank had previously concentrated its efforts on rural development projects. Indeed, the first urban project sparked certain ob- jections from strategists within the Bank who questioned the diversion of scarce investment resources from rural areas to the urban sector. Nonethe- less, an urban project for Bujumbura seemed timely while the city's growth patterns were considered manageable and could still be subject to rational planning. 2.03 In September 1978, GOB requested a loan of US$10 million to up- grade some 8,000-10,000 existing sites and develop 3,500-4,000 new ones (PCR, para. 2.02). The first Bank mission agreed to the GOB's targets, but proposed to increase the project cost to US$20 million, adding 300 low-cost sewered plots, 10,000 construction loans, schools, clinics, and support for the informal sector (or labor-intensive enterprises). Subsequent missions analyzed the feasibility of a health/nutrition component but dropped it for lack of government support, while adding an energy component. Finally, the project puzported to increase the coverage of Bujumbura's infrastructure and municipal maintenance service. The Audit notes, however, that, despite the substantial quantity of civil works in the project, insufficient engi- neering expertise appears to have characterized the pre-appraisal and appraisal missions, something which may have contributed to problems experienced during execution. 2.04 After presentation of an issues paper in late 1979, overall proj- ect size was reduced to US$16.7 million as a result of the decision to eliminate the health component and reduce the number of serviced sites. After circulation of the SAR, the project wai questioned by Bank staff for the following: (i) Concentration of Resources. "Devoting increasingly scarce re- sources to the growth of Bujumbura (which constitutes 3.5% of total population) would contradict the Government intention of decentralization and regional development." (ii) Increasing Dependency on Government. "Bujumbura's growth depends largely upon the increasing government intervention in the econ- omy, which has resulted in a rapidly swelling wage and salary bill...to 56.22 of ordinary budget." (iii) Procurement Issues. The controller's office recommended disburse- ment against 10OZ of expenditure for imported equipment and tech- nical assistance, on presentation of a statement of expenses. "All disbursements to be fully documented except for: civil works by force account; operating costs and salaries of project imple- menting agencies; and credits to artisans and home builders." Altogether, this excused about US$9.3 million, or 55Z of the proj- ect, from &ccountability requirements. (iv) Informal Sector or Artisan Promotion. The component was consider- ed too small to have significant impact on employment or income generation. (v) Cost Recovery. A specific warning about the difficulties of cost recovery in Burundi was answered by the project officer with a description of "a number of cost recovery provisions that either break new ground...or are common to urban projects but are new to Burundi." (vi) Tariffs. The Loan committee requested an increase in water tariffs and assurance that the city of Bujumbura meet its obliga- tions to pay the REGIDESO, an issue which was postponed and con- tinued (and continues) to beleaguer the Municipality. B. Project Objectives 2.05 The primary objective of the project, as stated in the SAR (para. 2.01), concerned the improvement of living conditions of the urban poor by "(a) redressing the most urgent deficiencies in existing neighborhoods and (b) laying a foundation for more efficient future growth". The SAR under- lin4s the importance of building organizational and managerial capacity to guarantee the maintwnance and improvement of facilities as measures which would "improve the access of low-income families to urban services" (SAR, para. 2.03). At the same time, project preparation documents (BTOR, February 1979) state " the project is best perceived not as a large-scale attempt to rehabilitate slums or eliminate poverty, but as a modest series of advances on a broad front." 2.06 The Audit points out that the project also served to enhance pol- icy dialogue by demonstrating feasible, low-cost and above all, cost-effi- cient approaches to urban development issues. Although unstated, these objectives are clearly implicit in the housing and upgrading components, as well as in the artisan training and support program and the urban services component. -6- C. Project Description 2.07 In accordance with its objectives, the project focused on improv- ing living standards through new housing and neighborhood upgrading pro- grams, on municipal development through the establishment of departments to manage maintenance, cost recovery and community relations, and on income generation through the provision of training and credit to informal sector artisans. The project, as appraised, included the following components, listed in order of th-4r relative share of total costs: (i) upgrading: provision of basic infrastructure to 6 low-income neighborhoods, affecting about 16,700 households and minor works in a seventh neighborhood (20Z); (ii) project management and technical assistance: focused on the devel- opment of a Municipal Housing Bureau, on municipal services'and on the artisan training program (20?); (iii) building loans for a total of about 2,000 home construction and improvement credits (19Z); (iv) municipal services improvement: provision of vehicles and equip- ment for street and building maintenance and for garbage collec- tion (13Z); (v) community facilitiest construction of 3 primary schools and reno- vation of 11 existing schools (12?); (vi) serviced plots: the development of about 780 urbanized plots (7Z); (vii) promotion of artisans: training, credit programs and the develop- ment of an experimental low-energy stove (5Z); and (viii) design, supervisi,n and studies for projece preparation (4%). 2.08 Total project cost was estimated at US$16.7 million of which IDA would provide 90X, including 100Z of the foreign exchange component. The disbursement period was estimated at four years in accordance with the Bank norms prevailing at the time, and 63% of total investment costs were to be recovered. Project design called for greater emphasis on operating costs, civil works by force account, and credit programs, all of which were sub- ject to reimbursement without documentation (SAR, para. 3.11) and together came to over 60? of project cost. These unusual terms for contracting and reimbursement were proposed on the basis that "civil works would be carried out by departmental forces of these units... Based upon local experience, this method would be more economical and expeditious than using private contractors" (SAR, para. 3.08), and by recommendation of the Bank's con- troller's office in a memo responding to the white cover SAR after its circulation and that of the above-mentioned Issues Paper (para. 2.04 (iii)). - 7 - 2.09 Project Management. The project required the creation of four new agencies, all of which were intended to be absorbed into existing institu- tions: the project management unit (to the Ministry of Public Works - Directorate of Coordination of Infrastructure); the housing bureau (to the Municipality); the collection bureau (to the Municipality); the artisan promotion program (to the Ministry of Social Affairs). It also required the reorganization of the municipal service divisions to form a single municipal technical services department. The proposed department was to take charge of garbage collection, maintenance of public buildings and roads, notwithstanding the existing road maintenance department in the Ministry of Public Works and the sanitation department in REGIDESO which managed stormwater drainage, sewers and septic tanks (para. 1.10). III. PROJECT IMPLEMENTATION AND ACHIEVEMENTS A. Time Delays and Cost Overruns 3.01 For the most part, delays in project implementation can be attri- buted to the overly optimistic implementation schedule (para. 5.01). In addition, the factors described in the following paragraphs came into play. 3.02 As the PCR points out (paras. 3.24 and 3.25), the SAR specifically favored construction by force account; "(civil works) would be carried out by departmental forces of these (government) units...Based upon local expe- rience, this method would be more economical and expeditious than using private contractors." (SAR. para. 3.08). As it turned out, implementation by state agencies (mainly REGIDESO and DPE) proved to take longer than expected. By mid 1982, the Ministry of Public Works completed the streets and drainage program for the first neighborhood (but not the largest or most populated). This represented one-sixth of the total component, and completion of serviced lot construction was seen extending into 1986. Only at this point did the project unit recur to a private contractor. 3.03 Coordination with the Municipality's various on-going projects and other donors led to setbacks. In the Buyenzi and Bwiza neighborhoods construction delays in a complementary sanitation project financed through AfDBIBADEA resulted in postponements of the Bank-financed projects to in- stall public standpipes and street-lighting. In the final analyses, up- grading of these neighborhoods was halted because of cost recovery prob- lems. 3.04 The public standpipe program was delayed, and even dropped tempor- arily from the project, for a variety of reasons. Initial problems stem- med from a design which used a siphon to eliminate loss from seepage at the nozzle. The Burundian Government objected, rather understandably, to the notion of each user's having to suck the nozzle to start up water flow. Once a faucet-controlled model had been accepted by all parties, the proj- ect was further delayed because REGIDESO refused to install new standpipes until the municipality repaid its outstanding debt for public water supply, accumulated since 1976. The issue of charging for water at public stand- -8- pipes is delicate, and had come up in earlier discussions following the presentation of the issues paper. The Bank proposed a user-payment system similar to those utilized by other francophone African countries. Although REGIDESO made peace with the Municipality in 1988 and the standpipe compo- nent was reinstated, the cost reccrery problem has never been resolved, and municipal debt to REGIDESO has grown accordingly. 3.05 The Municipal Technical Services Department took longer than anti- cipated to design and implement garbage collection. As no particular col- lection system had been specified at Appraisal, extra time was required to study appropriate equipment and develop collection routes. The situation was further complicated when the project engineer was dismissed and addi- tional time had to be spent to find an adequate replacement. Mid-1982 (the original start-up date) saw the call for bids for supplying equipment. Occupation of the serviced site project at Kwijabe likewise took a year longer than anticipated because the project unit undertook construction of 80Z of the houses. Families building on their own tended to move faster. 3.06 Problems of a personal and political nature affected the project adversely. The serviced lot project at Kwijabe was delayed because, in the words of the first project officer: "the Project Director was not convinced of the soundness of the project concept... a lack of commitment to the concept of serviced sites was the single most important reason why the Minister fired the Project Director in May, 1982." 3.07 Although this project enjoyed government support throughout its implementation, lack of counterpart funding held up construction in the final year (1986-87). In addition, the Municipality of Bujumbura embarked on an austerity drive in 1983, cutting back its contribution to the service departments' operating costs by 25Z, while-reducing all staff by 35Z. The project had already overrun its budget fo operating costs and salaries at the time, and the Bank agreed to designate unallocated funds for project carrying costs and continued technical assistance. A good part of the maintenance program, however, was shelved, orders for garbage collection skips and container trucks were cancelled, and the proposed formation of five maintenance teams (to clean drainage canals, markets and sports fields) was abandoned. 3.08 Major program delays also resulted directly and, in retrospect, unnecessarily from the Bank's strategy for getting the Municipality to apply revenue generating mechanisms (new taxes) by holding up investments in civil works, particularly for the upgrading component, pending applica- tion of the proposed "urban taxR. Upgrading costs were to be recovered through a parcel charge levied on each property, payable over 25 years at 12Z (SAR, para. 5.10), but no serious attempt was made during implementa- tion to apply these charges. Instead, according to the SAR, the Bank tied the upgrading component to the proposed urban tax, which had been designed to cover costs of overrll city maintenance (para. 5.02). 3.09 7-. general, new taxes tend to encounter resistance and the case of Burundi proved no exception, despite prior assurances from the Mayor's office that a simple mayoral decree would suffice. A new Congress formed - 9 - in 1981 with the power of tax review rejected the proposed urban tax, judging that it would place an --nacceptable burden on the city's popula- tion. As the national governmen. already levied property and rental taxes on all urban residents, the congressional argument against another tax for the same purposes made certain sense. Rather than create a new tax, Con- gress voted to pass the existing taxes to the Municipality in November 1983, thereby fulfilling, in effect, the Bank's stated condition for cost recovery. 3.10 Despite the ultimate compliance with the Bank's cost recovery requirements, and the establishment of a successful property tax system (para. 2.29), Bank resistance created further delays and an overall reduc- tion in the proposed civil works components (upgrading and city main- tenance). Towards the end of 1982, Bank supervi.ion missions expressed fear that the upgrading component could not be completed within the ap- praisal timetable or projected cost limits (BTOR, February 1983). Thus, when the Burundian Government requested an extension of the project to permit completion of physical works, the project officer moved instead to cut the uncompleted portion (75Z) of the upgrading component and to de- crease the overall project cost by 17Z, arguing that it made no sense to begin new civil works which could not be completed within the timetable (BTOR, February 1983). Although the project was, in fact, eventually granted two subsequent extensions, the Aide-Memoire of February 1983 posits the Bank as opposing any change in the original timetable. 3.11 The project officer in charge subsequently explained in a memo that it would have been imprudent to carry out the upgrading projects with- out first having tested the new property taxss as cost recovery mechanisms. It should be pointed out that no such "trial period" was foreseen at Appraisal, which only mentions 1985 as the date when major improvements in revenue generation could be expected (SAR, para. 5.08). In fact, municipal revenues tripled ftom 1983 to 19e5 with application of the new property taxes, according to the BTOR of November 1985. At the same time, during Appraisal, the Bank had not adequately considered the time required for registering and assessing properties. Property and rental taxes had previ- ously been applied only in Bujumbura's well-developed center, formerly the home of the European community representing no uore than 1O% of the city's population. Under Bank supervision, the project directors set out to reg- ister and assess properties throughout the rest of the city, amounting to a cadastre of over 15,000 residences. The cadastre, begun in 1983, was only completed during 1986. Thus, even if the urban tax had been accepted at the outset of the project, it could not have been functioning and "proven" operant before the end of the project. 3.12 In 1985, a new project officer for the Bank proposed extending the project to make use of the remaining funds (US$8.4 million, of which US$2.9 million had been tentatively cut from the project). The project was re- vised to emphasize community facilities (schools and markets) and a new serviced lot program to the north of the city (Gasenyi). When cost esti- mates came in, Gasenyi was deferred to the second urban project, and the first operation was closed. - 10 - 3.13 Costs ran high in the first three years, coming in from 152 to 145Z over appraisal estimates (dollar equivalents). School costs were initially estimated to be higher than the standard square meter cost, ac- cording to other Bank education projects, although actual construction costs seem to be in line with other Bank-financed projects. The greatest cost overruns, however, were in the categories of project management and studies, components which were bound to increase with the extension of th. project. The Appraisal Report foresaw covering expenses of project staff for two years. During implementation the limit was extended, apparently through project completion. B. Project Changes 3.14 In the final analysis, the project shifted from an urban develop- ment effort based on civil works, to an education project, with a large technical assistance component. Annex 1 delineates the final reprogramming, described in general terms below: (i) The upgrading program was reduced by a total of 75Z. Of the six neighborhoods originally scheduled to be rehabilitated, only one (Cibitoke) was completed as planned. In justification for the elimination of the upgrading component, the BTOR of July 1983 states: "Since the remaining neighborhoods will continue to bene- fit from street maintenance and other municipal services, this reduction (in the upgrading component) will not compromise the objective of improving living conditions for the major part of the population of Bujumbura." In fact, the maintenance component was cut back at the same time, affecting precisely the neighborhoods which formed the original target area of the project. (ii) The municipal services component was cut back by 40Z; three dump trucks, 6 maintenance vehicles, tools and equipment for building and drainage maintenance were eliminated. As a further repercus- sion, the projected road and building maintenance teams were elim- inated. The results can be seen today in the generally poor state of roads and drains in Bujumbura. (iii) The community facilities component was more than doubled, repre- senting about 30Z of the total, aR opposed to the 12% projected at Appraisal. The difference is explained by the construction or rehabilitation of additional schools, markets, an artisans train- ing center and facilities for the project unit. (iv) The technical assistance and project management component in- creased along with that of design, supervision and studies, by about 40Z. (v) The program to promote artisans shifted its emphasis from credit to training. - 11 - C. Project Achievements 1. Physical Works 3.15 The settlement of Kwijabe, with 778 serviced plots and minimum dwellings, was completed and turned over to beneficiaries within the pro- jected cost range even though program completion lagged two years behind schedule. As the photographs in Annex 2 show, it is a well-used and agree- able neighborhood and, judging from the improvements the residents have m&de to their homes, it is considered a good investment. Kwijabe ran into some physical problems which may be attributable to the original feasibil- ity study. As the first project officer states in a memo responding to the PCR: "The performance of the consulting firm which carried out the feasi- bility studies was uneven... Although this firm has gone on to acquire a good reputation in international work on urban development, this initial effort must be regarded as a learning experience--for both the Bank and the firm." (i) The settlement suffers from spot flooding due to inadequate drain- age, which, in turn appears to result from: (a) a faulty soils tests during project design; (b) overall site problems or grading; and/or (c) poor dimensioning of main street drains and the total absence of interlot drains. The problem has been compounded by the fact that the house design did not reinforce foundations and subsequent erosion is eating away the bases of many Kwijabe homes. (ii) Once settled, families protested the lack of secondary roadways in the development. The project unit (through SETEMU) opened, graded and provided drainage for a a series of secondary roadways in response to the residents' complaints. It did not charge costs to the residents. (iii) Septic tanks in Kwijabe were designed with a total capacity of lesa than 5 cubic meters and have filled up in the course of the first two years. The traditional septic tank in Bujumbura aver- ages 20 to 30 cubic meters in size. 82Z of Kwijabe's families were obliged to dig new septic tanks within their first year of occupation. (iv) The siphon-drawn standpipes were first installed in Kwijabe (para. 2.13 above). The Kwijabe residents' rejection of the model led to reconsiderations of its overall use. (v) In an attempt to support the use of local materials, it was pro- posed that the project build adobe houses for Kwijabe residents. - 12 - A few models were constructed, but they have not proven weather- resistant and the project has had to pay residents the cost of repairs or replacement (see photos, Annex 4). 2. Upgrading 3.16 Of the upgrading projects foreseen at Appraisal, Cibitoke's pri- mary roads were straightened, widened, graded and fitted with stormwater drains and public street lighting. In addition, a storm water drainage main was constructed in Kinama. Street drainage and electrification in Bwiza were completed during the project implementation period, but were financed through other sour.-s. Finally, sixty four public standpipes were installed throughout the peripheral neighborioods of Bujumbura. The up- graded programs not undertaken included the complete storm-water drainage plan for the peripheral areas (Kinama, Kamenge, and Nyakabiga) road im- provement in Kinama, Kamenge, Nyakabiga, Bwiza and Buyenzi and street lighting in Kamenge, Kinama, Bwiza and Ngagara (para. 4.06). 3. Development of Urban Serice Capability 3.17 The development of the SETEMU is described in the PCR (paras. 4.01-4.07). Insofar as the municipal services component was not planned in terms of the city's overall needs, and was, furthermore, substantially reduced in the course of project implementation, SETEMU functions at less than full-service coverage. At present SETEMU offers the following services: (a) septic tank clean-outs, charged to customers, a total of 1,300-1,500 m3/year, compared with an estimated demand for 5,000 m3/year; (b) garbage collection, 19,500 tons/year, which represents 58% of the city's annual production; (c) roar maintenance, regrading of 150 kms/year and resurfacing with gravel of about 40 kms/year out of a total of 250 kms. of unpaved roads in Bujumbura, and repairs of about 30% of the holes in paved roads/year; and (d) stormwater drain maintenance, cleaning out about 134 kms. of drainage ditches/year out of a total of 250 kms. of roads (500 kms. of drainage ditches). The services listed in items (b) through (d) are presumably financed through the property taxes charged by the Municipality. 4. Home Construction and Home Improvement Loans Program 3.18 The home construction loans program, financed 85% of the lot pur- chasers at Kwijabe. Two types of loans were offered: the "self-help", which accounted for 22? of loans, provided cash on the line, and allowed - 13 - families to hire their own builders and manage their own construction and, the "tacheron", which covered the vast majority of beneficiaries and meant that the project directors managed the construction for the residents at no extra cost. Interviews with both project directors and residents reveal a unanimous preference for the "self-help" scheme. Under the "tacheron" plan, the residents accuse project directors of inadequate construction management and the directors accuse residents of chronic complaining. At the same time, both groups appear to have underestimated construction costs which explains why the project gave out supplementary loans and the Govern- ment of Burundi subsidized the total loan program for a total of 20 million Burundian Francs (approximately US$170.000 or US$264/loan). The total amount loaned for home construction thus increased 75% over the estimate at Appraical, and the average loan amount increased accordingly. The home construction loans program, however, boasts a laudable repayment rate, with less than 3Z in arrears. 3.19 At Appraisal, 1,080 home improvement loans were foreseen for resi- dents in various peripheral neighborhoods of Bujumbura (i.e., those subject to upgrading programs under the project). Credit was to be offered in small amounts (US$300 to US$1,735 equivalent, at an average of US$990) to help families make essential improvements. The program got off to a slow start, generating demand for less than 500 loans within the first five years of the project. The project unit thus increased the amounts offered to a minimum of US$1,000, maximum of US$1,833 and average of US$1500. Demcnd increased rapidly such that the waiting list for credit was 2,500 by the end of the project. A total of 535 loans were made in all. Cost re- covery is not as successful for the home improvement loans as for home construction credits despite the home mortgages used in guarantee. Offi- cials at ECOSAT (para. 2.37 (iii)) relate that borrowers often feel little compunction to repay as they doubt that the State will move to evict them. Late payments were running at about 151 in July 1988, when ECOSAT took over cost recovery from the municipality. By the end of the year, they had increased to about 30%. ECOSAT intends to take strong measures to recover its portfolio during 1989. 5. Resource Mobilization and Cost Recovery 3.20 Although the urban tax was never applied exactly as proposed, the project can take credit for the establishment of an effective property tax system. Under project implementation the municipal tax base has indeed increased from 14% of the properties (representing only 41 of the total population) to 90? of the population and properties. The registering and assessment of approximately 15,000 additional households, no mean feat in itself, was carried out through the distribution of provisional occupancy titles (TOPs, PCR 4.07). Bank supervision missions appear to have offered significant guidance, as well as bringing strong pressures to bear (para. 2.17) in the area of municipal revenues, but this component also benefitted from an enthusiastic and capable local official who currently heads the Municipality's financial services. - 14 - 3.21 As reported in the PCR (para 4.10), the introduction of the two property taxes, one on rental income and the other on assessed value, in- creased municipal income by about 1302 in 1984. The two taxes accounted for -bout 50Z of Bujumbura's revenues over the following two years, falling to 45% by 1988. The decline in the property taxes' importance can be traced to three factorst the amount of the tax is fixed and does not in- crease with inflation like other municipal taxes; no new cadastres have been undertaken since 1983, and; the municipal tax collector's office con- siders the marginal gains from a new cadastre and assessment too low to justify their costs, although plans are underway for a second property tax campaign in 1990. Furthermore, while the actual operating costs of the property tax collection division are extremely low (estimated at 30,000 Fr.BUlmonth, or about 0.23Z of local revenues), the Municipality's finan- cial situation is tight and increased expenditures are difficult. 3.22 According to the SAR, revenues from the urban tax were to cover recurrent municipal service costs, public water consumption, investments in off-site infrastructure and in equipment, vehicles and materials for the technical service component. (SAR 3.04, 5.02, 5.03 and 5.06). Investments were to repaid through a loan from the national government to the Munici- pality of Bujumbura, water consumption was to be repaid to REGIDESO and the recurrent costs of technical services would be managed internally. In fact, the revenues generated from the property taxes have not covered the proposed categories, largely because of SETEMU's high operating expenses. At the same time, the Municipality's contributions to SETEMU do not, in fact, cover SETEMU's recurrent costs (let alone its investments) which receive a substantial amount of assistance through external financing. 3.23 Bank supervision missions stressed the idea that revenues from the urban tax should finance the investment costs of neighborhood upgrading projects, as well as maintenance, even though the SAR foresaw upgrading costs to be recovered "directly from beneficiaries' (SAR, para. 5.02). In fact, the annual assessed tax value from Cibitoke (FBu 6.828.575) corre- sponds closely to the debt service charges for upgrading (FBu. 5.923.700). But it is clear that if property taxes are to cover upgrading costs, then they may be insufficient to cover water consumption and waintenance costs as well. At the same time, the property tax assessments for other neigh- borhoods would justify upgrading projects there. In other words, if the parcel-charge system was abandoned in favor of a recovery system through property taxes, then five neighborhoods are currently paying for upgrading which never took place, while Cibitoke is actually paying back the costs of its own improvement. 6. Artisan Support Program 3.24 The artisan support prograr is well described in the PCR (para. 3.12). In the end, the project emphasized training rather than credits. Of the 360 small business loans anticipated, only 180 were made for a total amounting to about 18X of the appraisal estimate. A training center, which actually functions as a working cooperative, was built in Kwijabe. Twenty- seven carpenters are affiliated while others (averaging two per week) pur- - 15 - chase prepared wood from the center, but the use of the machines is limited to the affiliates and their employees. A total of six seamstresses/tailors have completed the three-year training course ard work at the center. More are paying off sewing machines, which are stored at the center. In the meantime, they prefer to work at home, closer to their clients. The sol- dering and metal work program has about ten affiliates, although they do not work steadily, but only according to demand. Training is given exclu- sively in the three areas mentioned, and emphasizes technical skills rather than business methods. In addition to the artisans' training programs offered by the project, the Municipality runs full-time adult education centers in two of the elementary schools built by the project (para. 2.36). No information is available as to the number of artisans benefitting from these training activities, but they appear to cover a wider range of skills, including leather work, shoemaking and mechanics. 3.25 Energy Component. As the PCR reports, (paras. 3.14 and 3.15) the component was successful in developing a low-energy, long-lasting (one year) stove for widespread domestic use. Although a total of about 12,000 stoves have been sold, production costs are still too high to compete with those of traditional stoves, which consume about one and a half times as much fuel, and must be replaced every three to four months. The project financed a workshop, where an average of ten artisans working at piece wages make the low-energy stoves, three sales-kiosks, a team of promoters who work door-to-door, and technical assistance from five different consul- tants, including preparation of a marketing study. According to the proj- ect unit and ONATUR, the sales price of the stoves only covers the arti- sans' labor costs, and no separate accounts have been kept to determine the total cost of the effort, nor the approximate benefits in terms of fuel saved. 3.26 Local Materials. Although not identified as a project component at Appraisal, the theme of local materials use was introduced during imple- mentation. Home construction in adobe and hand-formed clay tile roofs was encouraged in Kwijabe (para. 2.24 (v)) and in the project offices and SETEMU buildings. Outsie technical assistance was not utilized in either case, and would have serm.4 a useful purpose only had it been able to dis- courage the project from the outset. The clay-tile roofs added consider- ably to construction costs, while the overuse of wood in the substructure required to support their extra weight ostensibly defeated the conservation efforts of the energy subcomponent. The roofs on the Project Unit and the SETEMU buildings leaked and were replaced with lightweight asbestos cement panels within three years. 7. Community Facilities 3.27 The Municipal Department of Primary Education considers the proj- ect to be responsible for the construction of ten new schools (a total of 59 classrooms) and for repairs in fifteen schools. Apparently four of the existing schools, so deteriorated as to be out of use, were demolished and rebuilt under the category of repaired schools. Two of the newly- con- structed schools are used for adult education programs on a full-time ba- sis. The schools are judged by the Bank's Education Division to be "more - 16 - luxurious than the norm", but their square meter costs compare favorably with the official construction price of BPE (the state school construction enterprise). According to the project director, the schools responded to an urgent need in Bujumbura and, at the time (1985), the Bank was not sup- porting the construction of urban schools under its education project. Even though the education needs for Bujumbura were not formally assessed, the Bank's education officer for Burundi has confirmed that additional schools are always necessary and that the previous education loan did not include construction of urban schools. 8. Institutional Development 3.28 According to the Appraisal Report the various agencies established under the project were to be absorbed by the Municipality. The PCR reports quite accurately on each agency. The following list offers some additional observations about the various instititions created or strengthened under the project. (i) SETEMU has produced a certain amount of controversy, being an unplanned product of the project (PCR, paras. 4.02-4.06). Its institutional purpose is confusing as it is supported financially by the city of Bujumbura, but takes orders from the national gov- ernment. As cited in the PCR, SETEMU has dedicated itself to several national construction projects instead of municipal main- tenance, and has left the city of Bujumbura with the bill. The institution has recently adopted a policy of not investing in new projects, but of limiting itself to maintenance work. According- ly, the municipal budget for SETEMU fell to FBu 165 million in 1988, down from about FBu 500 million in 1987. If its budget remains on this order, SETEMU's recurrent expenses could be covered by the revenues from municipal property taxea and could conceivably increase to offer a wider coverage. At this point, the institution's greatest challenge involves finding ways to expand services while lowering its expenses. SETEMU technicians have explained that, given their limited equipment and the nature of the international assistance programs, it is often more advan- tageous to undertake new construction projects which are financed through donations (and which cost less to build new in cases such as paved roads), than to maintain the existing ones. (ii) The Collection Bureau established under the project management unit has been successfully incorporated into the Municipality as the City's Collection Department and Office of Financial Services. (iii) ECOSAT. Formerly the Urban Project Housing Bureau, ECOSAT was established in July 1988 under the Ministry of Public Works' Directorate of Urbanism and Housing. It exists as a counterpart to the ONL and SIP, but focuses on housing for low-income fam- ilies. ECOSAT now manages mortgage loan recovery from Kwijabe, from the project's home improvement loans and from the Musaga serviced lot program, developed with United Nations support at the - 17 - same time as the Bank's project. ECOSAT is currently initiating a new serviced lot project for .100 families in Bujumbura (Kanyosha) to be financed by the fund created from recovery of Bank project costs, by contributions from the Municipality and from the U.N. (FENU) and will also develop about 1,070 lots in Gasenyi (includ- ing some 620 new sites and services plots and the upgrading of 450 existing lots), under the Bank's second urban project in Burundi. Beginning 1990, it will offer home improvement loans to families in poorer neighborhoods and, will develop a secondary cities pro- gram in conjunction with the second Bank loan. (iv) Although not foreseen at Appraisal, the project management unit, known as "DUB" has become the prime mover of the artisan support program. It manages the workshop where the low-energy stoves are produced (sales and promotion are directed from ONATUR) inside its headquarters and continues to play a role in the development of artisans in the second urban project. IV. POINTS OF INTEREST A. Physical Impact 4.01 The project's physical impact can be seen in the new development in Kwijabe, in the upgraded areas of Cibttoke and in the installation of public standpipes and in increased municipal maintenance. Approximately 3,128 families, or 152 of Bujumbura's total population, were directly affected by the first two components. The second two are harder to quant- ify. Although municipal services are undoubtedly far better than before the project, the maintenance program cannot pretend to keep up with the city's problems. Garbage production has grown along with the population at 8? per year. The roads in Bujumbura and the number of cars which use and wear them down have also multiplied, while SETEMU's technical capacity remains fixed. Insofar as its activities are overshadowed by the growing needs for its services, SETEMU attracts more attention for its failures than for its achievements. 4.02 The upgrading project in Cibitoke has had a visible impact on subsequent development in that neighborhood. The widening of a main street has spurred private investments and improvements by homeowners, as well as a marked, but unquantified, generation of commercial and production activi- ties. The market place which was upgraded under the project has more than doubled in size following the intervention in spite of the lack of drainage or of sheltering elements such as solid roofs and slab floors. 4.03 Despite drainage and sanitary problems, [wijabe, housing some 780 families, is a notably pleasant neighborhood. It is considered a model in Bujumbura and one of the city's most desirable areas. Kwijabe's market clearly extends its influence well beyond that community, and has doubtless generated significant economic activity. - 18 - 4.04 The public standpipe program increased the number of standpipes from 116 to 180 throughout the city. Of an estimated 12.170 families (about 67% of Bujumbura's population) who depend on public standpipes, approximately 80Z now have access to water within 100 meters of their resi- dences. At the same time, the total water consumed through public stand- pipes represents less than 92 of Bujumbura's production. Thus city-wide water consumption was increased by approximately 3Z, through the project, while the number of families served has jumped by 552. The program's effect on public health has not been measured, but is assumed to be positive. 4.05 The extension of public street lighting into Cibitoke has made domestic electric connections possible in that neighborhood. REGIDESO reports that a significant number of households have requested, and are receiving, individual connections as a result. Since many small-businesses utilize the home, productive activities tend to increase with household electrical hook-ups. 4.06 On the negative side, the reduction in the upgrading component had some unfortunate side effects. Aside from the drainage problems in Kwijabe (para. 2.25), serious floods have developed in certain neighborhoods sur- rounding Cibitoke, apparently the result of cut-backs in the upgrading program. According to the technical directors of SETEMU, key primary rain- water collectors were eliminated from the program when the upgrading compo- nent was curtailed. Drainage ditches from Cibitoke consequently discharge into undersized, incomplete, or nonexistent street drains and have caused flooding in the adjacent neighborhoods. The problem has become particular- ly acute with this year's record rainfall. As SETEHU's capacity does not permit it to cover all the city's needs several of the infrastructure works undertaken by the project are already falling into disrepair, notably, storm water drains and streets in Cibitoce. 4.07 According to the SAR, the project would result in a more efficient pattern of land use in Bujunbura: "The project is expected to improve work-residence linkages by helping to fill in this open space (low-density Bujumbura) and in the process would raise the average city residential density by up to 10Z I It must be observed that, aside from the creation of Kwijabe, within the city's central core, the project has had little impact on overall urban growth patterns. The present situation follows the same pattern of extremely low-density in the original center, high density in the "indigenous" neighborhoods which ring the central areas, and medium densities in the emerging settlements in peripheral areas. B. Institutional Impact 4.08 The project was designed to reinforce local government by creating and/or strengthening several of its departments, which would, in turn, lead to a greater operational capacity on the part of the Municipality. The SAR pointed to its yearly budget surplus as indicative of "the city's inability to increase the level of operations rapidly enough to spend the resources it takes in." One of the project's most important achievements in the area - 19 - of municipal strengthening comes with the creation of the Municipal Collec- tion Bureau, the computerization of its financial affairs and the genera- tion of important new revenues. According to the Second Urban Project SAR "the project was able to put Bujumbura's finances on a sounder footing." (para. 1.28). As for operating capacity, several new agencies came into being which function within the metropolitan area of Bujumbura, but are not responsible to the Municipality. 4.09 The creation of SETEMU, (para. 2.37) actually reinforces the Ministry of Public Works, ECOSAT and DUB, the Directory of Housing and Urban Development. The fact that these institutions now function at a national level has facilitated the preparation of the second Bank project to be carried out in intermediate cities (para 4.27). Thus, while SETEMU was created specifically as a technical services agency for Bujumbura, it, nonetheless, expects to become active in the Bank's second urban project since it possesses most of the country's experience with the provision of urban services. 4.10 Although not contemplated at Appraisal, an important institutional consequence of the project has occurred at the level of the Superintendent of Schools in Bujumbura. Not only does this office enjoy newly renovated headquarters in one of Bujumbura's 44 schools, but it can also claim to direct a program which offers coverage to 10O% of Bujumbura's children of elementary school age. C. Policy Impact 4.11 The broadening of the Government's attitude toward urban develop- ment in general, and toward the informal sector and informal settlements in particular, represents an important area of policy development, whether due to this project or to the force of circumstances. The informal sector, seen, not so long ago, with more suspicion than enthusiasm, is now consid- ered a worthy object of government support, help and encouragement. It will be the focus of a major component in the second urban project and has already been a matter of several independent projects. Similarly, while government policy is not yet directed specifically toward the marginal or informal settlements, it has currently adopted a policy of tolerance, of granting occupancy rights and of extending services to settlements which used to be barred from the city limits. 4.12 The acceptance and city-wide application of the two property taxes likewise (para. 2.29) represent an important shift in policy. It should be recalled that prior to the Bank project a small minority of citizens supplied most of the city's revenues and a universal tax had been abol- ished for political reasons. Cost recovery for specific services remains uncertain, SETEMU has no difficulty collecting payments for cleaning out individual septic tanks, but it will be a long time before the Municipality charges for public water supply rejects its home subsidizing programs or charges full cost at Kwijabe (despite the commitment to do so in the Loan Agreement). - 20 - 4.13 The project has also seen an important new emphasis on maintenance in municipal services. SETEMU's recent decision to refrain from any new construction project represents an important policy stance. While it may come late and be insufficient, the recognition of the importance of mainte- nance services should not be overlooked. 4.14 As noted earlier, the Municipality's financial and collection services were considerably strengthened by the project, while its operating capacity did not develop at the same rate. In fact, the Municipality turned down the artisan (informal sector) support program, as well as the housing bureau (now ECOSAT), before these were passed to the Ministry of Public Works. While decentralization has become a more accepted general policy, the city itself appears to be cautious about accepting new respons- ibilities, a reaction perhaps, to its extremely rapid growth over the past six years, under the project. 4.15 Government housing policy has changed dramatically during the past six years. The Bank and the project have definitely helped to contribute to the notion that government should support low-income families, while stressing the development of serviced "reception areas" and minimum and basic dwelling units, as opposed to the high-standard housing for civil servants which comprised the only previous efforts in the housing field. At the same time, the Directorate of Housing and Urban Development has not been able to develop policies which meet the city's actual development needs, to admit the important reality of the spontaneous settlement process and to take their rapid growth rate into account. Rather, policies remain focused on serviced lot projects, following the Kwijabe model, for a limit- ed number of families of relatively high-income levels. Although Kwijabe was originally intended as a mixed-income neighborhood, current policy statements call for higher levels of services in government-supported hous- ing projects. It is now felt that Kwijabe taught an important lesson, that the poorest families should not define the upper limit of a project, but that standards should be set according to what better-off families can afford. At the same time, the Government does not seem comfortable with full cost recovery for what it considers to be social housing programs. The loan subsidies and the extra upgrading work at Kwijabe are an example. D. Poverty Impact 4.16 The Audit cannot support the statement in the SAR of the second Burundi urban project that: "Overall an important part of Bujumbura's population and virtually all households below the poverty threshold experi- enced an improvement in living standards [as a result of the first urban project]" (para. 1.27). While some very important advances were achieved, particularly in the areas of municipal strengthening and education, the project's direct poverty impact was minor. Urban civil works, i.e., hous- ing, upgrading and urban services, represented 182 of total project costs, but only the last has had a widespread, if weak, effect on the city's lower income groups. - 21 - 4.17 The serviced lots in Kwijabe represented only about 5Z of the total project cost. The beneficiary families were selected by lot from a group of 1,358 qualified applicants, who had been filtered from a larger group of 1,402 who fulfilled the following conditions: having family in- comes FBu 6.800 and 27,000 (which put them between the 15th and 65th per- centiles), having resided in Bujumbura for at least two years, having no other property and being head of a family. The project was designed to provide for a range of income groups, to avoid a ghetto-stigma. Similarly it was supposed to be open to nationals and non-nationals alike. At the first drawing, 202 of the beneficiaries were foreign residents. Shortly thereafter, however, the government decided that Kwijabe should be open only to Burundians and a second drawing was held to replace the non-nation- als and another 33 of the original group who were discovered not to qualify for one reason or another. 4.18 Despite the selection process, a participant observer study re- vealed that 25Z of the families at Kwijabe had incomes over 27,000 FBu, and another 20% had rented out their homes within the first year. The terms of reference for the study did not include information on income, so a full distribution was not given, but it was found that 702 of household heads are regularly salaried employees. Thus it appears that Kwijabe does not house Bujumbura's poorest residents or even a representative sample. Con- sidering its relatively high service levels, its central location and num- erous access roads, it should not be surprising that better-off families should move in. Given the basic premise, however, that 75% of the city's population is below the poverty level, the project had little cause to be particularly selective. In any case, Kwijabe does not represent a high- impact component, insofar as it is concentrated on less than 4% of the city's families. 4.19 Cibitoke, where the upgrading component was carried out, has 2,350 families, or about 112 of the city's population. The upgrading component directly benefitted the families on the main street with electric lighting (about 92 of the area's population) and those on the secondary streets with drainage (about 272 of the local population) and street improvement. Given the recent torrential rains, the poor maintenance and the incomplete drain- age system, most of the streets and drains are in poor shape making the benefits of upgrading hard to detect. Once again, the impact was not wide- spread, nor was it particularly strong. 4.20 Urban services, albeit incomplete and/or infrequent, can be appre- ciated city-wide. The nunicipal service component came to less than 82 of project costs, but its impact is indeed visible, insofar as it has induced people to throw garbage in a single place (even though it might stay there for a week or so), kept roads open for public transportation (even if pas- sage is not always comfortable) and maintained the flows in key drains at times of heavy rains. 4.21 The artisan support program reached only 180 out of an estimated 3,000 craftsmen in the city. Follow-up of beneficiaries of this program revealed that craftsmen's incomes ranged from 16,000 FBu/month (US$100) to - 22 - about 20,000 FBu/month (US$126), but no research was done to determine how craftsmens' incomes changed, if they did. Insofar as it concentrated ex- clusively on carpenters, metalworkers and tailors/seamstresses, the pro- gram's effect could not have been very widespread, and its cost/benefit ratio does not appear very favorable. The total program, including cred- its, training and the training center, came to about US$4,000/artisan. The low-energy stove project, which currently employs ten craftsmen as occa- sional manufacturers of DUB 10 stoves, cannot be said to have had any sig- nificant poverty impact. On the other hand, the Borrower points out that many people have directly benefitted from the service provided by the a.isans supported through the program, while a significant number of artisans have made use of the supply facilities offered by the Artisan Promotion Center. 4.22 The project's indirect effects, through the new markets at Kwijabe and at Cibitoke, through the various construction projects generated as a result of home-building and home improvement loans have probably had a greater impact on income and jobs than the component specifically designed for this purpose. Unfortunately, the project's indirect impact was not evaluated, and its effect on poverty can only be guessed at. The school construction program probably benefitted a wide range of the population. Since this component was not subject to appraisal, nor to a complete ex- post evaluation, its impact on the low-income population of Bujumbura can- not be assessed. E. Other Points of Interest 4.23 Impact on Local Contractors. The reliance on state agencies for the civil works components not only prolonged the project, but also had a negative effect on local businesses. When government contracts dwindled, the number of independent contractors in Bujumbura also fell, from 39 to 9. While it is doubtful that this project alone could, or should, have saved the contracting businesses, it is ironic that the Bank should have been financing a project to support the private construction industry at the same time (Credit 1230-BU). 4.24 Economic Reevaluation. The PCR includes a reasonable reevaluation of costs and benefits of the Kwijabe serviced site program. Kwijabe how- ever, represented less than 5Z of the total project cost. A complete eco- nomic evaluation should compare the rate of returns from the investment in schools and technical assistance to that of urban upgrading and maintenance services which it replaced. It should also compare the cost and the like- lihood of obtaining future financing for schools versus urban upgrading and services. At the same time, it would have to consider the revenues gener- ated by the new property taxes in terms of the costs involved in cutting back the upgrading and service components, i.e., the strategy used to achieve the new tax policy. 4.25 Procurement Issues. The PCR accurately describes certain irregu- larities which occurred in procurement (PCR, para. 3.26) and the misuse of project funds by certain employees (PCR, para. 4.08). A series of letters - 23 - between the Bank Division Chief and the Minister of Public Works during the second half of the project dealt with another instance of a temporary dis- appearance of funds, this time an account that was supposed to be estab- lished within the Municipality of Bujumbura. The Audit suggests that such abuses were made easier by the project terms which left the bulk of the project expenses to the disposition of the project unit, subject to reim- bursements based on statement of expenses, with no supporting documentation (para. 2.04 (iii)). 4.26 Sustainability and Replicability. This project has proved more successful than most in applying a mechanism to recover costs, to repay intermediate loans (Bujumbura to the national government) and to continue investment in urban development. However, it is unclear which of the proj- ect's components are, in fact, sustainable. The following list analyses each component: (i) Cost recovery from home finance credits (new housing, home im- provement) does not cover administrative costs, nor costs of late payment, which are increasing. Thus, replicability is not cer- tain. (ii) The new housing programs at Kwijabe were subsidized by the Burun- dian Government both up-front and in terms of additional infra- structure for which there was no cost recovery. (iii) The loan terms to artisans, at 8%, are below the commercial inter- est rates and are not sustainable for this reason. In addition late payment and default appear to pose serious problems for replicability. (iv) The costs of urban upgrading were not recovered by parcel charges, as specified in the SAR. However, the amount collected from prop- erty taxes in Cibitoke is adequate to pay off the debt for works done in that part of the city. This implies, however, the elimi- nation of other components which should also be financed through the property taxes (according to the SAR), such as maintenance services, public water supply and the cost of investment in equip- ment for municipal services. (v) In principal, the city-wide revenues from the property taxes can cover the costs of investment in municipal service equipment (pay- ing back the Central Government's loan), public water supply and maintenance services (SETEMU), as long as SETEMU does not charge for investment projects. However, SETEMU's operational costs reflect a partial service coverage. If SETEMU were to increase its operational capability to offer complete urban services, it is not clear that the revenues from the property taxes would suffice. 4.27 The Second Project. Preparation of a second urban project was included as a component in the first operation. The second project will emphasize housing, upgrading of infrastructure, community facilities, and - 24 - artisan promotion in Bujumbura and five secondary urban centers. The main- tenance and energy components have been dropped from the formula (the form- er giving rise to speculation by SETEMU as to who is going to maintain the new infrastructure in secondary towns). The establishment of a system of generalized property taxes to be established and applied in each of the secondary towns to pay for costs of investment in infrastructure appears to be a direct lesson taken from the first urban project. Another lesson emerges in the destination of US$5 million, or 20Z of total project cost, for undefined uses, to be allocated in the course of project imple- mentation. 4.28 Technical Assistance. The PCR's assessment of the technical assistance component seems accurate. It should be mentioned, however, that the original feasibility study appeared to confuse certain issues, and can be blamed for certain failures in physical design. V. THE ROLE OF THE BANK 5.01 The Bank's influence is clear in the project's size and composi- tion. It appears, however, that the Bank may have proposed and supported the idea of a project larger than the Government of Burundi could readily har'le, as the following, from a 1979 Back to Office Report, suggests: "The Government proposed a $12 million project, and that is probably about the limit of its investment capacity over a 3 to 4 year period. The mis- sion identified a larger program.. .tentatively estimated at $20 million.' At the same time, the Bank limited the scope of the project by restricting its activities to Bujumbura (rather than also including Gitega), dropping the serviced site of Gasenyi at Appraisal, and not including a health com- ponent because full agreement could not be reached with the Health Ministry on its composition. 5.02 The Bank, nevertheless, played an important role by introducing, for better or worse, a wide range of programs: a local materials component subsequently rejected by the Burundians; the artisan support program, con- tinued in the second project and copied at a local level by the Municipal- ity of Bujumbura, and; the revenue mobilization effort, which proved quite successful in the end, but was not repeated in the second project. The Governnent, it should be recalled, requested a simple housing and urban upgrading program. The Bank's role in project preparation included a long diagnostic process to determine what other areas of urban development could be addressed in Bujumbura. 5.03 The Bank must accept partial responsibility for the problems which affected the civil works components. Given the fact that over half of the project's resources were directed toward construction and civil or sanitary works, greater participation of a civil or sanitary engineer would have been highly recommendable at Appraisal, or during the first five years of project development, and particularly when the cutbacks were discussed. Supervision missions during these years averaged a total of 70 person- days/year, compared with 9.5 person-days/year during the last four years, and included many consultants, but emphasized sociological, institutional and design, as opposed to engineering, aspects. - 25 - 5.04 This project offers an interesting example of the importance of the project officer in setting the tone for the role of the Bank during supervision. The professional backgrounds of the two project officers clearly marked the project's evolution. During the first five years, guid- ed by an urban economist, the project emphasized cost recovery and finan- cial development, even at the expense of physical works (para 2.19). While this financial strategy worked, the cutback of programmed civil works ap- pears to have had a negative environmental impact (para. 4.06), and may not have been necessary. The first five years are also characterized by a certain optimism in supervisiou reports; the project is labelled "problem- free" in all but one BTOR. 5.05 In 1985, a new project officer took charge, this time an engineer. The focus shifted back to civil works and by the fastest possible means. In order to make use of the remaining funds, the school construction compo- nent was increased, an artisan training center constructed, a market built, the public standpipe program resurrected, and a new serviced site effort, as well as the second urban project, were designed. In the first case, the high priority placed on cost recovery would have determined a loan consist- ing almost exclusively of technical assistance to develop a system to pay back the loan, putting a cost recovery mechanism in place with no costs to recover except that of its own installation. In the second, the execution of the project became more important than its replicability. In the first instance, the Bank can be seen pushing for the establishment of a municipal property tax system and in the second, pushing to get the project done. Indeed, the rate of execution virtually doubled in the final two years. 5.06 Bank personalities played another key role in project evolution. The first project officer acted to curtail disbursements when faced with a deadline, rather than to extend the project. The second project officer took a diffe.ent approach, prolonging the project by two years to spend the remaining funds, even though the investment program changed significantly in the interim (see Annex 1). The question raised is interesting; how far should the Bank go in extending project deadlines? In this case, given the precedents in project preparation the question is more complicated: should the Bank overload a country's investment capacity if it is not willing to grant an extension to complete the investment program? In retrospect, the Audit finds that the original execution period should have been extended, because subsequent events have demonstrated that Burundi has moved to establish viable cost recovery mechanisms, which could have been applied and strengthened through the upgrading and maintenance programs. In gener- al, cost recovery instruments, as well as institutional development, re- quire more time than the early Bank urban projects allowed. VI. LESSONS LEARNED A. Lessons for Physical Works 6.01 The results of the project suggest the following lessons relating to the physical works components, upgrading, maintenance services and serviced lots: - 26 - (i) Engineers from relevant fields should participate in appraisal and supervision of projects with a high percentage of physical works. Timely consultations with experts versed in soils, engineering, drainage, Aivil construction and sanitation might have saved the project from certain negative results. (ii) Drainage projects, like maintenance programs should be developed for total, rather than partial, coverage. As this project demon- strates, it may be preferable not to build drainage canals at all than to build them without providing adequate collectors. Simi- larly, garbage collection programs which allow the refuse of an entire neighborhood to fester for a week before removal may be worse than no collection program. This project opted for a series of half-dosages, a certain amount of drainage maintenance, a cer- tain amount of garbage collection, a certain amount of road work, etc., but the proposed technical service capacity was not calcu- lated in terms of the city's needs. Instead a sort of "seed capi- tal' or pump priming philosophy seems to have been applied which financed start-up equipment in several areas of maintenance, rath- er than complete service needs in any one area. In the future, a more focused approach should be considered, for example, full road maintenance, but no drainage clean-outs or garbage collections. As this project demonstrates, the piecemeal approach fails to ensure maintenance in the end, while it may also weaken the tech- nical service agency. (iii) Certain areas of maintenance are better dealt with through private enterprises or even community participation. SETEMU has found that it is more efficient to hire seasonal teams (for six weeks twice yearly) to clean out drainage ditches, a labor-intensive activity in any case, than to maintain a full-time staff for +-he same purpose. It might prove cheaper in the long run to pay or require communities to look after their own drainage systems. (iv) In general, private sector contractors can move faster than public sector agencies for civil works projects. Timely project imple- mentation may have suffered as a result of using force account rather than local bids. B. Lessons for Municipal Strenathening Programs 6.02 It is telling that the Municipality of Bujumbura turned down or lost control of several of the agencies which were strengthened or devel- oped by the project. Institutional strengthening programs run several risks made apparent by the project: the first, of overloading the local government beyond its operating capacity, as has apparently occurred in certain instances; the second, of creating institutions which are too ex- pensive or too demanding for absorption by a municipality by virtue of their higher salary and technical assistance standards; and the third, of creating a dependency between the special institutions and the Bank, which supports salaries, operating expenses, etc. - 27 - C. Lessons for Urban Development 6.03 Serviced lot projects were originally conceived as a means of controlling urban sprawl and stemming the disorderly spread of squatter and spontaneous settlements. The serviced lot component in this project was implemented on a scale too small to affect overall urban development and its solutions favored the better-off low-income urban families. Meanwhile, squatter settlements proliferated on the Bujumbura's outskirts at an esti- mated growth rate of 82/year. By virtue of their requirements of residency and long-term payment, the serviced lots were beyond the reach of most urban migrants or newly-formed families. Long-term rational planning for urban development must be based on mechanisms other than serviced lots. 6.04 Household electrical hook-ups appear to be in general demand among unserviced neighborhoods. Public street lighting projects should take this into account in project design and promotion, and include domestic connec- tions where possible. D. Lessons for Economic Development Programs 6.05 Local materials, appropriate technologies and low-energy systems may not be compatible with economic development or with employment genera- tion programs because, in many instances, they cost more than standard technologies and are not affordable by low-income, informal sector fami- lies. This project offers some pertinent examples: (i) Houses of adobe walls and tile roofs represent high maintenance costs and low-durability in the context of a low-incom. community in Bujumbura, or even in municipal buildings. (ii) The low-energy stove program was developed as part of the artisan support component. While the low-energy stove was developed suc- cessfully as part of the project, its marketing failed for two reasons. First, the program linked sales to the manufacturers and artisans do not represent an appropriate group to promote new products. As the project's experience with carpenters, metal- workers and tailors/seamstresses should point out, there is a wide difference between a craftsman and a salesman. Second, new prod- ucts rarely get marketed from the bottom up. Most become popular among higher-income groups before filtering down to lower income groups. Insofar as it failed to create a market for the stoves, the project neither increased artisans' earnings, nor promoted significant use of low-energy stoves. 6.06 Income generation programs may have more success working with a wider range of skills, and, in particular, working with merchants and trad- ers rather than with craftsmen. Observation of the project suggests that the creation of markets (even indirectly as in Cibitoke) has a greater impact than building workshops. Several studies of the informal sector suggest that small manufacturing offers significantly less possibility of generating income and employment than do commerce and trade. A recent - 28 - experiment by the Socio-Economic Research Center of the University of Bujumbura (CURDES), financed by the Bank, suggests that business and simple accounting are of more benefit to small business and craftsmen -.han voca- tional training. E. Lessons for Project Design 6.07 The larger and more complicated the project, the longer it will generally take to execute. After a decade of experience with urban proj- ects in Africa, this lesson should come as no great surprise and should cause the Bank no shame. The second Burundian project is given an imple- mentation period of seven years from the outset. A corollary to this is that projects which hinge on the introduction of new laws or cost recovery practices, particularly taxation, will eenerally face resistance. An ap- oropriately ample time frame should thus be applied. VII. CONCLUSIONS 7.01 In the final analysis, the combination of contrasting backgrounds and personalities involved on the Bank's behalf worked to the project's benefit. The project managed to fulfill its program both on the physical works side, and, a rare achievement, on the revenue mobilization side with- in a time frame which compares favorably with urban projects worldwide and more specifically in East Africa (PCR-Annex 4). The project's success on each side, however, must be qualified: with reference to cost recovery, Bujumbura's new financial capacity is not matched by an increase in its operating capacity, and regarding the civil works projects, the increased schools and technical assistance components neither addressed the areas of greatest urban need, nor contributed to the increasing the Municipality's operating capacity. 7.02 In terms of its objectives, while municipal institutions were not strengthened to the extent anticipated by the project, they probably ab- sorbed as much as they could under the circumstances. Among other objec- tives, those of rationalizing urban planning and of improving the standard of living for the majority of Bujumbura's residents were not realized in full. But insofar as the strengthening of municipal revenues and the provision of schools can be influential in "laying a foundation for more efficient future growth", the project achieved a good part of its stated objectives. 7.03 During implementation the project suffered from setbacks, some built in to project design (dependency on public sector construction, re- quirement for urban tax application), others not. The problems with pro- curement and misuse of funds are disappointing, but not totally unexpected in public sector projects. The Audit concludes that extension of the project's implementation round was justified. - 29 - ANNEX 1 COMPARISON OF FROJECT COSTS (in US$ '000) Appraisal Completion $ z $ 2 1. Neighborhood Upgrading 3,470 20.0 960 5.7 2. Serviced Lots 1,170 7.0 810 4.8 3. Building Loans 3,140 9.0 3,010 8.0 4. Promotion of Craftsmen 940 5.6 720 4.2 5. Community Facilities 2,200 13.0 4,800 29.0 6. Municipal Services 2,.00 12.0 1,270 7.5 7. Project Management and Technical Assistance 3,200 19.0 4,300 25.0 8. Design, Studies 400 2.8 800 4.5 TOTAL 100.0 100.0 - 30 - 30 ANNEX 2 Page 1 of 8 BURUNDI PHOTOGRAPHS 1. Kwijabe is a pleasant, well-equipped neighborhood close to the center of Bujumbura. MOP. 2. Resident families have invested in improving and maintaining their houses. -31- ANNEX 2 Page 2 of 8 3. Responding to local pressure, the project unit added these secondary streets after Kwijabe was otherwise completed, both to ease circulation and to provide additional site drainage. t- ' t~.. 4. Interlot drainage was not contemplated at Kwijabe. It is compounded by the latrine overflows, and by inadequate foundation. design. Notice how the foundation line is shored up by additional stonework -32- ANNEX 2 Page 3 of 8 5. The drainage problem at Kwijabe means that families must invest in additional inf ill. 11 6. An adobe house with clay tile roofs could not withstand Bujumbura' s weather; notice the cracking walls and sagging roof. The local materials were rejected by Kwijabe residents. - 33 ANNEX 2 Page 4 of 8 ~ .."'~\'fI . 7. Informal sector activities are spurred indirectly by the u-pgrading and development projects which provided workshops: IIwo 8. ....and markets. Here an extension of the Kwijabe market houses carpenters' sales and some stray goats as well. The market area financed by the bank, with cement floors, roofs and drainage, has generated additional market activities, spilling over into the surrounding area. - 34 - AN1EX 2 Page 5 of 8 9. The market at Cibitoke was neither planned nor improved by the project. Nonetheless, the upgrading there opened roads which in turn opened the way for a new market to spring up. 10. In the workshop financed through the project, carpenters produce more sophisticated products ... -35 -ANNEX 2 Page 6 of 8 *ir 11. ...applying an assembly line process and using power tools. 197J 744 It, I 12. -36- ANNEX 2 Page 7 of 8 -.ACV~ 13. The workshop for a seamstress' and tailors' cooperative has not gained as much momentum. Researchers from the Univegsity of Bujumbura have counted over 1.500 seamstresses and tailors working in the city. They prefer to work at home, or directly in the market place where they can meet their clients. 44 41 14. The welding machine is parked temporarily in the sewing center... - 37 - AIUEX 2 Page 8 of 8 AJL siu iAAV 15. While the metalworkers appear to be experiencing a slack period for lack of demand. - 38 - ANNEX 3 Page 1 of 2 COMMENTS FROM THE BORROWER REPUBLIC OF BURUNDI Ministry of Public Works and Urban Development Office of the Minister Bujumbura, June 8, 1989 Our ref.: 720/590 Mr. Alexander Nowicki Division Chief Operations Evaluation Department The World Bank Washington, D.C. Subject: First Urban Development Project (Credit 1049-BU) Dear Sir: Further to your letter of April 26, 1989 and your fax No. 1877 of May 16, 1989, we have the honor to send you our comments on the Project Performance Audit Report for the First Urban Development Project (Credit 1049-BU). We approve this report overall, which has the merit of drawing attention to design problems, particularly as regards drainage systems and latrines in newly developed plots, the need for a proper sanitation system in residential districts, and above all the role of projects of this type in urban development planning policies. Projects of this type must form part of a national urban policy. The scope of certain activities, and the degree of satisfaction among beneficiaries of specific project activities, was generally limited because of the initial design. We emphasized this point in the May 1988 Project Completion Report. Other activities, on the other hand, had a definite positive impact. Notable among these were the strengthening of the service capability of the Municipality of Bujumbura, with the establishment of the Regie Municipale des Recettes charged with identifying and collecting taxes and other levies, and of Municipal Technical Services to take responsibility for garbage collection, road and drain maintenance and the institution of a cost-recovery system. The construction and rehabilitation of primary schools has been of great benefit to students in Bujumbura and to their parents and teachers, contrary to what the Report states. The importance of a completed program should be evaluated in terms of the degree of satisfaction among program beneficiaries. All of these activities should therefore be entered on the Project's plus side. - 39 - ANNEX 3 Page 2 of 2 As originally conceived, the program for promotion of artisanal activities failed to achieve the hoped-for results. However, the new approach adopted in conjunction with the Bank did produce significant results, even if the number of artisans involved remains small. This situation will be corrected in the Second Urban Development Project, which is to start up shortly. It is not correct to say that the program for artisanal promotion has had zero impact on the nonstructured economy simply because only a small number of artisans in the city have been affected by it. Its impact is in fact quite large when measured in terms of the number of c.ders placed by all categories of the urban population with artisans supported by the Project and the supply facilities offered to all artisans in Bujumbura by the Centre de Promotion Artisanale built by the Project to provide effective promotional services to craft workers. The construction and rehabilitation of housing has created many jobs for the nonstructured sector. In view of the shortage of small and medium-sized private contracting firms in the construction industry, the people who have built and/or rehabilitated all that housing are artisans in the nonstructured sector, working either as "tAcherons" (subcontractors) or as "self-help" builders. Bricklayers, carpenters, joiners, producers and suppliers of building materials, tailors/seamstresses, casual laborers, etc. -- all have found work under this heading. The impact on the nonstructured economy is thus very apparent. In conclusion, therefore, even though the Project was an initial experiment in the area of urban development, it made a highly significant contribution on the institutional level in particular, even if some of the outcomes left something to be desired. We appreciate the new system of ex post evaluation of projects that the Bank has introduced, as it allows objective analysis of problems in the design and implementation phases and enables lessons to be drawn in good time so as to avoid repeating the same mistakes. We wish to thank the Bank for taking account of the comments we made in the First Project Completion Report in preparing and appraising the Second Urban Development Project. - 40 PROJECT COMPLETION REPORT BURUNDI FIRST URBAN DEVELOPMENT PROJECT (CREDIT 1049-BU) July 13, 1988 Africa Regional Office - 41 - REPUBLIC OF BURUNDI FIRST URBAN DEVELOPMENT PROJECT - (CREDIT 1049-ISU) PROJECT COMPLETION REPORT I. INTRODUCTION 1.01 Burundi's population at the time of project identification in early 1978 was about 4 million. The population growth rate, at 2.2% per year, was relatively low for the region. About 95% of the population was scattered in rural areas. Since its independence, in 1962, urbanization was confined to Bujumbura. As a consequence the city grew rapidly during the 1960's at annual rates of 6-9%, but its growth slowed down during the 1970's to a current figure of 5% per year. In secondary centers, growth rates have been highly unstable, with periods of negative growth or stagnation following restrictions on informal activities and trade. 1.02 Housing production in the formal sector, i.e. housing conforming to all zoning and building regulations, was not keeping up pace with population growth. Private individuals produced all of the non-permanent and semi-permanent housing built during this period, which constituted 39% and 52% of new housing respectively. Formal housing (only 9%) was provided by two public housing corporations responsible for construction of relatively high-standard housing, chiefly provided for civil servants: the National Housing Office (Office National du Logement, ONL ) and the Public Housing Corporation ( Societe Immobiliere Publique, SIP ). 1010 units at an average cost of FBu.4,654,910 (US$ 38,000) were built by SIP with a Government subsidy amounting to 52%. This cost was obviously unacceptable for the national budget. 1.03 The principal institution in the urban sector is the Ministry of Public Works, Equipment and Housing. The Minister was the Chairman of the National Urban Commission, composed of 24 members representing all ministries with an interest in urban development, plus the Municipality of Bujumbura and the Chamber of Commerce for the private sector. Bujumbura is the only municipality in Burundi, headed by the Mayor, who is appointed by the President of the Republic and responsible to the Minister of the Interior through the Province Governor. The Municipality's major responsibilities are security, garbage collection, and maintenance of markets, secondary roads and drainage ditches. The Municipality was unable to carry out its responsibilities in 1978, because of insufficient human and financial resources. - 42 - II. PROJECT PREPARATION AND APPRAISAL Identification 2.01 In March 1978, a Bank mission initiated discussions with the Government on the preparation of an urban project. This mission war combined with a mission from the United Nations Center for Housing, Building and Planning. The Government proposed a $12 million project but the mission identified a larger program, in Bujumbura and one or two small towns, estimated at $20 million. Since the U.N.Center did not have sufficient funds to finance the feasibility studies, the mission recommended that the study be financed from the IDA Technical Assistance Credit (613-BU), or from the Preparation Project Facility (PPF). 2.02 In July 1978, a second mission completed the identification and initiated the preparation of the project. Agreement was reached on terms of reference and on a short list of consulting firms for the feasibility study. In order to keep the project on a manageable scale, the mission decided not to include Gitega, the second largest city of the country, and proposed to prepare a project based on an IDA credit of US$ 10 million. The mission agreed with the Government request consisting of tae upgrading of 8,000-10,000 plots and the development of 3,500-4,000 plots in new settlement areas, btc proposed to reduce the pumber of housing loans from 10,000 to 5,000, and to include a health/nutrition component. Preparation 2.03 The consulting firm was selected in December 1978 and a PPF advance of US$ 410,000 was approved on February 16, 1979. In January- February 1979, a preparation mission helped finalize the contract for feasibility studies, which was signed at the end of February 1979. In April and July 1979, two other preparation missions reviewed the consultant's first and second interim reports. A preappraisal mission took place in October 1979; only six weeks prior to appraisal. The reason is that this project was subject to substantial pressure throughout FY80 to ensure Board presentation before the end of the FY. Appraisal 2.04 The project was appraised in November-December 1979 by two Bank staff assisted by five consultants. The main change was the giving up of the health component, due to a lack of cooperation from the Minister of Health, who wanted sophisticated health centers or nothing. The Issues Meeting, held on January 3, 1980, discussed the urban tax proposed by the appraisal mission. It was agreed that this tax should be adopted as the preferred cost recovery mechanism. Nonetheless, this urban tax has never been implemented (see para.3.03). The meeting also pointed out that the proposed 12% interest rate to beneficiaries of housing loans, although equal to forecast annual inflation of 12% after 1982, was higher than the - 43 - nominal interest rates charged at that time in Burundi for housing loans. In particular, long-term loans for hit .-standard housing were bearing an interest rate of 8.5%. In addition, Government was subsidizing 20% of construction costs as well as interest rates on housing loans for civil servants. It was agreed that the proposed 12% interest rate should be kept as a starting point for negotiations with the government on the basis of the need for a non-negative real interest rate, but that the Bank should be flexible and accept a lower interest rate, if the Government raised this issue at negotiations (which was not done). Negotiations and Approval 2.05 A post-appraisal mission took place in February 1980 to ascertain the progress being made in satisfying the conditions of negotiation, review with the new Public Works Minister the content of all project components and confirm the financial and institutional arrangements. BNDE informed the mission that it could not handle housing and artisan credits as agreed at appraisal. BNDE had agreed to acz as fiduciary, keeping records of inflows and out flows of loanable funds and taking charge of the administrative aspects of the credits. In light of this situation, the mission had to fall back on the Project Unit to handle the housing and artisan credits directly. On February 29, 1980, the Government requested a supplemental PPF consisting of: (i) US$ 27,000 to complete preparation of detailed design and (ii) US$ 150,000 to create and start a Project Unit with equipment and technical assistance. This supplemental PPF was approved on May 1, 1980. Nonetheless, an amount of US$ 222,024.96 was undisbursed when the Credit was declared effective. The Credit was negotiated on May 15-16, 1980, approved on June 24, 1980 and signed on July 3, 1980. Major Objectives of the Project 2.06 The major project objectives were: (a)s to improve the living conditions of the urban poor; (b): to demonstrate the feasibility of low-cost housing at affordable standards; (c): to strengthen urban institutionts; and (d): to lay a foundation for more efficient future growth. Project Description 2.07 The project consisted of the following elements: (a) Upgrading of 6 low-income neighborhoods; (b) Preparation of 780 sites and services plots; (c) Housing loans for construction and improvement; (4) Promotion of artisans through training units and credit, and promotion of improved charcoal stoves; (e) Community facilities: construction of 3 primary schools and renovation of 11 existing schools; (f) Improvement of municipal services; (g) Technical assistance and training; and (h) Studies. - 44 - III. IMPLEMENTATION Effectiveness 3.01 The conditions of effectiveness included the signing of the subsidiary loan agreement between the Government and the Municipality of Bujumbura, and the deposit of FBu.25 million (US$ 280,000 equivalent) in a revolving fund. The subsidiary loan agreement was signed on September 21, 1980, and the Credit was declared effective on October 31, 1980, after a first extension of the effectiveness date (initially October 3, 1980). Project Management 3.02 The Project Unit was established by decree No 100/16 of February 11, 1980, within the Ministry of Public Works, Equipment and Housing. The Project Director (Mr.Jean de Dieu Basabakwinshi) was appointed by decree No 100/51 of March 11, 1980. He was replaced, on June 1, 1982 by Mr.Daniel Niro, previously Project Adviser. On May 10, 1983, Mr.Alphonse Niyongere, previously Chief of the Artisans Bureau, was appointed Co- director of the project. He became Project Director when Mr.Niro left the country, in February 1985. Although not included in the original project description, offices were built in Kwijabe and completed in June 1983. A. Maior Revisions 3.03 The major revisions were due to the delays in the finalization and application of cost recovery mechanisms. The SAR proposed that an urban tax (para.5.06) be levied on dwelling rental values in Bujumbura to cover repayment of the Municipality's subsidiary loan from Government and project-related costs of maintenance and garbage collection. This was confirmed in the Project Agreement, section 2.06.b. The Government and the Municipality stated at appraisal that the urban tax, being a municipal tax, could be created by Mayoral Decision (Decision du Maire). This was in fact done, by the Mayor of Bujumbura at the time. In December 1981, at the time the consitution was promulgated, the Government view was that in order to give full credence and credibility to the new experiment with legislative decision-making in Burundi, it was best to re-examine the question of such an important tax in the capital city. It was for this reason that, despite the fact that the urban tax has already been created by the method the Government itself considered most appropriate, the legislature was asked to examine the question. The Government prepared a municipal tax code, which included a provision for an urban tax. The code was approved by the Council of Ministers in November 1982, and presented to the National Assembly. Although approving the principle of the code, the Assembly requested more information on its impacts and urged a reduction of the proposed urban tax rate of FBu 55 per fiscal m2 (physical built and unbuilt surface adjusted for neighborhood, house quality and other factors). As a matter of fact, it was a mistake to go for the full revision of municipal taxes, when all that was required for the project's cost recovery was the urban tax. In this case, the best was the enemy of the good. There is a good case that - 45 - the Bank should have urged the authorities to proceed in phases, first to confine the attention to the urban tax, and only later to entertain a proposal for a renovation of all municipal taxes. 3.04 In March 1983, the Council of Ministers rejected a fifth version of the municipal tax code. Since no progress had been made by June 1983, a supervision mission proposed that project activities be curtailed. The rehabilitation program was stopped after the first phase (Cibitoke). The second phase was limited to construction of two outfalls and the third phase was cancelled. Therefore, the number of neighborhoods scheduled for upgrading was reduced from six to one. 3.05 Instead of an urban tax, as agreed at appraisal, the National Assembly, in November 1983, voted to transfer all receipts of the property and rental income taxes to the Municipality (Law no.1/002 of April 2, 1984 and law no.1/001 of February 2, 1984). This decision was deemed satisfactory. 3.06 Although the Closing Date was December 31, 1985, a supervision mission, in March 1985, proposed to reactivate the rehabilitation program. The municipal technical services (SETEMU) were required to prepare bid documents for improvement of streets and drainage for an amount of about US$ 1.0 million. This component was not executed, as SETEMU wanted to carry out the works themselves, on force account. The following supervision mission did not accept. 3.07 As a consequence, the cost of the upgrading component was reduced from about US$ 3.5 million to 1.0 million. A supervision mission recommended to use the amount made available for the construction and renovation of more schools, for which there was no cost recovery problem. As designed at appraisal, the project included the construction of three primary schools and the rehabilitation of 11 existing ones. Actually, 6 new schools viere built and 19 existing schools were renovated. Annex 1 shows the composition of planned and actual project components. B. Physical Proaress Upgrading of 6 Neighborhoods 3.08 It was decided to execute the works in three phases. The first upgrading phase (street improvements and drainage canals) was started at Cibitoke in May 1981, and completed in March 1982. The final design for the phases 2 and 3 was carried out in 1982, but a supervision mission, in June 1983, proposed to limit the second phase to the construction of two outfalls and to cancel the third phase, because the cost recovery mechanisms were not yet operational, particularly the urban tax, as agreed at appraisal (see para.3.04). The only sub-component not cancelled, because the contract had already been awarded, was the installation of 64 standpipes in low-income neighborhoods. The national water company, Regideso, refused to connect the standpipes to the network, arguing that the Municipality never paid the standpipes' water consumption. After years of discussions, the 64 standpipes were finally installed in 1988. - 46 - Serviced Plots at Kwilabe 3.09 A 33 ha site, 1.5 km. from the city center, was to be graded, filled and provided with a street lay-out, standpipes, and public lighting. PAfter completion of the site survey, construction of the streets and drainage network started in August 1981, and was completed at the end of December 1981 as had been forecast. The lay-out plan was completed by the end of September 1981. It contained 757 plots for housing, with a superficy of 250, 300 or 400 m2, and 21 commercial plots: - 325 plots of 250 m2 at 307 FBulm2 - 316 plots of 300 m2 at 349 FBu/m2 - 116 plots of 400 m2 at 393 FBu/m2 The launching of the works was delayed for 18 months because the project director was not convinced of the soundness of the project concept. In addition to the lack of progress in Kwijabe, the realization by the Minister of Public Works that it was due to a lack of conmitment to the concept of serviced sites was the single most important reason why the Minister fired the Project Director in May 1982. It was also the reason why the Minister chose to appoint an expatriate as Project Director. Despite the very real political difficulties in doing so, the expatriate was the person most likely to get houses built in Kwijabe. The target population was the households whose incomes are between the 15th and the 65th percentiles of the Bujumbura income distribution. The 757 housing plots were allocated by lottery on December 1 and 15, 1981, from a list of 1402 applicants, of which 1358 had met all critbria. Five prototype houses were built and proposed to the Kwijabe allottees. In an attempt to maintain costs within the affordable range for beneficiaries, the housing bureau modified the design of the core unit. Contract for water supply was awarded on April 25, 1982. The works were completed in March 1983 and 10 standpipes were put into operation. Construction of the houses began in July 1982 and was completed at the end of December 1987. 591 houses (78%) were built by small contractors (tacherons), and 166 houses (22%) were built by the owners (self help) but really with the help of small contractors and the technical assistance of the project. Housing Loans 3.10 Housing loans were granted in Kwijabe for the purchase of plots and the construction of houses. 757 loans were granted for the plots under a 25-year lease-purchase agreement. 643 beneficiaries applied for a construction loan with an interest rate of 12% per annum, for a 15-year term, including a 9-month grace period (maximum amount US$ 2,500 equivalent). In addition, 535 loans were granted in upgrading areas for home improvement (maximum amount US$ 1,735). Annex 7 gives the number of - 47 - loans granted each year, their average amount and the total amount per year, for plots, for construction as well as for rehabilitation. The total amount of construction and rehabilitation loans reached FBu.346.5 million, compared to FBu.255.7 million expected at appraisal. The table below summarizes the data concerning the loans : Type of Loan Number Total Amount Average Plots 757 72,892,214 96,291 Construction 643 171,670,208 266,983 Rehabilitation 535 101,859,013 190,391 3.11 Although the granting of full title to all plotholders was not a project objective, an agreement with the Land Title Office empowered the Housing Bureau, within the Project Unit, to deliver Provisional Occupancy Certificates (TOP) to applicants. At the end of December 1987, 10,039 TOPs had been delivered for a total of 12,333 plots in the 8 neighborhoods ( Bwiza, Buyenzi, Nyakabiga, Ngagara, Kamenge, Cibitoke, Kinama and Musaga ). Promotion of Artisans 3.12 The feasibility studies and the appraisal mission had recommended that trainers identified from the ranks of neighborhood artisans carry out the artisan promotion program, operating from Housing Bureau local offices, estabished in headquarters of the UPRONA party. This arrangement was neither realistic nor feasible. Artisans were reluctant to work in the party's headquarters and to spend their time as trainers. As a matter of fact, artisans were assisted by the construction of an Atisan Center at Kwijabe, consisting of 20 workshops, and the installation of a training and service center. Initially located in a building owned by the Ministry of Public Works, it was transferred in 1986 into new facilities built in Kwijabe. It was expected at appraisal that the program would have trained about 500 artisans by the end of the project. A carpentry workshop was in operation since June 1982, and was producing doors and windows for Kwijabe residents, as well as a range of finished products for the public. Metalworkers and tailors were also provided with workshops. Nonetheless, the goal of 500 artisans has not been reached, because the carpenters who regularly attended the center considered it more as a service center than a training center. Therefore, the carpentry workshop was operating at full capacity, but generally with the same carpenters. The total number of artisans recorded and reached by the project can be estimated as follows: Recorded Reached Carpenters : 133 81 Apprentices : 91 Metalworkers : 66 18 Tailors 145 25 Total : 344 215 - 48 - 3.13 It was decided, in March 1984, to transfer the Artisan Promotion Bureau to the Ministry of Labor as of January 1, 1985. A Transfer Agreement was prepared and approved by the Bank. However, the Ministry refused to take over the Artisan Promotion Bureau as it had not sufficient funding in its budget for this activity. Under the Second Urban Project, the training and service center of Kwijabe will be transferred to the Chamber of Commerce, Industry, Agriculture and Artisanat of Burundi (CCIB). 3.14 An important part of the Artisans component was dedicated to the improved charcoal stoves sub-component incorporated with the artisans component by a letter of the Minister of Public works, dated December 11, 1980. This program had been designed to support the testing, production and sale of an improved version of the traditional "imbabura" (charcoal cookstove), to help reduce the heavy expenses for energy by the poor families. This expense represents 27% of monthly income on average and up to 40% in some neighborhoods. The project tested ten different stoves, and for three of them energy savings averaged 35% to 40%. About 1,700 of these stoves (DUB2,DUB7 and DUB10)were sold at neighborhood and central markets, but the population was reluctant to use these new stoves, even if the savings were acknowledged. It was necessary to carry out a market survey to determine the most promising design and the DUB10 model was selected. The project launched several publicity campaigns and from June 1986 to the end of 1987, a total of 9,314 improved stoves (DUB 10) were sold, at a unit price of FBu.250. The users' general consensus, as gauged through an evaluation survey, is that the DUB10 is performing well, saving about 40% of fuel. It is produced by informal sector artisans, who are organized by the DUB Project in some kind of a cooperative. 3.15 The Energy Project, Credit 1593-BU, has an amount of US$ 200,000 earmarked for the continuation of the improved stoves project. In February 1988, it was decided to transfer this activity to ONATOUR (National Peat Corporation). The completion mission, in April 1988, held meetings with the directors of the Project Unit and ONATOUR and helped them to reach an agreement for the transfer of personnel and equipment to ONATOUR. Up to now, the best peat stove is the DUB10. Nevertheless, production of stoves through the current cooperative organized by the project is not a long term solution, Stoves should be produced by the artisans in their own workshops which requires that they earn more money when producing improved stoves. The design of DUB10 has been changed to reduce the production cost. This new model (DUB14) should be taken into production as soon as possible in order to recapture the market before people fall back on their traditional stove. Artisans Credits 3.16 The project provided credit to artisans for the installation and equipment (at 8% interest rate, 9 years, including a 2-year grace period), and for working capital or supply (at 8% interest rate, 18 months, including a 6-months grace period). The most interested were the tailors who borrowed to buy sewing machines. The carpenters were not interested in - 49 - equipment loans because their machines are too expensive. They preferred to use the service center. Some carpenters received installation loans to buy workshops built by the project at Kwijabe. The detail of loans granted is shown in Annex 7. Their total amount is FBu 10.5 million, compared to FBU 58.3 million expected at appraisal. However, it should be noted that the project bought raw materials for delivery to artisans, for an amount of FBu.35.6 million. Community Facilities 3.17 The project included construction of 24 classrooms in three primary schools: 12 classrooms in Kwijabe, 6 in Kamenge and 6 in Cibitoke. An additional 11 primary schools were to be renovated. According to the SAR, construction would have been undertaken by the Education Project Unit (BPE) of the Education Ministry, which had carried out the design, construction and equipping of rural primary schools under the first Bank- assisted education project (CR.679-BU). Nevertheless, by letter dated July 23, 1981, the Director General of Primary Education informed the Project Director that BPE'workload was such that it would not be able to carry out the primary school construction and repair component of the project. Therefore, it was decided that construction and renovation of primary schools would be undertaken through Local Competitive Bidding. The project unit launched a single Local Competitive Bidding for all the construction and renovation works, but divided it into two phases: (i) construction of three new schools and rehabilitation of three existing schools; and (ii) construction of three other new schools and rehabilitation of three other existing schools. 3.18 A contract was signed for the entire program. Three schools were built in Kwijabe, Kamenge and Cibitoke, and three schools were renovated, one in Ngagara and two in Buyenzi. In June 1983, a supervision mission recommended that the second phase (construction of three schools and renovation of three existing ones) not be undertaken, because their completion prior to the Closing Date (12/31/85) was highly unlikely. Since the contract was already signed for this second phase, another supervision mission, in March 1985, proposed to undertake the works and to launch another Local Competitive Bidding for renovation of 13 existing primary schools (see para.3.07). This was completed by the end of 1987. Finally, 6 new schools were built and 19 existing schools were renovated. This was done by local contractors and not by the Ministry of Education, on force account, as proposed at appraisal. 3.19 Three of the five existing markets in the upgrading areas should have been renovated under the project, but nothing was done because of the cancellation of phases 2 and 3 of the upgrading works, with the exception of Kwijabe where the market was renovated and upgraded, near the sites and services area. A sports field was also installed in Kwijabe, but five others in upgrading areas were cancelled for the same reason. - 50 - Improvement of Municipal Services 3.20 This component aimed at 1l1ping the Municipality of Bujumbura improve its capacity in: (a) maintenance of streets and drains; (b) maintenance of public buildings; and (c) solid waste collection and disposal. Existing functional units would have been reorganized into a Technical Services Department (TSD). To support this new department, the IDA Credit financed the purchase of equipment, trucks, vehicles and tools, and the construction, in Kwijabe, of new offices, mechanic workshop and maintenance facilities. Two technical advisers were also financed by the Credit. After the creation of SETEMU (see para.4.02), in 1984, their contracts were renewed along with SETEMU. Technical Assistance 3.21 Technical Assistance was provided under the project to the Project Unit and to the Municipality. At the beginning of the project, there was an excessive turnover, due to strained relationships between the Project Director and the foreign experts. The actual total number of man/months has been 362.5 m/m compared to 282 m/m expected at appraisal. The reasons of the difference (82.5 m/m) were mainly that : - the implementation period was longer than forecast; - the Housing Bureau was not transferred to the Municipality, and the Project Unit insisted to keep a construction officer (60m/m), not included in the appraisal estimate, until the end of the project. The details on the technical assistance provided are given in Annex 6. In addition, 94 m/m of expatriates were used for hierarchical positions, instead of technical assistance. For example, the senior adviser became Project Director in May 1982 (para. 3.02). C.Implementation Schedule 3.22 Project implementation was originally scheduled for a four-year period, January 1981-December 1984, but was not completed until the end of 1987, i.e. seven years, due mainly to the delays in deciding on the transfer of the land tax and of the rental income tax to the Municipality. In addition, when this decision was made, in January 1984, the Bank gave no reaction until March 1985. In addition to the cancellation of five neighborhoods upgrading, the rise of the dollar had left a large surplus that both the Bank and the Government wanted to see spent, particularly for the expansion of components included in the project, such as primary schools. 3.23 In summary, the original four-year schedule was not realistic since the disbursement profile for urban projects worldwide shows an average of eight years. Moreover, preparation works and tendering were to start in July 1980, when the Credit was approved on June 24, 1980 and signed on July 3, 1980. The appraisal estimate and actual implementation schedules are shown in Annex 2. - 51 - D.Procurement 3.24 It was forecast at appraisal that contracts for an aggregate value of US$ 3.86 million equivalent would be awarded on the basis of International Competitive Bidding, for procurement associated with major civil works and equipment. A number of small contracts totalling US$ 0.7 million equivalent were to be awarded under Local Competitive Bidding. The rest of the works should have been done on force account by departmental forces of the Ministries of Public Works, Education and Agriculture, for a total amount of US$4.0 million. Nonetheless, these ministries were unable to carry out these works (see para.3.07) and contracts were awarded after Local Competitive Bidding. 3.25 The only works carried out by departmental forces of Government agencies, REGIDESO and SETEMU, were electricity, public lighting and drainage. For the latter, after Local Competitive Bidding, the contract was awarded to the Sanitation Department of REGIDESO, which later became SETEMU (para.4.02). However, due to the bad quality of the works, the Project Unit had to terminate the contract and give the remaining works to a private contractor. For electricity, REGIDESO carried out the works with considerable delays. Although the contractual completion date was January 31, 1983, the works were neither completed in Kwijabe nor started in Cibitoke at the end of March 1983. 3.26 Bid documents were often submitted to the Bank even if the cost estimates were below the ceilings of US$ 250,000 for works and US$ 75,000 for equipment, as stated in the Credit Agreement, section E.1. of Schedule 3. Generally, the Project Unit had no difficulty in complying with the Bank guidelines for procurement, except in two cases: At the start of the project, in October 1980, the Project Director requested approval of an immediate purchase of four vehicles for the Project Unit, to be acquired through local shopping. The Bank replied that the two vehicles could be acquired through local shopping, as an exception, but that the others would be procured through Local Competitive Bidding, as set forth in the Credit Agreement. As a matter of fact, all four vehicles were acquired through local shopping prior to receipt of the Bank's reply. A supervision mission, then in the field, proposed that the Project Unit order two extra vehicles under competitive bidding and sell, transfer, or otherwise dispose of the two vehicles acquired inappropriately. This proposal was not accepted by the Government. Following consultation with Bank management, the mission agreed to let the situation stand. - 52 - - In October 1985, the Project Unit proposed to award a contract for the purchase of tires and spare parts to the second lowest bidder (FBu 2,816,300), because the lowest bidder (FBu 1,855,196) had not specified the delivery time. The Bank replied that this was not a sufficient reason to pay 52% more. In addition, the lowest bidder was the only one who had accepted the payment schedule of the bid documents. The Project Unit agreed. E.Costs 3.27 The comparison between appraisal estimates and actual costs of the project is shown in Annex 3. When measured in US dollars, actual costs exceed appraisal estimates by only 0.6%. Because of the depreciation of the Burundian Franc against the dollar, however, project costs in FBu exceeded appraisal estimates by about 22.6%. The exchange rates varied from 90 FBu for 1 US$ in 1980 to 123.5 FBU for 1 US$ in 1987. Since the Credit was expressed in US$, the project benefitted by the strength of the dollar from 1983 to 1985. 3.28 Although the actual cost in dollars is about the same as estimated at appraisal, the average hides the large disparity among cost components: - Actual cost for neighborhood upgrading was 72% lower than the appraisal estimate because of the cancellation of five out of six neighborhoods. - For coumunity facilities, actual cost was 120% above appraisal estimate because 6 primary schools were built and 19 renovated, instead of 3 and 11, respectively. - Technical assistance exceeded estimates by 34% because the project was implemented in seven years instead of four. - Actual cost of studies was 60% higher than estimated because, in addition to the studies needed for the project, it was decided in 1985 to finance, from the Credit, the feasibility studies of a second urban project. F.Disbursements 3.29 Actual disbursements from the Credit by quarter compared with appraisal estimates are given in Annex 4, page 1. Disbursements took place over eight years, compared to the appraisal estimate of five years. The delay was due primarily to delays in making decisions on cost recovery, as explained above (see para.3.22). The original closing date of December 31, 1985 was extended by two years to December 31, 1987. - 53 - 3.30 Actual disbursements from the various categories of the Credit, compared to appraisal estimates, are shown in Annex 4, page 2. A graph also gives a comparison with the disbursement profiles for urban projects worldwide and for urban projects in East Africa (Annex 4, page 3). G.Performance of Consultants and Contractors 3.31 The consulting firm which carried out the feasibility studies was competent and produced the reports on time. The basic design criteria and approaches they proposed were followed, except for the artisans component, where more realistic arrangements had to be made during implementation. The overall performance of the foreign experts recruited under the technical assistance component was good. Nonetheless, they did not dedicate enough time to the training of the local staff. Consequently, many extension requests were submitted by the Government and accepted by the Bank. Finally, the share of the technical assistance in the total cost of the project was too important. 3.32 The overall performance of private contractors was very gcod. The only problems occured with SETEMU, a Government agency which was unable to complete satisfactorily drainage works in the suburbs (see para.3.25). H. Reporting 3.33 A series of 11 quarterly reports were submitted in a timely manner to the Bank from October 1, 1980 to June 30, 1983. Then, the Project Unit produced a series of half-yearly reports through June 30, 1987. The reports were comprehensive and accurate, and were used by the Project Unit, the Mintstry of Public Works and other departments for monitoring the progress of the project. 3.34 An independant auditor was retained by the Project Unit (see para.4.08) and audited project accounts were prepared regularly through 1987. Delays occurred at the beginning, but generally the audit reports were received within the eight month period following the end of the fiscal year, as required by the Development Credit Agreement (section 4.01 c). - 54 - IV. INSTITUTIONAL, FINANCIAL AND ECONOMIC PERIORMANCE A.Institutional Aspects 4.01 One of the main objectives of the project was to strengthen the Municipality of Bujumbura. To this end, a Technical Services Department was established and reinforced by the provision of equipment, training, technical assistance and by the construction of offices and workshops. However, in May 1983, a Government decision reversed this policy. 4.02 The Minister of Public Works informed the Bank, by telex dated May 25, 1983, that the Council of Ministers of May 11 had decided to create an autonomous institution called Municipal Technical Services (SETEMU), under the responsibility of the Ministers of P.W. and Interior. The telex proposed that the transfer of TSD to SETEMU be reviewed by the next supervision mission. This mission took place on June 20-23, 1983, but neither the aide-memoire left in the field , nor the confirmation telex, dated June 30, 1983, mentioned the creation of SETEMU. In an action letter to the Minister of Public Works dated July 26, 1983, it was specified that this important proposal deserved serious study, but that there was little urgency to this examination. The letter suggested that this proposal be studied in more detail during a forthcoming supervision mission. 4.03 The decree establishing SETEMU, No.100/126, was issued on July 12, 1983, and sent to the Bank who aknowledged receipt on September 19, 1983. Although this telex specified that the decree had been forwarded to the legal department for comment, no comment was sent to the Government. On January 17, 1984, a telex, to the Minister of P.W., requested that he keep the Bank informed on progress of SETEMU's implementation, including schedule for transfer of responsibilities from TSD to SETEMU. 4.04 The creation of SETEMU could have been a good solution, if the Director General had been responsible to the Mayor. Although the Mayor was the Chairman of SETEMU, there was no coordination beween him and the Director General of SETEMU, because of a conflict of personalities. As a matter of fact, the Director General reported directly to the two Ministers, of P.W. and Interior, who took the decisions with him without the Mayor's concurrence. 4.05 For example, in 1987, the Municipality was required by the Central Government to pay for civil works not incorporated in its own program. The reconstruction of the Ntahangwa Bridge, flood control measures and the " Boulevard du 18 Novembre " represented total expenditures of FBu 325 million, which should have been paid by the Ministry of Public Works. In addition, SETEMU built a bus and truck station at the Municipality's expense. In 1987, the Municipality had to pay FBu 500 million for SETEMU, when its total revenue was only FBu 362 million. 4.06 The project was to strengthen the municipal technical services, but the creation of SETEMU jeopardized this component, as these technical services were no longer within the municipality. Taking back technical services to the municipality is an objective of the Second Urban Development Project. - 55 - 4.07 The municipal collection service became a Collection Bureau under the project having responsibility for collection of project related charges and taxes, and of housing and artisans loans, in addition to current municipal taxes. Nonetheless, the Housing Bureau was never transferred to the Municipality, because the Government wanted to keep the control of housing, through the Ministry of Public Works and Urban Development (General Directorate of Urbanism and Housing). This decision had a negative impact on the institutional strengthening component. Nevertheless, a Land Title Office has been established within the Municipality to deliver Provisional Occupancy Certificates (TOP), to update regularly the land registry and to maintain the list of taxpayers. B. Financial Performance Project Unit 4.08 An independent auditor acceptable to the Association was recruited in April 1982, and the first audit mission took place in May 1982, for the audit of the 1981 accounts. The auditor insisted every year on the need for more effective financial management within the Project Unit. In particular, some employees defalcated funds for their own use (about US$ 13,000 equivalent). A part of this amount has been reimbursed but, at the end of 1986, four employees were still indebted, and two of them for FBu 598,397 each (about US$ 5,000). Cost Recovery 4.09 The setting up of the recovery system was delayed for about two years (see paras.3.03-3.05), and implementation only started in 1984. Due to the transfer of the land and rental income taxes, the Municipality revenue was raised from FBu 99.5 million in 1983 to FBu 248.0 million in 1984. A Collection Department was established within the Municipality, with a view to recovering the project investment and maintenance costs. Its second purpose was that of evolving into the main instrument for general reorganization of the municipal tax system. The collection rate improved, based on a basic file of about 15,000 owners of improved and unimproved properties in Bujumbura. The project played an important role in delivering the Provisional Occupancy Titles (TOPs), which are the basis for taxation. 4.10 The table below shows how the financial situation of the Municipality improved in 1984 and 1985, then deteriorated because of the weight of SETEMUt (FBu million) Year 1981 1982 1983 1984 1985 1986 1987 Revenue 86 97 99 248 345 369 362 Expenditures : other than SETEMU 74 86 94 152 171 205 222 SETEMU 85 112 466 500 Total Expenditures 74 86 94 237 283 671 722 Balance 12 11 5 11 62 (302) (360) - 56 - The Government required the Municipality to bear investment cost not programmed by itself, such as the work on the reconstruction of the Ntahangwa Bridge, improvement of rivers in the wake of flooding, and the Boulevard du 28 Novembre, representing an overall cost of FBu 518 million. For the financing of these works, the Municipality had to borrow FBu 150 million from the Government, and from CAMOFI: a 203 million loan at 16% interest for 5 years, and a 151 million loan at 15% interest for 6 years. The Municipality cannot finance these operations for which it is not responsible, and the takeover of these two CAMOFI loans by the National Budget is a condition of effectiveness of the Second Urban Credit. It seems clear that if the tasks incumbent on the SETEMUs were undertaken directly by the Municipality, their cost would be substantially reduced, and the service undoubtedly improved. Housing Loans and Artisan Loans 4.11 The reimbursement of the housing and artisan loans is collected by the Collection Bureau, within the Municipality. During project implementation, recovery rates were stable, as shown in the table below : Housing Loans Artisan Loans Construction Rehabilitation 12/31/83 77% 37% 87% 12/31/84 79% 53% 85% 09/31/85 72.2% 57.5% 83% 03/31/86 74.4% 54.5% 80.4% 03/31/87 74% 46.5% 77% 12/31/87 76.5% 57.6% 75% The recovery rate for the rehabilitation loans is very poor because their average amount is low, about FBu 190,000, i.e. US$ 1,500. Undoubtedly, lack of motivation among staff of the Municipality of Bujumbura, and the absence of a proper litigation department, account for a good deal of. In addition, the Municipality receives no remuneration for this work, despite the provisions of Credit Agreement 1049-BU stipulating that it would be entitled to 25% of interest collected. Although Bank missions insisted that this clause of the Agreement be observed, the Minister of Public Works never authorized the Municipality to appropriate its own remuneration. The totality of the interests was deposited in a blocked account, No 150L, in the "Banque de la Republique du Burundi" (BRB). Nonetheless, a withdraw of FBu 10 million was used for operating expenses of the Project Unit in 1984, and a second one, on May 30, 1984, to pay for road works in Musaga. The balance of this account will be used for new sites and services operations. C.Economic Justification 4.12 In re-evaluating the economic rate of return for the sites and services component, the same methodology utilised at appraisal was utilized for the analysis. The actual costs of land filling, site preparation, on- site and off-site infrastructure, core housing, labor and materials to complete core housing and commercial and community facilities were included - 57 - in the analysis. Also included in the cost stream were the future maintenance costs estimated at 5% of the investment costs (estimated at 3% at appraisal). The actual costs were deflated to represent prices in 1982 which served as the baoe year (Annex 8). Costs and benefits were considered over the 1982=2000 period. 4.13 Benefits for the housing and commercial areas of the development were based on imputed rental values of the units. The average rental valu, was estimated at FBu 5,000 per month. The appraisal estimate was FBu 3,80 which did not take into account the commercie.1 benefits. On the basis of these assumptions, which are shown in Annex , the economic rate of return - was calculated to be 13.3%, compared to 11% at appraisal. 4.14 No economic re-evaluation has been made for the upgrading component, since only one neighborhood was upgraded instead of six. - 58 - V. THE ROLE OF THE BANK 5.01 The Bank played an important role in Burundi by demonstrating the viability of upgrading as an alternative to slum clearance and reduced construction standards as a means of making shelter affordable by the urban poor. Considerable discussion took place between the Government and the Bank during implementation about the construction materials to be adopted, compatible with the low -income population target. The Bank was also instrumental in convincing the Government of the feasibility of cost recovery, and much attention was given to this issue early in the project. No decision on cost recovery mechanisms would have been obtained without the strong pressure exerted by the Bank. 5.02 Bank supervisi.n was relatively continuous, as only one change in project officer occurred during the life of the project. During the implementation period, a total of 13 supervision missions visited Burundi. The number of staff-weeks per year spent in the supervision effort in the field is given below : 1980 1981 1982 1983 1984 1985 1986 1987 2 11 7.2 4 8.2 2.6 1 1 On the consultants' side, there was a lack of continuity, since 12 different consultants participated in Bank missions. Six of them were financial analysts, and only one participated in two missions. - 59 - VI. POST EVALUATION 6.01 A participant-observer evaluation, or beneficiary assessment as it is now known, was carried out on this project during the 1985-86 period. The purpose of this evaluation was to determine the point of view of the project beneficiaries, actual and potential, regarding new housing, home improvement credit and the artisan employment-generation components. 6.02 The evaluation methodology consisted of conversational interviews with random representative samples of beneficiaries and participant- observation, residenc' in housing sites for protracted periods of time (three months). The evaluation was conducted for nine months by a five- person indigenous team of trained interviewers, all of which had post graduate degrees, from the Social and Economic Research Center (CURDES) of the University of Burundi, at a total cost of US$ 19,000. 6.03 There were a number of findings from this qualitative assessment which were considered useful by project management. These include : (a) People in tha new housing areas preferred to have control over the construction of their houses rather than have the project carry out the construction, even though the construction period was longer. They did express and demonstrate a need, however, for technical assistance, particularly regarding estimates on costs. (b) The reason underlying the low demand for home improvement credit in the upgrading areas, aside from the little knowledge of this component's existence, was found to be people's resistance to assu ming long-term credit obligations for fear of adverse reprisals, should they be unable to meet repayment terms. Given the unstable employment pattern endemic to the urban poor, preference and custom is for financing home improvements in small increments, as time and money allow. (c) The world of the artisan was poorly appreciated by project planners, e.g., tailors and seamstresses preferred their home/workshops to the training center because both clients and materials were close to the former but far from the latter. 6.04 This post-evaluation has been used for the preparation of the Second Urban Project, with the view to increase community participation to ensure that beneficiaries are well aware of the work that will be carried out and if the procedures for land tenure regularization and paymenit. - 60 - VII. CONCLUSIONS 7.01 Overall, the project was successful in addressing deficiencies in housing availability, housing policies and cost recovery mechanisms. The project demonstrated that through the provision of appropriate design standards, new housing can be provided to lower income families which is affordable while at the same time, investment costs can be recovered. Concerning the Municipality of Bujumbura, its revenue jumped under the project from FBu 85.6 million in 1981 to FBu 368.8 million in 1986. 7.02 The project benefitted most of the low-income population of Bujumbura through the upgrading of Cibitoke, the improvement of drainage and solid waste collection, the construction or renovation of 25 primary schools. In Kwijabe, the households had access to the plots from the 15th to the 65th percentiles. This means that the beneficiaries had a monthly revenue of less than FBu 22,500 (US$ 180), which is below the poverty level. 7.03 The major cause of delay was due to the difficulties faced by the Government in deciding on cost recovery mechanisms. The main lesson learned is that two years are necessary to obtain institutioal and financial decisions which require a vote by a National Assemoly. This should be taken into account for the preparation of future projects. BURUNDI - 61 - 44E), FIRST URBAN PROJECT Credit 1049-8U Conposition of Planned and Actual Project Components Project Component Planned at Appraisal Actual A.Neighborhooa Upgraoini Streets and Drainage CIBOTOVE 1981 Lcie 1961 KINAMA,KAliENGE 1982 Not Done. Only kinama Outfall 1982 NYAKABIGA,BWIZA,BUYWAZI Not Done Electricity CIBITOKE,BWIZA 1983 Done 1984 KAMENGE,KINAMA,NSAGARA Not Done Water Supply BUYENZI,BWIZA,KAMENGE Done 1986 NYAKABISA 1982-83 B.Serviced Plots Site preparation 1981 * Done 1981 Streets and Drainage 1981-82 Done 1981-82 Water Supply 1982 Done 1983 C.House Construction KWIJABE.Number of Plots 781 778 Housing Credits: 1080 1186 New Houses 1981-84 651 ,1992-87 Improvement 1981-84 535 ,1983-96 D.Promotion of Artisans Number of Artisans 500 190 Number of Apprentices 90 91 Training Center Not Forecast In Kuijabe, 1986 Artisans Credits 360 180 E.Comaunitv Facilities Construction New Schools V,in 1982-84 3,in 1982-84 31in 1985-86 Rehabilitation of Schools 11,in 1982-84 3,in 1982-84 3,in 1985-86 13,in 1986-87 Markets 3,in 1982-84 1, in 1987 Sports Fields Not Forecast 3, in 1987 F.Municipal Services Construction Headquarter 1982-84 Done 1982-84 Maintenance Workshop 1984 Done 1984 Equipment 1983-86 Done 1983-86 G.Technical Assistance Total Ran/Months 282 362.5 Personnel recruited as Project Officers: 0 94 Total: 282 456.5 ==-_ =-X - 62 - BURUNDI ANNEX 2 FIRST URBAN PROJECT ---- IMPLEMENTAI10 SCHEDULE Project Element 1981 1982 1983 1984 198 1986 1987 A.Neighborhood Upgrading --------------------- B.Serviced Plots -------------- ----- C.House Canstruction ---------------------- D.Prototion of Artisans ----------------------------- E.Conounity Facilities -------- F.Municipal Services ------ - 6.Technical Assistance ------------- ------Appraisal Estioate ====2==Actual May 28, 1988. 2:37 pa. - 63 - - BURUNDI ANNEX 3 FIRST URBAt PROIECT ------- PROJECT COSTS:APPRAIBA6 EIMATES AN' ACTUAL COSTS --------------------EtrF -u -li on-------------- Caponent Apprais;l ICtual Vifference A.Neighborhood Jporajia . . -205.50 -6b.40 .5erviced plots 104.10 53.30 -15.E0 -15.13 C.House Constructiori 279.6V 34.30 54.70 19.56 D.Promotion of Artisans 83.50 77.60 -5.90 -7.07 E.Coasunitv FEcilitie= 195.40 525.20 329.80 168.78 F.Municipal Services 182.60 138.00 -44.60 -24.42 *6.Technical assistance 285.80 467.30 181.50 63.51 H.Studies 42.90 83.60 40.70 94.87 Totals 1483.40 1818.30 334.90 22.58 ------------------US$ thousands------------------- Component Appraisal Actual\l Difference 1 A.Neighborhood Upgrading 3477.50 959.14 -2518.36 -72.42 B.Serviced plots 1169.70 814.35 -355.35 -30.38 C.House Construction 3141.60 3083.10 -58.50 -1.86 D.Prosotion of Artisans 93B.20 715.67 -222.53 -23.72 E.Community Facilities 2195.50 4843.68 2648.18 120.62 F.Hunicipal Services 2051.70 1272.71 -778.99 -37.97 B.Technical assistance 3211.20 4309.69 1098.49 34.21 H.Studies 482.00 771.00 289.00 59.96 Totals 16667.40 16769.34 101.94 0.61 ,i:Figures in US$ have been calculated on the basis of the following exchange rates: Annual Averages 1980 1981 1982 1983 1984 1985 1986 1987 US$ 1.00 = F8u 90.00 90.00 90.00 93.00 119.70 120.70 114.00 123.50 June 4, 1988. 3:50 pa. - 64 - BURUNDI ANNEX 4 FIRST URBAN PROJECT ------ ESTIMATED AND ACTUAL DISBURSEMENTS oaqe I of 2 US$ '000 ------- IDA FY Appraisal Estimates Actual ----- Quarter --------------- ----------------------- Ending Quarterly Cumulate Percent Quarterly Cusulate Percent 1981 12/31/80 598 598 3.99 03/31/81 600 600 4.00 141 739 4.93 06130181 200 800 5.33 91 830 5.53 1992 09/30/81 400 1200 8.00 286 1116 7.44 12/31/81 600 1800 12.00 193 1309 R.73 03/31/82 700 2500 16.67 419 1728 11.52 06/30/82 700 3200 21.33 613 2341 15.61 1983 09/30/82 800 4000 26.67 541 2882 19.21 12/31/82 800 4800 32.00 493 3375 22.50 03/31/83 1300 6100 40.67 862 4237 28.25 06/30/83 1900 8000 53.33 582 4819 32.13 1984 09/30/83 1300 9300 62.00 619 5438 36.25 12/31/63 1200 10500 70.00 492 5930 39.53 03/31/84 1100 11600 77.33 387 6317 42.11 06/30/84 1000 12600 84.00 615 6932' 46.21 1985 09/30/84 700 13300 88.67 528 - 7460 49.73 12/31/94 700 14000 93.33 343 7803 52.02 03/31/85 500 14500 96.67 446 8249 54.99 06/30/85 500 15000 100.00 320 8569 57.13 1986 09/30/85 495 9064 60.43 12/31/85 449 9513 63.42 03/31/86 814 10327 68.85 06/30/86 596 10923 72.82 1987 09/30/86 206 11129 74.19 12/31/86 865 11994 79.96 03/31/87 813 12807 85.38 06/30/87 682 13489 89.93 1988 09/30/87 899 14388 95.92 12/31/87 274 14662 97.75 03/31/88 338 15000 100.00 Iay 28, 1988.3.43pa. - 65 - ANNEX 4 BURUNDI FIRST URBAN PROJECT page 2 of 2 Disbursement of Credit by Category ------- US$ '000 Category Amount allocated in Amount disbursed Difference Credit Agreement ------- --- ------- (1)Street improvement, drainage and building 2000 1265 -735 construction for TSD. (2) Othe, Civil Works. 1425 4605 3180 (3) Equipment, vehicles, 1350 600 -750 materials and furniture. (4) Same for Municipality. 1200 950 -250 (5) Consultants'services. 2200 3550 1350 (6) Operating costs. 750 790 40 (7) Artisan credits. 360 260 -100 (8) housing credits. 1700 2615 915 (9) PPF. 587 365 -222 (10) Unallocated. 3428 0 -3428 TOTAL: 15000 15000 0 June 4, 1988. 4:36 ps. Comparison Between Disbursement Profile Appr Estimate and Actual 100 . 90 - 80 - 70 - 60- 400 30 - (L ~40 30 - 20 - 10 1.1 1.2 2.1 2.2 3.1 3.2 4.1 4.2 5.1 5.2 6.1 6.2 7.1 7.2 8.1 8.2 9.1 9.2 00 Years, Semesters + Urb Proj East Afr 0 Appraisal Estimates 3 Urban Projects Worldwide A Actual --67 - ANNEX 5 Page 1 of 2 BURUNDI: FIRST URBAN PROJECT MUNICIFALITY OF EUJUMBURA: SOURCES AND APPLICATIONS OF FUNDS (FBu Million) A. SOURCES ---------- 1981 1982 1983 1984 1985 1986 1987 Doerating Budget ---- ---- ---- --- ---- ---- ---- Doerating ReceiDts 4.855,18 9,291,714 12,683,728 16,837,912 19,774,193 23,587,636 25,834,658 Income from PropErty :3,682,226 40,929,829 34,197,664 38,884,802 56,218,155 67,381,350 69,491,007 Vehicle Taxes 264,745 118,200 75,050 81,600 205,400 153,800 230,300 Commercial Taxes 46.779,800 45,534,547 52,522,409 51,879,195 74,036,286 83,677,525 84,959,046 Rental Income Tak 124,236,821 138,001,946 148,795,613 139,485,714 Land Tax 15,886,833 55,103,730 44,995,161 40,140,286 Other Receiots 259,580 1,213,912 254,056 1,819,091 Sub-Total E5,581,939 96,874,290 99,478,851 248,066,743 344,553,622 368,845,141 361,960,102 Investment Buiget 1981 1982 1983 1984- 1985 1986 1987 Subsidiary Loan 5,900,542 Sovernment Loans 128,451,532 21,548,468 A0OsFI Loans 88,310,799 266,112,990 Sub-Total 0 0 0 0 0 222,662,873 287,661,458 TOTAL SOURCES 85,581,939 96,674,290 99,478,851 248,066,743 344,553,622 591,508,014 649,621,560 May 30. 1988. 8:43 AM, nI Meail - - ANNEX 5 Page 2 of 2 SURUNDI: FIRST URBAN PROJECT MUNICIPALITY OF BUJUMBURA: SOURCES AND APLICATIONS OF FUNDS (FBu Million) 6: APPLICATIONS Operating Sudget 1981 1982 1983 1984 1985 1986 1987 ater/Electricitv 737,673 2,965,237 12,404,286 27,620,618 5,054,867 33,469,732 5,101,801 Fuel/Oil 4,351,600 6,065,483 7,173,477 3,638,886 3,175,504 3,074,151 3,515,698 Supplies 4.176,260 3,613,300 2,654,309 4,296,290 9,177,142 9,421,332 8,733,286 maintenance 3,371,609 3,894,141 3,098,582 3,832,171 6,532,260 6,991,485 8,657,243 Social Activities 2,944,446 2,218,503 1,731,806 3,023,219 5,204,657 9,398,516 9,771,018 Payroll 34,137,248 40,904,549 47,625,272 50,314,784 69,607,738 75,038,078 93,077,655 Interest Pavaents 707,834 Sub-Total 4?,718,836 60.461.213 74,667,732 92,725,968 98,752,168 138,101,128 128,856,701 Capital Expenditures 1981 1982 1983 1984 1985 1986 1987 BNDE Loan 1,288,900 2101-,700 1,808,040 1,049,040 1,048,440 - 742,689 1,048,440 CAr.OFI Loan 10,000,000 IDA Subsidiary Loan 6,020,063 State Advance 30,000,000 Kanyosha Buildings 1,300,000 600,000 666,000 600,000 Public Gardens 1,539,063 433,200 Transport Equipment 2,673,590 5,100,000 1,817,574 - 5,690,724 2,468,996 Furniture & Ecuipeant 6,943,204 3,198,989 4,521,998 3,203,568 3,035,293 445,205 3,543,000 Sub-Total 10,905,694 11,614,689 8,747,612 10,938.671 4,683,733 16,417,681 37,493,636 External Agencies 1981 1982 1983 1984 1985 1986 1987 Police 13,243,207 13,868,147 10,422,925 23,778,568 30,915,107 23,031,612 35,065,503 SETEMU 84,869,432 111,569,791 466,274,889 499,578,020 First IDA Urban Project 18,580,000 25,554,470 15,214,250 9,542,415 Social Housing Project 6,200,000 11,008,000 10,000,000 11,263,431 Sub-Total 13,243,207 13,868,147 10,422,925 133,428,000 179,047,368 514,520,751 555,449,369 TOTAL AFPLICATIONS 73,867,737 85,944,049 93,858,269 237,092,639 282,483,269 671,039,560 721,799,706 May 18, 1988. S:37 AM. B.Teaaille - - 69 - FROUECT UNIAT: Senior Adviser 54 alm 14 al -40 ale Municipal Engineer 1o mf 45 alf 9 m/e Housing Bureau Adviser 42 ala 46 *a 4 GIs Administrative Assistant 24 m/m 25 a/ 1 aIl MUNICIPALITY Fiscal Adviser 12 aa 52.5 alc 40.5 ala MUNICIPALITY then SETEMU Technical Adviser 24 alm 61 aIm 37 el@ Maintenance Chief 30 ai 57 ala 27 al& ARTISAN PROMOTION Senior Adviser 30 Gil 55 al 25 alm Technical Trainers 18 ala 0 ala -18 al Short Term Assignaents 12 aim 7 /a -5 Gl Sub-Total 282 a/a 362.5 e/a 80.5 ala - - - - - - - - - - - - - Al Str,,ice" Flcts al, kwilazE ER Nuber Plots Censtructior :ota Personal Totai - Loans Loans Laans Contribution Cost 1982 88 7.810,944 24,709,517 -2.520,461 13,909,483 46,429.944 1983 333 33,315,931 73,483,250 10,799,181 23,778,329 130,577,510 1984 226 21,307,223 48,023,613 69,330,83b 13,077,804 82,40B.640 1985 77 7,129,081 18,101,765 25,230.846 5,877,168 31.108,014 1986 23 2,177,360 4,944,400 7,121,760 2,449,000 9,570,760 1987 10 1,151,675 2,407,663 3,5C9,338 0 3.559,338 TOTALS : 757 72,892,214 171,670,208 244,562,422 59,091,784 303,654,206 8: House Ifprovement Loans YEAR Number A@ount Average Neighborhood Nb.of LoansTotal Aaount Average 1983 13 1,822,000 140,154 Buyenzi 29 5,268,130 181,660 1984 68 10,313,900 151,675 Bwiza 81 -14,496,564 178,970 1985 763 48,940,792 186,087 Cibitoke 124 23,784,305 191,809 1986 185 39,494,721 213,485 Kamenge 103 19,681,850 191,086 1987 6 1,287,600 214,600 Kinaza 81 16,303,480 201,278 - = Ngagara 5 1,085,700 217,140 TOTALS : 535 101,859,013 190,391 Nyakabiga 112 21,238,985 189,634 Total 535 101,859,014 190,391 C: Artisans Loans IEAR Loans for Supply Equipaentlnstallation fOT.L 1982 458,136 458,136 1983 4,196,845 .62,284 4,859,129 1984 1,155,997 342,094 1,498,091 1985 134,042 1,253,710 1,397,752 1986 121,660 1,500,000 1,621,660 1987 170,907 218,490 ,00,000 689,397 TOTALB : 6,237,587 -76,578 1,800,000 10,514,165 - 71 - BURUNDI AN' E FIRST URBAN PROJECT 'ost and Benefit Analysis KWIJABE.Serviced Flots. YEAR Deilator Site Prepar ConstructiLi Recur,Cost Total Post Houses Benefits Casn Flow 1982 100 8,679,050 38,619,000 433,952 47,732,002 10 600,000 (47,132,002) 1983 99 36,730,942 96,505,660 2,270,500 135,507,102 259 15,540,000 (119,967,102) 1984 87 20,655,216 53,307,159 3,303,260 77,265,636 570 34,200,000 (43,065,636) 1985 83 6,581,124 19,921,760 3,632,3117 30,135,200 690 41,400,000 11,264,800 1986 79 1,910,539 5,838,499 3,727,844 11,476,882 725 43,500,000 32,023,118- 1987 75 953,530 1,794,037 3,775,520 6,523,087 751 45,060,900 38,536,913 1988 70 0 0 3,775,520 3,775,520 778 46,680,000 42,904,480 2000 70 0 0 3,775,520 3,775,520 778 46,680,000 42,904,480 Net Present Value: 429,418,331 Internal Econotic rate of Return: 13.3 %