India - Second Madras Urban Development and Kanpur Urban Development Projects

Document of The World Bank FOR OFFICIAL USE ONLY MICROFICHE COPY Report No. 10579-IN Type: (PPR) TWEDDLE, E/ X31707 / T9 073/ OEDD3 PROJECT PERFORMANCE AUDIT REPORT INDIA SECOND MADRAS AND KANPUR URBAN DEVELOPMENT PROJECTS (CREDITS 1082-IN AND 1185-IN) APRIL 22, 1992 Operations Evaluation Department Jocument has a restricted distribution and may be used by recipients only in the performance of official duties. Its contents may not otherwise be disclosed without World Bank authorization. GLOSSARY OF ABBREVIATIONS DHRW Department of Highways and Rural Works DOH Department of Health DSW Department of Social Welfare GTN - Government of Tamil Nadu HUDCO - Housing and Urban Development Corporation IRR - Inner Rural Road KDA - Kanpur Development Authority KUDV - Kanpur Urban Development Project Metrowater Madras Metropolitan Water Supply and Sanitation Board MMA - Madras Metropolitan Area MMC (or MC) - Madras Municipal Corporation (or Madras Corporation) 'A Madras Metropolitan Development Authority :1UDP 1 - First Madras Urban Development Project ..UDP ii - Second Madras Urban Development Project PC - Public Convenience FTC - Pallavan Transport Company PWD - Public Works Department SIDCO Small Industries Development Corporation TIIC Tamil Nadu Industrial Investment Corporation TNHB - Tamil Nadu Housing Board TNSCB Tamil Nadu Slum Clearance Board TNUDP - Tamil Nadu Urban Development Project UPUDP - Uttar Pradesh Urban Development Project THE WORLD BANK FOR OFFICIAL USE ONLY Washington. D.C 20433 U.S.A. 4 Director-General ktans [Valation April 22, 1992 MEMORANDUM TO THE EXECUTIVE DIRECTORS AND THE PRESIDENT SUBJECT: Project Performance Audit Report on India Second Madras and Kanpur Urban Development Projects (Credits 1082-IN and 1185-IN) Attached, for information, is a copy of a report entitled "Project Performance Audit Report on India - Second Madras and Kanpur Urban Development Projects (Credits 1082-IN AND 1185-IN)" prepared by the Operations Evaluation Department. Attachment 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 PROJECT PERFORMANCE AUDIT REPORT INDIA SECOND MADRAS AND KANPUR URBAN DEVELOPMENT PROJECTS (CREDITS 1082-IN AND 1185-IN) TABLE OF CONTENTS Page No. Preface ............................................................ i Evaluation Summary ................................................. vi I. PROJEC2 BACKGROUND ........................................... I A. Urbanization in India ..................................... I B. National Urban Policy through the Mid-1980's .............. 2 C. The Project Cities ........................................ 4 D. Bank Role and Strategy in the Urban Sector ................ 6 II. THE PROJECTS ................................................. 6 A. Project Preparation ....................................... 6 B. Project Appraisal and Negotiations ........................ 9 C. Project Objectives ........................................ 12 D. Project Description and Principal Features ............... 12 1. Project Components ..................................... 12 2. Project Coordination and Implementation Arrangements ... 14 3. Project Costs and Financing ............................ 15 III. PROJECT IMPLEMENTATION AND RESULTS ........................... 16 A. Time Delays ............................................... 16 B. Cost Variations ........................................... 19 C. Project Results ........................................... 20 1. Physical Achievements and Environmental Impacts ........ 20 2. Rates of Return and Poverty Impact ..................... 23 3. Financial Performance .................................. 26 4. Institutional Development .............................. 27 5. Policy Impact .......................................... 27 IV. POINTS OF SPECIAL INTEREST ............................... 29 A. Project Impact on City Development ........................ 29 B. The Role and Advantages of Follow-on Projects ............. 31 C. The Filtering-Up of Shelter Improvement Benefits .......... 32 D. Sustainability ............................................ 35 V. THE ROLE OF THE BANK ........................ ........ 36 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 (Cont'd) VI. CONCLUSIONS AND LESSONS LEARNED .............................. 38 A. Conclusions ............................................... 38 B. Lessons Learned ........................................... 40 PROJECT PERFORMANCE AUDIT REPORT INDIA SECOND MADRAS AND KANPUR URBAN DEVELOPMENT PROJECTS (CREDITS 1082-IN AND 1185-IN) PREFACE 1. This is a Project Performance Audit Report (PPAR) on the Second Madras and Kanpur Urban Development Projects in India. Approved on December 16. 1980, the Second Madras Urban Development Project was supported by IDA Credit 1082-IN in the amount of US$ 42.0 million. This credit was fully disbursed. The Kanpur Urban Development Project, approved on October 27, 1981. was partially financed by Credit 1185-IN in the amount of US$25.0 million. US$ 3.3 million of this credit was eventually cancelled. 2. The PPAR consists of an Evaluation Summary and a Project Performance Audit Report (PPAR) prepared by the Operations Evaluation Department (OED). Project Completion Reports (PCRs) for the Kanpur operation prepared by the Infrastructure Division of the Technical Department of the Asia Region (ASTIN) and for the Second Madras project prepared by the Population. Human Resources and Water Operations Division of Asia Country Department IV (AS4PH) were previously sent to the Board as OED Reports No. 8402, dated February 23, 1990, and No. 8688, dated May 23, 1990, respectively. 3. The PCRs provide accounts of the project experiences and indicate their achievements and failures. To give a wider perspective OED has provided additional information about the history of the two operations, their main implementation problems and outcomes, the issues raised, and the principal lessons learned. The PPAR is based on the aforementioned PCRs, the respective SARs, project files, legal documents, discussions at Board presentation, and other relevant material. It also drew from OED interviewa with Bank project staff, discussions with representatives of the Indian central and state (Tamil Nadu and Uttar Pradesh) governments, including all major coordinating and executing agencies, and site visits carried out during an Audit mission in March- April 1990. 4. The Audit generally agrees with the findings of the PCRs and examines several of the key issues raised by the projects in further detail. 5. The draft PPAR was sent to the Borrower for comments but none were received. ii PROJECT PERFORMANCE AUDIT REPORT INDIA MADRAS SECOND URBAN DEVELOPMENT PROJECT (CREDIT 1082-IN) BASIC DATA SHEET CREDIT POSITION (Amounts in US$ Million) As of Feb. 28, 1991 Credit Original Disbursed Cancelled Repaid Outstandint IDA 42.0 42.0 - 0.23 46.03 CUMULATIVE ESTIMATED AND ACTUAL.DISBURSEMENTS FY81 FY82 FY83 FY84 JX8i. FY86 FY87 rY88 FY89 Appraisal Estimate (US$ millions) 0.3 6.6 18.0 31.5 40.5 42.0 Actual (USS millions) 0.0 2.57 12.28 14.2 18.33 21.68 28.21 36.57 42.0 Actual as 2 of Appraisal (2) 0.0 0.38 0.68 0.45 0.45 . 0.51 0.67 0.87 100 Date of Final Disbursement: October 3, 1988 PROJECT DATES Oriainal Actual Identification 06/78 06/78 Preparation Mission 10/15/78 10/15/78 Appraisal Mission 10/08/79 10/08/79 Negotiations 11/10/80 11/10/80 Board Approval 12/16/80 12/16/80 Signing 01/14/81 01/14/81 Effectiveness 04/14/81 04/14/81 Credit Closing 03/31/86 03/31/88 Credit Completion 06/30/85 12/31/87 iii STAF INPUTS Stage of Project Cycle To.al Staftheeks Through Appraisat 47.8 Appraisal through Loard Approval 30.6 Board Approval through Effectiveness 0.5 Supervision 106.6 Total 185.5 MISSION DATA No. of No. of Staff Month/Year Weeks Persons Weeks Through Appraisal 10/78-07/79 8.0 18 144.0 Appraisal thru Board Approval 10/79-08/80 7.8 11 85.8 Supervision I 05/81 2.8 4 11.2 Supervision II 11/81 2.8 3 8.4 Supervision III 03/82 0.3 1 0.3 Supervision IV 07/82 3.2 5 16.0 Supervision V 04/83 2.0 3 6.0 Supervision VI 12/83 2.8 3 8.4 Supervision VII 04/85 1.6 1 1.6 Supervision VIII 08/85 4.2 2 8.4 Supervision IX 12/85 2.2 !. 2.2 Supervision X 05/86 2.4 1 2.4 Supervision XI 10/86 2.8 3 8.4 Supervision XII 03/87 3.2 5 16.0 Supervision XIII 10/88 0.3 2 0.6 OTHER PROJECT DATA Borrower/Executing Agency: Government of India Follow-on Proiect(s): Project: Tamil Nadu Urban Development Project Credit No.: 1923-IN Amount: US$300.20 Board Date: June 15, 1988 iv PROJECT PERFORMANCE AUDIT REPORT INDIA KANPUR URBAN DEVELOPMENT PROJECT (CREDIT 1185-IN) BASIC DATA SHEET CREDIT POSITION (Amounts in US$ Million) As of Feb. 28, 1991 Credit Original Disbursed Cancelled Repaid Outstanding IDA 25.0 22.26 3.33 0.0 27.31 CUMULATIVE ESTIMATED AND ACTUAL DISBURSEMENTS FY82 FY83 FY84 FY5 FY86 FY87 FY88 Appraisal Estimate (US$ millions) 1.0 7.2 14.0 19.5 22.0 Actual (US$ millions) 0.0 2.1 6.5 9.4 11.8 14.3 19.2 Actual as Z of Appraisal (%) 6 29 47 48 53 64 90 Uate of Final Disbursement: April 25, 1988 PROJECT DATES Original Actual Negotiations 02/81 09/81 Board Approval 04/81 10/27/81 Signing - 02/04/82 Effectiveness - 04/22/82 Credit Closing 06/86 06/30/87 Completion 06/86 03/31/88 V STAIF INPUTS Stape of Proiect Cycle Total Staffweejs Through Appraisal 207.0 Appraisal through Board Approval 4.8 Board Approval through Effectiveness 18.9 Supervision 168.8 Total 399.5 MISSION DATA (Not available) OTHER PROJECT DATA Borrower/Executing Agency: Government of India Follow-on Proiect(sY; Project: Uttar Pradesh Urban Development Project Credit No.: 1780-IN Amount: US$130.0 Board Date% April 21, 1987 vi PROJECT PERFORMANCE AUDIT REPORT INDIA SECOND MADRAS AND KANPUR URBAN DEVELOPMENT PROJECTS (CREDITS 1082-IN AND 1185-IN) EVALUATION SUMMARY Introduction 1. The Second Madras and Kanpur 3. Located on the Ganges River and Urban Development Projects (MUDP II having a population of 1.7 million, and KUDP), approved in December 1980 Kanpur was India's eighth largest and October 1981 respectively, were metropolis in India in 1981 and the the Bank Group's sixth and seventh most populous and industrialized city operations in the urban sector in in Uttar Pradesh. Since the 1960's, Tndia. They followed the Madras Kanpur has grown more slowly than Urban Development Project (MUDP I), other cities in the state due to the approved in March 1977, three relative stagnation of the operations in Calcutta -- one of traditional industries which form which was for urban transport -- and much of its economic base. Because an urban transport project in Bombay. of its economic decline and KUDP was rollowed by the multi-city insufficient infrastructure Uttar Pradesh Urban Development investments over the past several Project (UPUDP), approved in April decades, at the time KUDP was 1987, and MUDP II by the Tamil Nadu appraised, health and environmental Urban Development Project (TNUDP), conditions in Kanpur were worse than approved in June 1988. in most other large Indian cities. 2. Madras, with a metropolitan Obiectives and Descrintio population currently exceeding 5.5 million, is India's fourth largest 4. As a follow-on operation, MUDP urban agglomeration and the largest Ills main objective was to further city in Tamil Nadu state. As an develop and expand low-cost solutions important port and regional to urban problems initiated under commercial center, growth has MUDP I. especially in the shelter occurred rapidly in Madras over the sector. The project was specifically past several decades, particularly on inte-ded to support the continued the metropolitan periphery. As is reorientation of housing and also the case with other large Indian infrastructure investments in the cities, Madras suffers both from Madras metropolitan area to make them basic service -- especially water more responsive to the needs of the supply and sanitation -- deficiencies urban poor. It also aimed at and seriously inadequate housing strengthening local institutions conditions. responsible for shelter, urban infrastructure, and transport vii provision. MUDP II was likewise Implementation Experi6nce expected to support improved metropolitan planning, capital 7. MUDP II suftered substantial programming, and budgeting. implementation delays. as the original closing date was extended 5. The objectives of KUDP were twice, from March 1986 to March 1988, similar. Its main goals were to rohile physical completion took three increase the supply of serviced land years longer than anticipated. to the poor through the provision of Despite quick start-up of sites and low-cost residential and small services development and rapid bus business plots and to address the procurement, the project ran into worst environmental conditions in several major stumbling blocks slums located on privately-owned land including land acquisition and in and around the city core. As in construction management problems. Madras, the transfer of land Delays likewise resulted from IDA's ownership to project beneficiaries suspension of credit disbursements and phased housing construction were when the state government failed to the "cornerstone" of the operation. meet financial obligations covenanted Other objectives were to strengthen in the Project Agreement. the institutions responsible for the financing and management of urban 8. KUDP was also subject to services and to develop a statewide implementation delays. While these urban development strategy. were shorter than those experienced by MUDP II. requiring only a one year 6. Both operations involved a extension of the closing date until combination of shelter and non- June 1987, they occurred for some of shelter investments. Sites and the same reasons. Unlike MUDP II services and slum improvement however, physical execution started accounted for more than two-thirds of more slowly in Kanpur due to a total project costs in both cases. combination of insufficient staff and MUDP II also had a major transport -- land acquisition problems. The sites primarily bus procurement -- and services component was further subproject and smaller components for hindered by poor coordination with solid waste management, municipal the local electric utility, which maintenance, and technical resulted in a significant delay in assistance. KUDP, in turn, supported providing connections needed for two relatively sizeable urban sewage pumping stations. In infrastructure (water supply, addition, financial and institutional sewerage, and drainage) investments weaknesses in two executing agencies and contained small solid waste, required KDA to assume direct maintenance, traffic management, and responsibility for the sewer institutional strengthening connection and drainage components. components. The lead project agencies were the Madras Metropolitan Results Development Authority (MMDA) and the Kanpur Development Authority (KDA). 9. MUDP II exceeded its physical targets for sites and services, slum upgrading, and bus procurement. This was largely due to the two-year extension of the project implementation period, together with devaluation-based "savings" of part viii of the IDA credit. However, the impact of land and low-cost housing number of small industry sheds in development in rapidly growing Indian sites and services areas and tenure cities, where the supply of both transfer agreements and home serviced land and adequate shelter is improvement loans in upgraded slum often severely constrained. On the neighborhoods were well below less positive side, this has also appraisal projections. KUDP, in contributed to a tendency for some turn, substantially underran its serviced sites and housing units to targets for slum upgrading and water "filter up" to higher income groups connections and only slightly due to inadequate plot allocation exceeded that for sites and services procedures, transfers of plot despite the one-year extension. In ownership, or rentals. In slum contrast, it more than doubled the areas, in turn, it has led to the appraisal target for sewer eviction of some low-income squatter connections outside shelter residents of larger plots by their improvement areas. original occupants. Slow occupation of sites and services plots has been 10. In both cities, sites and a persisting problem in both cities services and slum upgrading have due to a combination of poor physical resulted in significant environmental access, an initial lack of community improvements in the neighborhoods facilities and on-site security, directly affected. UpZraded insufficient resources to proceed infrastructure, together with the with housing construction, and provision of land tenure, has induced speculation. increased private housing investment in numerous elum areas. Storm 12. Despite these difficulties, the drainage components, on the other experience in Madras and Kanpur hand, have produced only mixed confirms that sites and services and, results, particularly in Madras. In particularly, slum upgrading are much those areas where drainage works have more cost-effective ways of improving been completed, annual monsoon and the shelter situation of the urban other flooding has greatly poor than the approach previously diminished, but poor maintenance followed in India involving the threatens the sustainability of the construction of often highly associated benefits. In addition, subsidized finished housing, whether MUDP II did little to resolve serious for middle and upper-income families citywide environmental sanitation or to accommodate relocated slum problems. KUDP's sewer and drainage dwellers. Despite this and the components, together with privately- Bank's attempts to use the projects operated community toilet and bathing (especially MUDP II) to achieve a facilities in numerous slum areas, substantial redirection of shelter constitute a more satisfactory investments, public sector housing experience from this standpoint. agencies in both cities continue to KUDP's traffic management component provide housing for non-low-income has reportedly also had a positive groups and to relocate slum impact. households into tenenents. Furthermore, despite their role in 11. Both projects had acceptable increasing the supply of improved re-estimated rates of return, shelter for low-income families, especially for sites and services and neither operation was able to slum upgrading, confirming the dramatically reduce existing local potentially significant economic housing deficits. ix 13. Financial performance of the public agencies; and better urban two operations was mixed. The public planning and management more transport company in Madras and the generally. In neither project were water agency in Kanpur both these objectives fully met. Although encountered problems in complying MUDP 1119 policy goals were more with project financial covenants. ambitious than those of KUDP, the Even though cost recovery from sites former had the advantage of being and services, thus far, has been able to build on the gains already quite positive in both cases, that achieved under MUDP 1. Numerous of from slum improvement has been its efforts to extend these considerably less so, as has that accomplishments, however, were from the sewer connections program in frustrated as the result of Kanpur. Attempts to enhance insufficient local political support. municipal revenues likewise fell KUDP ran into similar difficulties. short of appraisal expectations. Partly as a result, inadequate Sustainability coverage and poor quality of municipal maintenance and other 16. The sites and services services remain serious problems in components of both projects are both cities. likely to be sustainable given their good cost recovery record and visual 14. Achievement of project evidence that many occupants mobilize institutional development goals was additional resources to expand and only partial. While MUDP II improve their housing. The only introduced improvements in municipal lingering concern is the provision of and se,.tor agency accounting and community services and municipal management information systems, maintenance of project efforts to enhance MMDA's overall infrastructure, which are even more metropolitan planning, capital problematic in upgraded slum areas. programming, budgeting, and The bus transport component of MUDP performance evaluation roles were II is also likely to be sustainable largely unsuccessful. In Kanpur, in in the short run because of the good turn, although a Traffic Engineering operating performance of the bus and Monitoring Cell was established company and its awareness of the need in the local authority and cost to periodically adjust fares in order accounting, administrative, and to cover costs. However, its longer- auditing improvements occurred within run sustainability is less certain. the water agency, attempts to improve In the same vein, the sustainability the latter's metering, billing, and of project benefits from the water leak detection activities met with and sewerage investments under KUDP only relative success. is uncertain since the willingness of the water utility to increase tariffs 15. The two operations were and improve its collection of expected to have a local development outstanding bills remains policy impact in terms of: increased questionable. public sector provision of and cost recovery from basic urban Findinas and Lessons infrastructure, transport services (Madras), and shelter; enhanced 17. Despite significant execution municipal revenue generation, delays, MUD? 11 exceeded most of its maintenance, and service delivery; targets, particularly for sites and more efficient operation of local services, slum upgrading, and bus x acquisition. The project also problematic, as does the financial achieved many of its financial goals situation of the local water utility. and was able to further demonstrate the viability of the sites and 19. Among the general lessons which services and slum upgrading can be drawn from these projects are approaches to low-income shelter the following: provision. Other positive aspects include environmental betterment in (a) It is necessary to take the sites and services and slum upgrading legal, regulatory and areas, major road improvements which institutional frameworks (eg. have supported broader urban Land Ceiling and Rent Control development tendencies in the Acts) and their implications metropolitan region and the fact that for the supply of serviced land a large share of the beneficiaries of and housing more clearly into the shelter components were at or account in designing future below the poverty level. Many of urban projects in India and these families have reportedly elsewhere. In addition, such experienced substantial increases in operations should seek to income after taking possession of reduce the distortions urbanized lots or receiving tenure in generated by these non-physical slum areas. Given existing constraints on the provision of distortions in urban land and housing urban services and shelter. markets, however, some shelter benefits have inadvertently "filtered (b) Both operations reveal the need up" to higher income groups. for a better ex-ante understanding of the 18. For the most part, KUDP was intricacies and dynamics of also a satisfactory operation. The land and shelter occupancy, target for the project's largest informal "tenure" component (sites and services) was relationships, and social slightly exceeded, although that for organization more generally in slum upgrading was not achieved due slum areas. There is also a to previously anticipated need to develop more agile and difficulties in purchasing slum land market-sensitive mechanisms for from private owners. In the absence public sector land acquisition of legislative changes, this problem and improvement-cum-tenure continues to place a constraint on transfer to occupants of future slum improvement activity in privately-owned slum areas. Kanpur and other Indian cities. With the exception of slum upgrading and (c) Similarly, there is a need to residential water connections, obtain better knowledge as to appraisal targets for project what happens to the initial components were largely met, while occupants of sites and services substantial improvements have plots and of slum properties reportedly occurred in terms of storm once they have been tenured and drainage and solid waste management, improved. It is particularly generating relevant environmental and important to gather information health benefits. As in MUDP II, it regarding the fate of the appears likely that benefit flows lowest income residents of such will be sustained for most project areas to determine the extent components, although, as in Madras, to which, over time, they may municipal maintenance remains be displaced, voluntarily or xi involuntarily, by more affluent (g) Even though it was not fully households. successful with respect to its ambitious "software" (d) Both projects also show that objectives, MUDP II clearly there is often a considerable illustrates the advantages of a time lag between the physical follow-on project in terms of completion of sites and consolidating gains made under services plots and their an earlier operation and occupation by beneficiaries. attempting to move forward on The relative importance of the the often difficult sector various reasons for this (see policy and institutional para. 11 above) needs to be development policy fronts. better understood and measures taken to accelerate the (h) Both operations likewise occupation process so as to demonstrate that the prospect minimize the risk of of Bank Group financing for underutilization of project future projects in the same infrastructure investments. sectors and localities and/or involving the same executing (e) The experience with both agencies can provide operations suggests that considerable leverage with interventions that are mainly borrowers in terms of inducing concerned with shelter their compliance with improvements are not the most politically sensitive financial appropriate vehicle to bring and other covenants. about changes in the organization, financial (i) The two projects illustrate the management, and quality of importance of the continuity of service delivery on the part of qualified and motivated local governments. Such technical and managerial staff objectives are likely to be among executing agencies for much more effectively achieved successful project performance. through specific state-wide Such continuity produces municipal development projects. economies in terms of the "learning curve" involved in (f) Independently of the degree and preparing and carrying out type of local government Bank-supported projects and is participation in future urban one of the benefits of a development operations, the longer-term Bank-borrower planning of such projects relationship in the same sector should give greater attention and locality. It also to issues of institutional facilitates the creation of coordination and, especially, working "alliances" among key post-construction maintenance project agencies and their of project-related personnel. infrastructure and facilities. Local communities should also (j) Finally, MUDP II, in have a more active role in the particular, reveals the operation and maintenance of advantages of strong continuity project-provided services and in project task management infrastructure. within the Bank. This is especially important for multi- xii component, policy-oriented urban development operations such as those undertaken in Madras and Kanpur. 1 PROJECT PERFORMANCE AUDIT REPORT INDIA SECOND MADRAS AND KANPUR URBAN DEVELOPMENT PROJECTS (CREDITS 1082-IN AND 1185-IN) I. PROJECT BACKGROUND 1.01 Credit 1082-IN for US$ 42.0 million for the Second Madras Urban Development Project (MUDP II) was approved on December 16, 1980. This was the Bank's sixth operation in the urban sector in India and followed the First Madras Urban Development Project (IMUDP I). partially financed by Credit 687-IN, approved in March 1977 and closed in December 1982. 1 The credit, which was fully disbursed, was closed on March 31, 1988, after two one year extensions. It has been followed by the state-wide Tamil Nadu Urban Development Project (TNUDP, Credit 1923-IN), approved in June 1988, which is currently under implementation. 1.02 Credit 1185-IN for US$ 25.0 million was approved on October 27, 1981 to help finance the Kanpur Urban Development Project (KUDP). This was the Bank' s seventh urban development operation in India. Other than a water supply and sewerage project for five towns and rural areas in the state of Uttar Pradesh (Credit 585-IN, approved in August 1975 and closed in December 1982), it was also the Bank's first urban development effort in Kanpur. The credit was closed on June 30, 1987, after a one year extension. The last disbursement was made on April 25, 1988, at which time an unused balance of US$ 3.3 million, or 13% of the initial IDA credit, was cancelled. A follow-on multi-city operation, the Uttar Pradesh Urban Development Project (UPUDP, Credit 1780-IN and Loan 2797-IN), was approved in April 1987 and is presently in execution. A. Urbanization in India 1.03 In demographic terms, India remains a predominantly rural country, although the urban population is growing much more rapidly than that in rural areas. Thus, while only 27% of the total population resided in towns and cities in 1988, the urban growth rate was on the order of 4Z a year between 1980-88. as 1MUDP I was audited in 1986 (OED Report No. 6330, dated June 30, 1986). In addition to MUDP I, the Bank's earlier urban development projects in India included three operations in Calcutta (Urban I, II, and Urban Transport, approved in August 1973, December 1977, and June 1980, respectively) and one in Bombay (Urban Transport, approved in October 1986). The Bank's urban development experience in Calcutta is also the subject of an EDI case study undertaken by Biswapriya Sanyal and Meenu Tewari, entitled "Politics and Institutions in Urban Development: The Story of the Calcutta Metropolitan Development Authority," (March 1990), while that in Madras was recently surveyed by Melvin M. Webber, Anna Raines, and Chetan Vaidya, "World Bank Urban Projects in Madras - A Review," Infrastructure and Urban Development Department Diceussion Paper, April 1991. 2 compared with 2.2% for the population as a whole. Even more importantly, despite their relatively low share of the total populption, urban centers contained over 220 million people in 1989. 2 Altogether, roughly two-fifths of the Indian urban population live in cities having 500,000 inhabitants or more, and the number of such centers increased impressively from 11 in 1960 to 36 in 1980. The country's three largest metropolitan areas, finally, are projected to be among the world's principal "megacities" by the end of the century. 1.04 This demographic shift, which is due in good measure to rural-urban migration, has accompanied the much more rapid growth of industrial and service than agricultural output, such that, by 1988, the two former sectors accounted for an estimated 68% of GDP, as compared with 56% in 1965. A Bank-sponsored study of Indian urbanization indicates, however, that the main force behind urban population growth over the past several decades has been the significant decrease in rural-based household manufacturing employment and the corresponding increase 4 in urban factory jobs, rather than the expansion of industrial activity _er se . More generally, the study concludes that recent Indian urbanization has been directly associated with economic growth, especially during the period since Independence, and that, based on the experience of other countries, India is no more urbanized than would be expected at its present stage of development. It also finds both that total factor productivity increases with urban size and that -- through 1981 and with the exception of Delhi -- the largest cities were growing less rapidly than the urban population as a whole, suggesting that, at least from an economic standpoint, government fears of "overurbanization" and excessive growth of the principal metropolitan areas are largely unjustified. B. National Urban Policy through the Mid-1980's 1.05 At the time MUDP II and KUDP were approved, India was in the initial years of the Sixth National Development Plan, which covered the 1980-85 period. According to the SAR for MUDP II (para. 1.02), the Sixth Plan indicated that urban "expansion should take place in medium-sized cities and small towns, while there should be an investment moratorium in very large cities." It also noted that government programs in support of the Plan had tended to "neglect the 2 The corresponding figures for 1971 and 1981 were 107 and 156 million, respectively. 3 UN projections indicate that Calcutta (16.6 million), Bombay (16.0 million) and Delhi (13.3 million) are likely to be among the world's ten largest urban agglomerations in the year 2000, while Madras is expected to have a population of some 8.2 million. 4 Edwin S. Mills and Charles M. Becker, Studies in Indian Urban Development, Oxford University Press, New York, 1986, pp. 39-41. See alsi, Rakesh Mohan, "Urbanization in India's Future," Urban Development Department Discussion Paper, Report No. UDD-61, February 1984, and Margret Thalwitz, "India - Water and Urban Strategy Paper," draft, January 26, 1990, Chapter 1. 5 For a discussion of more recent policy initiatives, see Thalwitz, op. cit., Chapter 2. 3 complementarity of rural and urban development in achieving basic national objectives...( and that] despite major efforts by state governments to formulate and implement policies in the largest cities (including Madras), comprehensive and articulated national and state urban policies remain at an early stage of development in India." Furthermore, even though outlays for housing and urban development had increased somewhat relative to earlier 5-year plans, the absolute amounts of public investment for shelter were "relatively small, compared to the size of the problem." 1.06 The SAR for KUDP (para. 1.04), in turn, observes that specific responsibility for the formulation and implementation of urban development programs in India rests with state governments and local bodies, within the overall policy framework established in national planning documents. "Planned urban development," however, did not receive explicit GOT attention until the 1960's. The Third (1961-66) and Fourth (1969-74) Plans aimed at controlling urban land values, land use planning, standards for housing and other services, and strengthening municipal administrations. The Fourth Plan also recommended that state governments enact comprehensive Town and Country Planning Acts and stressed "the need for positive steps to prevent unrestricted growth of metropolitan areas, by a balanced regional approach to urbanization." The Fifth Plan (1974-79) further emphasized these objectives and highlighted the need for the "integrated development of urban settlements" and to improve the housing conditions of the urban poor. 1.07 The Sixth Plan, finally, recognized the need to provide affordable shelter, together with safe water supply and adequate sanitation, to the urban poor. To achieve these goals, special attention would be given to the modification of existing legislation, land use controls, and minimum plot size requirements, while efforts would also be made to "address the problems of slums by improvement and upgrading rather than by relocation, to maintain easy access to employment centers and avoid destruction of existing housing stock, however substandard it may be." Under the Plan, continuing attempts to "curb migration from rural areas to large cities" would be made through the "distribution of economic and job opportunities in such a way that expansion takes place in medium-sized cities and small towns." 1.08 Various instruments were used in the effort to slow growth of the largest cities by attempting to control or redirect the location of industrial employment. These included: (a) the licensing of all new or expanded industrial activities, which, after 1977, was prohibited in all metropolitan areas and large urban agglomerations; (b) direct investment by government-owned enterprises in low-income and rural areas and in small towns and cities; (c) equalization of the prices of basic intermediate inputs such as coal, steel, and cement across regions, independently of their transport costs; (d) promotion of small-scale industry; (e) development of industrial estates; (f) metropolitan planning, including measures to restrict the expansion of manufacturing employment within large cities, leading plants in some instances to relocate on the immediate outskirts of such jurisdictions; and (g) programs to develop backward districts. Research carried out in the early 1980's, however, suggests that these programs had relatively little impact on industrial location and, since manufacturing accounts for a less than a third of total urban employment, on relative city sizes and growth rates. 4 1.09 In terms of central funding during the 1970's, the Government of India (GOI) initiated ambitious programs "to begin tackling urban deterioration" in larger metropolitan areas, first in Calcutta, then in Bombay and Madras. In addition, a National Minimum Needs Program was introduced under the Fifth Plan to help support slum improvement activities and the Housing and Urban Development Corporation (HUDCO) was established in 1970 to finance state government shelter and urban development programs. 6 Through the late 1970's, about 65% of the housing solutions financed by HUDCO were for Economically Weaker Section (EWS) families having incomes below US$ 45 per month, while another 32% were for Low Income Group (LIG) families with monthly incomes between US$ 45 and 75. 1.10 In general, however, the development of large agglomerations received relatively low priority at both the national and state levels dude to the emphasis given to rural development, the expansion of small and medium-sized cities, and the creation of industrial centers in "special growth zones" in the attempt to contain migration to large urban areas. Partly as a result, in the words of the SAR for KUDP (para. 1.07), "urban service delivery continue(d] to deteriorate, creating adverse effects on the urban economy and on the welfare of the population, especially for the urban poor." The existing regulatory environment, particularly the Land Ceiling Act and rent control measures, moreover, "severely constrained" private sector participation in urban land and housing development, while municipal financial resources were often insufficient to provide "a reasonable level of services." On the more positive side, there was increasing official recognition of the need to improve urban management and reorient investments to the benefit of the urban poor. C. The Proiect Cities 1.11 Madras, with a metropolitan population currently exceeding 5.5 million, is India's fourth largest agglomeration. a The Madras Metropolitan Area (MMA) covers 1,167 square kilometers and is composed of Madras City -- which occupies the core area of 172 square kilometers -- four municipal towns, four townships, and twenty panchavats, or non-municipal urban areas. Madras is the largest city in Tamil Nadu state, situated in the southeastern part of the country on the Bay of Bengal. Tamil Nadu, which has a population of over 50 million, is one of the most populous and industrialized, but one of the less 6 For specific information on shelter programs and finance in India, see Ranjana Madhusudhan, "Housing and Housing Finance in India: An Overview," Urban Development Department Discussion Paper, Report No. UDD-33, August 1983. ' For a discussion of the Urban Land (Ceiling and Regulation) Act, enacted in 1976, rent control, and other constraints on housing supply in India, see Devendra B. Gupta, "Urban Housing in India," World Bank Staff Working Papers, Number 730, April 1985. a For more detailed descriptions of urban development tendencies and problems in Madras, see K.C. Sivaramakrishnan and Leslie Green, Metropolitan Management: The Asian Experience, Oxford University Press, New York, April 1986, and John P. Lea and John M. Courtney (eds.), Cities in Conflict: Studies in the Planning and Management of Asian Cities, The World Bank, 1985. 5 affluent, states in India. 9 It is also the second most urbanized state in the country (af,-er Maharashtra). even though its urban population growth fell from roughly 3.3% a year in the 1960's to 2.5% in the 1970's due to a slowdown in the regional econcmy, especially in the industrial sector. 1.12 As an important trade and commercial center and the fourth largest port in the country, Madras has nevertheless grown rapidly over the past several decades, primarily at the result of migration, and registered the highest growth rate (roughly 5%) among Indian metropolitan areas during the 1960's. Areas on the metropolitan periphery continue to gain population at a rapid rate. As is also the case with other large Indian cities, Madras suffers both from basic service, especially water supply and sanitation, deficiencies 10 and inadequate shelter conditions. At the time HUDP II was appraised, about 25% of the metropolitan population lived in unserviced slum hutment areas, while another 25 to 30% resided in older slums and/or unauthorized and poorly serviced settlements. Due to continued population growth, it is estimated that between 30,000 and 40,000 additional families need to be housed in the MMA each year. u 1.13 With a population of 1.7 million, Kanpur was the eighth largest metropolis in India in 1981 and the most populous and industrialized city of the north central state of Uttar Pradesh. 1 Located on the southern bank of the Ganges River, it is the principal urban agglomeration in the central region of India. Unlike Madras, however, it is not the state capital, which is located at Lucknow. Uttar Pradesh is demographically the largest, but also one of the poorest states in the country, having a population of some 110 million in 1981 and a per capita income roughly 30% below the national average. 1 Although most of the state's population continues to be rural, urban areas -- particularly medium-sized and large cities -- are expanding rapidly. Demographic growth in Kanpur, however, was estimated at only 2.8% a year in the 1970's, compared with 4.9% for the state's urban population as a whole. 9 In 1977-78. per capita income in Tamil Nadu was 360 Rupees, which was below the average for India as a whole (385 Rupees). 10 Madras, in fact, has perhaps the poorest water supply and sewerage systems among major Indian cities. Due to its low-lying flat terrain, drainage is also a serious problem, especially following the annual monsoon. The Madras Water Supply and Sanitation Project (Credit 1822-IN), approved in June 1987, is attempting to address part of these deficiencies. 1 A. Lakshmanan and E. Rotner, "Madras, India: Low-Cost Approaches to Managing Urban Development," in Lea and Courtney (eds.), op. cit., pg. 88. See also Chapter 14 in Sivaramakrishnan and Green, op. cit., especially pp. 233-234. 12 Kanpur's population in 1989 was estimated to be slightly over 2 million, making it the ninth largest urban agglomeration after Bombay, Calcutta, Delhi. Madras, Bangalore, Ahmadabad, Hyderabad, and Pune (which was smaller than Kanpur at the beginning of the decade). 13 In 1977-78, per capita income in Uttar Pradesh (275 Rupees) exceeded only that of Bihar (231 Rupees) and Madhya Pradesh (270 Rupees). 6 1.14 Kanpur's industrial base developed initially in the mid-19th century with the establishment of major cotton and wool textile mills and leather industries. Large enterprises for fertilizer and armaments production and many smaller manufacturing ventures were added after World War II. The city's industrial activity has declined in the post-war period, reflecting the relative stagnation of traditional industries throughout India and sluggish agricultural growth in Uttar Pradesh. Due to its economic decline and insufficient infrastructure investments over a period of several decades, at the time KUDP was appraised, health and environmental -- including housing and basic urban service -- conditions in Kanpur were worse than in other cities in the state and elsewhere in India. The large number of slums located on privately owned land, known locally as "ahatas," was a matter of particular concern. D. Bank Role and Strategy in the Urban Sector 1.15 As all of the projects approved through the early 1980's suggest, the Bank's urban lending program in India initially had a strong metropolitan focus, concentrating either on urban transport or taking a broader cross-sectoral or "integrated" approach. More generally, the Bank Group's strategy for the Indian urban sector was to assist government initiatives which contributed to improved efficiency of the urban economy and werL directed toward the needs of low-income groups and the adoption of replicable low-cost service systems. The strategy also focused on the strengthening of local institutions in support of their efforts to expand and sustain urban development programs. II. THE PROJECTS A. Proiect Preparation 2.01 According to the PCR, ' the design of MUDP II was firmly based on the "positive lessons and negative experiences" of the first urban project in Madras. In this connection, it sought to increase the scale and scope of the sites and services and slum upgrading activities supported under MUDP I and to place the proposed shelter investments in the context of the needs of the metropolitan area as a whole. 2.02 The main problem areas encountered by MUDP I, which were addressed with varying degrees of success during preparation of the second project, were: (a) land acquisition - prior acquisition of the areas required for sites and services was initially set as a condition for negotiations, but this was subsequently relaxed by the Bank, creating difficulties later on; (b) land tenure for slum improvement beneficiaries -provision of several thousand lease-cum-sale tenure agreements and the effective initiation of cost recovery from MUDP I slum upgrading beneficiaries was a condition of negotiations for the second project; (c) bus fare revisions - which were obtained, albeit temporarily, prior to negotiations; (d) improvements in municipal finance and services - through the establishment of an urban surcharge on the state sales tax and agreement with OED Report No. 8688, dated May 23, 1990, paras. 9-11. 7 state authorities on related financial and institutional strengthening measures to enable the Madras Corporation to improve infrastructure maintenance and other municipal services; and (e) expansion of the supply of serviced land through the public sector - by restricting government provision of finished housing for middle and upper income groups and tenement construction for relocated slum dwellers. Serious problems nevertheless persisted in rnst of these areas. 2.03 Design of the second project also incorporated "tested innovations" introduced under MUDP I, including: (a) full cost recovery and unsubsidized sites and services and slum upgrading solutions; (b) affordable planning and engineering standards; (c) differential pricing of plots for different income groups within the target population; and, (d) provision of long-term tenure to squatter households in slum improvement areas. Several additional innovations were introduced by MUDP II, specifically: (a) a "city strategy approach" to housing provision which "placed the supply of shelter under the project in the context of total supply from all sources and its relation to shelter needs;" (b) home improvement loans and grants for slum upgrading beneficiaries of both MUDP I and II; (c) the funding of NGO-operated "medicare" centers in sites and services and slum improvement areas; and (d) inclusion of a modest storm drainage component. The project likewise attempted to introduce slum improvement on private lands, but discussions on this issue broke down at negotiations. 2.04 Project preparation was carried out by the main implementing agencies, under the coordination of the Madras Metropolitan Development Authority (MMDA) and following procedures and design standards adopted during MUDP I. Previous participation of most of these agencies in the first project greatly facilitated preparation (and subsequent implementation) of the second operation. As suggested above, however, despite the serious problems encountered in this connection by MUDP I, the land required for sites and services development under MUDP II was not fully acquired prior to project approval, while a bus fare increase agreed with IDA as a condition for negotiations was later partially rescinded by the state government without informing the Association, in the process undermining the revenue projections made for the bus company at the time of appraisal. Furthermore, even though MUDP I succeeded in demonstrating that the provision of serviced urban land, the transfer of land tenure, and in situ slum improvement were cost-effective ways of meeting the shelter needs of the urban poor, IDA's attempt to modify the traditional finished housing approach of state housing agencies under MUDP II was to prove only partly successful. 2.05 A review of the project files reveals that MUDP II as eventually appraised was considerably less ambitious than originally conceived. In addition to endorsing the state government's desire to expand low-income shelter improvements, including provision of as many as 30,000 additional serviced sites and alum upgrading for another 75,000 families over a three year period, early IDA preparation missions gave considerable emphasis to Madras' serious drainage problems. These, in fact, were the specific subject of a consultant's report, which recommended development of a longer-term investment program whpse initial works could be financed as part of the second project. Attention was also given to employment generation and, with UNICEF's assistance, to maternal and child health care measures. As a result, the initial project profile proposed separate components both for flood protection and employment creation in addition to sites and services, slum upgrading, bus transport, road construction, traffic 8 management, and technical assistance. Physical design improvements in the shel-er components were similarly stressed in early mission reports. As concerned bus transport, the focus was on improving the financial situation, operating conditions, and cost-effectiveness of the bus company. 2.06 As preparation proceeded, the drainage component was sharply scaled down because the larger program of works originally proposed was not expected to be ready in time for appraisal, and shelter improvement targets were neduced, with a corresponding decrease in total project costs. The principal issues identified during the latter stages of preparation were: (a) sites and services - land acquisition, the number of plots to be developed, shelter design, plot layouts, and infrastructure standards; (b) slum upgrading - plans for tenure transfer, user charges, and cost recovery; and, tc) bus transport - the need for financial reforms. While no specific policy differences with the state government were identified, a possibly expanded role for the Madras Corporation in the slum improvement component and the need to improve its revenues, services to project implementing agencies, and maintenance of completed project facilities were highlighted as issues to be addressed during appraisal. 2.07 KUDP's design was described in the respective PCR 2 as having been based on "a well conceived approach" by the principal executing agencies and the Bank Group "toward addressing the problems in Kanpur." Project preparation was characterized as having been "very thorough, giving all parties concerned a clear idea of the objectives and how they would be realized." Much of the credit for this is attributed to the Project Monitoring Cell (PMC), composed of "experienced senior officers with strong motivation," whose coordinating role was expected to continue throughout implementation. The only major problem related to project preparation cited by the PCR involved legal complications associated with land acquisition. Litigation with private owners resulted in the loss of land at one of three planned sites and services locations, while similar difficulties plagued attempts to acquire "ahatas" for slum upgrading. 2.08 KUDP was based on a "project approach document" prepared by the Kanpur Development Authority (KDA). This report presented a comprehensive overview of the urban problems facing Kanpur, giving particular emphasis to environmental sanitation and poverty/employment issues, and proposed a project containing three main components: (a) investments to meet basic shelter and environmental infrastructure needs, especially of low-income families; (b) employment generation and productivity enhancement; and (c) improvements in the planning, financing, and management of urban growth. As in Madras, shelter improvements were to involve both sites and services (15-20,000 plots) and slum upgrading (potentially benefitting as many as 130,000 families in central Kanpur). The design principles and criteria to be adopted were essentially the same as those in Madras. Environmental sanitation investments would include both citywide interventions and works linked directly to specific sites and services and slum upgrading schemes, together with improvement of the operational capacity of the local water agency. Small-scale enterprises were to be supported through a combination of financial and technical assistance, and attention would be given to road improvements and traffic management. 2 OED Report No. 8402, dated February 23, 1990, para. 7. 9 2.09 IDA's identification mission stressed the need to pursue "appropriate program policies" and municipal finance and taxation reforms so as to improve local capacity to manage urban development. It concluded that the project should give special emphasis to the "ahatas" in order to "eliminate the worst public health hazards within residential areas and provide the urban poor with increased access to basic infrastructure services." It also recommended that the project's environmental sanitation activities be "limited to support of highest priority functions," especially solid waste management, sever clearance, and drainage, and affirmed that an important objective of the operation should be to improve the ability of local institutions to provide and maintain urban services, requiring significant strengthening of both KDA and the municipal government. The main issues raised in the first two project briefs were the institutional improvements required by the local water agency -- which was experiencing considerable difficulties implementing an existing IDA credit -- project land requirements, the potential problems associated with application of acquisition procedures under the Uttar Pradesh Slum (Improvement and Clearance) Act of 1962, which had not yet been tested in the courts, and the need to define "appropriate interest rates" for the proposed shelter solutions, in light of the state government's "long-standing practice of interest subsidies." 2.10 Subsequent missions recommended that project scope be cut back due to insufficient progress with respect to "ahata" acquisition and slow preparation of environmental sanitation investments. As in Madras, it was suggested that measures to support small-scale enterprises be incorporated in the sites and services and slum upgrading components and it was proposed that traffic improvements be implemented through a "Traffic Engineering and Management Cell" to be set up in the municipal government, which would initiate "a limited road and sidewalk rehabilitation and maintenance program." It was further recommended that the project contain equipment, material, and manpower resources to improve municipal maintenance activities. These proposals were reflected in the third project brief, which divided the operation into three main components -- shelter, environmental sanitation and maintenance, and technical assistance. B. Proiect Alppraisal and Negotiations 2.11 MUDP II was pre-appraised in June and appraised in October 1979. Issues raised at the pre-appraisal stage included: (a) progress in the provision of land tenure ("patta") to and collecting improvement charges from MUDP I slum improvement beneficiaries; (b) the need for fare increases on the part of the bus company; (c) the need for financial strengthening of the Madras Corporation; (d) project costs and financing in relation to the state governmeat's overall investment program; and (e) the executing capacity of project implementing agencies. The focus on the project's financial scope vis-a-vie Tamil Nadu's broader public investment program reflected IDA's preoccupation with the state government's ability to meet expected counterpart funding requirements for other proposed operations, including the Madras water supply and sewerage operation and a statewide nutrition project, which were expected to be implemented in parallel to MUDP II. 2.12 It was suggested that one way to reduce pressure on the state budget would be through the reduction of expenditures for non-Bank assisted shelter programs, particularly slum clearance/tenement building and public housing 10 1 provision for middle and upper-income groups. The potentially high concentration of investments in Madras relative to the rest of the state if MUDP II and the proposed water project were to be carried out at the same time was also a concern. 3 It was also recognized that a "realistic approach" to cost recovery should be adopted, given likely "political constraints" to raising bus fares and (under the proposed water supply operation) water tariffs, while introducing cost recovery for slum improvements. 2.13 At appraisal, the scale of the sites and services component was reduced from 26,500 to 14,900 plots and the target number of households to be benefitted from alum upgrading was decreased from 65,000 to 50,000. In retrospect, this scaling-down of the operation was justified in light of the implementation requirements for meeting the more ambitious shelter targets. However, it also meant that MUDP II was not able to have as large an impact on the metropolitan housing deficit as originally desired. This trade-off between the "implementability" of Bank-supported integrated urban development projects and the scale of their physical and socio-economic impacts in relation to local needs, particularly in large metropolitan areas, has constrained their scope and effectiveness and contributed to an increasing reliance on more streamlined subsectoral or "wholesaling" approaches, which, through separate, less place- specific (in the sense of not being pre-identified and designed) operations, have focused either on urban transport, shelter provision, housing finance, or municipal development. 2.14 A number of "experimental measures," including the purchase of hand carts, bins, and other equipment, construction of a transfer station, acquisition of vehicles for refuse collection and of vehicles and equipment for sanitary landfill disposal areas, together with related consultant services, were proposed for a pilot solid waste management component. It was also agreed that, as in MUDP I, the second operation would include consultant services to MMDA and other project agencies to continue development of medium and long-term investment programs for the metropolitan area, carry out special studies to identify poverty alleviation interventions, and assist the Madras Corporation to evaluate and improve its maintenance activities. A 12% interest rate on loans for shelter improvement was proposed, as was recovery of 75% of the chargeable cost of land and on-site infrastructure in slum upgrading areas and 80% recovery of what was characterized as an "innovative mixture" of home improvement loans and grants. 2.15 Conditions for project negotiations included: (a) a 15% bus fare increase; (b) the issuing of land tenure agreements to 2,000 dwellers of alums improved under MUDP I; (c) "demonstrable progress" with respect to the physical occupation of sites and services plot.x developed under the first operation; (d) design and installation of systems for the collection and accounting of alum improvement charges and administration of home improvement loans; (e) submic3ion of the approved final valuation statement for one of the areas to be used for sites and services development "as evidence of adequate progress with respect to 3 These concerns were later lessened when it was decided not to go ahead with the Madras Water and Sewerage Project at that time. However, this meant the need to defer investments of the highest priority for the metropolitan area which were not addressed by MUDP II. 11 land acquisition;" and (f) improvement of the financial situation of the Madras Corporation. Limitation of publit; housing expenditures for non-low-income groups and slum clearance/tenement construction was also proposed as a condition for negotiations. 2.16 Dissolution of the state government in early 1980 resulted in unanticipated delays with respect to GTN's compliance with the conditions for negotiations, particularly the proposed bus fare increase, the issuing of lease- cum-sale agreements to slum dwellers, and submission of specific proposals to increase municipal revenues. These difficulties led to a "pre-negotiation" mission in March 1980, at which time state authorities informed IDA that assurances could not be given with respect to the limitation of public housing investments for groups other than the target populations. Such restrictions, however, were later incorporated as specific covenants in the Project Agreement. 4 The mission concluded that prospects for completing the Bank Group's other requirements were "reasonably good," and it was estimated that negotiations could be held in September 1980. Despite this, a second "pre-negotiation" mission proved necessary in August-September 1980 in order to consolidate agreements with the new state government. Negotiations finally occurred in November 1980, more than a year after the appraisal mission. 2.17 KUDP was pre-appraised in September 1980 and appraised in February- March 1981. As in MUDP II, the main issues at the pre-appraisal stage included interest rate levels and land acquisition, together with the need to strengthen the local water supply agency. Following the mission, it was recommended that an existing state slum improvement program, which operated through grants and did not provide land tenure, be discontinued because it conflicted with the project's approach entailing the transfer of land tenure to and cost recovery from slum residents. The pre-appraisal mission also suggested that KUDP be viewed as the first phase of a larger urban development program, which later materialized as the UPUDP. 2.18 At appraisal, it was decided that an urban community development project sponsored by UNICEF would be implemented concurrently with KUDP. As in Tamil Nadu, the state government agreed that a 12% interest rate be applied to the financing of shelter solutions under the operation. Because of the anticipated demands on existing institutional capacity, the project execution period was extended from three to four years. An ambitious sewer connection program, that was ultimately expected to benefit some 20,000 "ahata" households and 40,000 dwellings in other parts of the city, was added to the project at this time. One of the motivations for this program was the existence of recently installed sewer lines that were seriously underutilized due to an insufficient 4 Sections 3.02. and 3.03, respectively. The former restricted annual investments by the Housing Board for non-EWS groups to not more than 55Z of its total housing and plot investments starting in April 1982, while the latter limited the Slum Clearance Board's slum clearance-cum-tenement expenditures to no more than Rs 37.500,000 a year between April 1981 and March 1986. s To define the details of this project and its articulation with KUDP, a UNICEF staff member participated in the IDA appraisal team. 12 number of residential connections. It was, thus, proposed that a loan fund be established under KUDP to finance about 30,000 sewer connections over four years. C. Proiect Obiectives 2.19 As a follow-on operation, the main objective of MUDP II was to further develop and expand low-cost solutions to urban problems initiated under MUDP I. especially in the shelter sector. The project was specifically intended to support the continued reorientation of shelter and infrastructure investments in the Madras metropolitan area so as to make them more responsive to the needs of the urban poor. MUDP II also aimed at strengthening local institutions responsible for the provision of shelter, urban infrastructure, and transport. To ensure replicability, the principle of full cost recovery for sites and services, introduced under the first operation, was to be continued, while actual cost recovery from slum upgrading and other investments was to be increased. The project was likewise expected to provide continuing support for "effective" metropolitan planning, capital programming, and budgeting. 2.20 The objectives of KUDP were largely similar. Its main goals were to increase the supply of serviced land to the poor through the provision of low- cost residential and small business plots and to address the worst environmental conditions prevailing in slums located on privately owned land and other squatter settlements in and around the central city. The transfer of land ownership to project beneficiaries and the phased construction of dwellings were to be the "cornerstone" of the operation. Full recovery of all allocable costs for sites and services and slum upgrading and "the reduction and elimination of direct subsidies inherent in the ongoing 'minimum needs program' and other shelter programs" would also be pursued. Other key project objectives were to strengthen the insti4utions responsible for the financing and administration of urban services and to develop a strategy for statewide urban development and management. Institutions to be reinforced under the operation were: KDA; Kanpur Nagar Mahapalika (KNM), the municipal authority; and Kanpur Jal Sansthan (KJS), the local water supply and sewerage company. D. Project Description and Principal Features 1. Proiect Components 2.21 To achieve its objectives, MUDP II, as appraised, consisted of the following components (and approximate cost shares, excluding administration and contingencies): (a) Sites and Services (30%): construction of serviced residential plots, core housing, and community facilities and provision of shelter loans for about 15,000 predominantly low-income households (or some 83,000 people), together with serviced land, sheds, and machinery loans for industrial and commercial activities at two sites covering a total of 180 hectares; (b) Slum Upgrading (38%): improvement of on and off-site infrastructure and provision of land tenure, home improvement loans, and community facilities for roughly 50,000 families, 13 plus home improvement loans for households in slums improved under MUDP I; (c) Transport (25%): provision of about 550 buses and depot improvements for the Pallavan Transport Company (PTC), construction of 6 km of the Inner Ring Road, and general improvements to 14 km of the Madras-Tiruvellor Road; (d) Solid Waste Management and Maintenance of Municipal Services (5%): provision of civil works and equipment to the Madras Corporation (MC) to upgrade local solid waste management and the maintenance of municipal infrastructure, together with consulting services for the review of organizational, technical, and financial aspects of these activities; and (e) Technical Assistance (2%): to MMDA, the Small Industries Development Corporation (SIDCO), PTC, and MC for studies to identify priorities and prepare programs for mediun and long- term investments, together with institutional strengthening and training, particularly for MMDA. 2.22 With the exception of bus transport, the components of the Kanpur project were largely the same as those of MUDP II. KUDP, however, contained a more ambitious environmental sanitation component. The projected cost shares of the two main shelter components were also different between the two operations due to the greater "weight" of sites and services in relation to slum upgrading in KUDP vis-a-vis MUDP II. The breakdown of project components and subcomponents (and their approximate cost shares, excluding administration and contingencies) in KUDP, as appraised, was as follows: (a) Shelter (i) Sites and Services (47%): about 14.800 residential plots, together with core housing, shelter loans, and community facilities, and 540 serviced small business plots at three locations, benefitting an estimated 83,000 people; and (ii) Slum Upgrading (24%): upgrading of "ahatas" and other slum areas, including grants of land tenure, improved infrastructure services, home improvement and sanitary core loans, community facilities, and small business support, benefitting some 112,000 residents. (b) Environmental Sanitation, Maintenance, and Traffic Management (i) Water Supply, Sewerage, and Drainage (18%): measures to maximize the use of existing trunk infrastructure in central Kanpur, extension of water supply, sewerage, and drainage networks directly affecting proposed shelter components, and a sewer connections loan program; 14 (ii) Solid Waste Management (4%): improved facilities for the collection and disposal of refuse and night soil, together with workshop and depot improvements; (iii) Maintenance (3%): equipment and tools for improving the cleaning, repair, and maintenance of roads, sewers, and drains and (iv) Traffic Management (22): low-cost measures to improve traffic flows in areas adjacent to proposed shelter components and for enforcement of traffic regulations. (c) Technical Assistance and Institutional Strengthening (31): consultant and advisory services to carry out studies and training for state and local agencies. 2.23 Much of the focus in both operations was on low-cost shelter improvements. Together, sites and services and alum upgrading accounted for 68% of projected total costs of MUDP II and an even larger share (71%) of those for KUDP. MUDP II, although involving interventions in several subsectors (is.. shelter, road improvements, bus transport, solid waste management, as well as institutional strengthening), was less ambitious than its predecessor, which also contained specific components for small-scale business development, maternal and child health care, and water supply and sewerage. In part, this reflected difficulties encountered during the implementation of MUDP I and in part IDA/ Borrower decisions to deal with sectoral investment needs through independent operations (ie., the Madras Water Supply and Sanitation Project) and/or at the state level, as through the Tamil Nadu Integrated Nutrition Project (Credit 1003- IN, approved in April 1980 and closed in March 1989). 2. Proiect Coordination and Implementation Arrangements 2.24 As in the first operation, overall project coordination under MUDP II was entrusted to MMDA, 6 while implementation of individual components involved a host of other state and local government entities. The principal executing agencies were the Tamil Nadu Housing Board (TNHB) for sites and services, the Tamil Nadu Slum Clearance Board (TNSCB) for slum improvement, PTC for bus operation, the Department of Highway and Rural Works (DHRW) for road construction and drainage, and the Madras Corporation for solid waste management and maintenance. The Departments of Social Welfare (DSW) and Health (DOH), together with voluntary organizations, were responsible for health, nutrition, and child welfare interventions in sites and services and slum upgrading areas, while SIDCO was to play a similar role in terms of small business development. 7 Other sectoral agencies were involved provision of water and sewer, electricity, 6 MMDA was created in 1972 to coordinate the provision of services by local and state agencies within the Madras metropolitan area. ' At MMDA's request, responsibility for the provision of support to small businesses in the sites and services and slum upgrading areas was transferred to the Tamil Nadu Industrial Investment Corporation (TIIC) in November 1983. 15 street lighting, drainage, and educational services in connection with the sites and services and upgrading components, while the Collector of Madras of the Tamil Nadu state government was responsible for land surveys and acquisition. 2.25 In Kanpur, in turn, KDA a was responsible for overall project coordination at the local level. Unlike Madras, however, where MMDA's role was restricted to planning and coordination, KDA was also directly responsible for implementing the project's sites and services and slum upgrading components. KINM (solid waste management, maintenance, and traffic management) and KJS (water supply, sewerage, and drainage) were also involved in project execution, as was the Kanpur Electricity Supply Administration (KESA). A state-level steering committee was created during project preparation and continued to function during implementation to provide guidance on policy and project-related matters, while a Project Management and Monitoring Committee was set up to perform coordination, control, and problem solving functions. Educational activities in relation to child and maternal health and nutrition were carried out in slum improvement areas under a community development project sponsored by UNICEF. 3. Proiect Costs and Financing 2.26 Total cost of MUDP II, as appraised, was estimated to be Rs 73.9 crores, or US$ 87.9 million, including an IDA credit of US$ 42 million. As indicated above, slum upgrading (US$ 22.7 million), sites and services (US$ 18.1 million), and bus transport (US$ 10.8 million) were the largest components of the project in terms of cost. Including related off-site infrastructure and community facilities, core housing, and home improvement loans, the appraised average cost (excluding contingencies) of serviced sites and upgraded slum areas were US$ 1,215 and US$ 454 per household, respectively. 2.27 As appraised, total cost of KUDP was projected to be Rs 41.4 crores, or US$ 51.7 million, which was to be partially financed by an IDA credit of US$ 25 million. Sitea and services (US$ 17.3 million, excluding contingencies), slum upgrading (US$ 8.7 million), and sewerage (US$ 4.7 million) accounted for the largest shares of total project costs. Including associated ancillary costs, average cost (net of contingencies) per sites and services plot at appraisal was estimated to be US$ 1,128, while the corresponding figure per slum Lmprovement beneficiary household was US$ 435. 8 KDA was established by the Uttar Pradesh Urban Planning and Development Act of 1973 under the State Housing and Urban Development Department. 16 III. PROJECT IMPLEMENTATION AND RESULTS A. Time Delays 3.01 MUDP II suffered substantial implementation delays. The original credit closing date was extended twice, from March 1986 to March 1988, while physical completion took three years longer than anticipated at appraisal. Despite the rapid start of sites and services development on land already in the possession of the Housing Board and with respect to bus procurement, the project later ran into several major stumbling blocks. Land acquisition proved to be problematic for the rest of the sites and services component, as well as for road and drainage works. Construction management problems affected execution of both the sites and services and road components, while the need to remove squatters from drainage sites also led to delays. Construction of the Inner Ring Road sections took twice as long as originally anticipated. 3.02 Delays likewise resulted from IDA's suspension of disbursements for the project between June 1984 and May 1985. This occurred because the state government failed to approve the bus fare increases required in order for PTC to comply with a financial covenant contained in the Project Agreement. 1 Bus fare hikes were a politically sensitive issue and had already caused problems when IDA's first supervision mission discovered that the state government had rescinded 50% of the earlier fare increase enacted as a condition for credit negotiations. In addition, together with the fare revision delays, the single procurement of 550 buses and large increases in operating costs due in part to unexpected OPEC oil price increases created both a substantial additional debt burden for PTC and augmented its operating losses. 3.03 The suspension of disbursements for MUDP II is noteworthy both because it represents one of the very few occasions where the Bank Group has taken such an action in an urban development project 2 and because it did eventually lead to an improvement in the financial and operational situation of the metropolitan transport company. According to interviews carried out with company officials during the Audit mission, PTC recognized the need for the fare increases from the standpoint of its own financial health, ' but was unable to Section 3.09. Among other measures, this covenant required that the Government of Tamil Nadu should cause PTC to take all action necessary to ensure that its operating costs did not exceed 95% of its operating revenues. 2 The possibility that IDA might suspend disbursements over the bus fare issue was first communicated to the Tamil Nadu Government in April 1983, following the fifth supervision mission. The actual decision, however, was not made without some reluctance on the part of Bank management and the formal suspension of disbursements on June 1, 1984 came only after an informal suspension as of on February 1, 1984 had not produced the desired results. 3 A similar view was expressed by the former MMDA staff member (now a participant in the Project Management Group for TNUDP) who monitored the bus component of MUDP II and was the Authority's main liaison person with PTC. 17 enact them lacking state government authorization. It later viewed IDA's decision to suspend disbursements as having been an effective way of inducing state authorities to take a politically difficult action. State officials, in turn, were able to publicly point the finger at the Bank for requiring such an unpopular move. Even though supervision missions were not undertaken during the period when the suspension was in effect, execution of project components moved ahead, albeit at a slower pace, in the expectation that funding would eventually resume, and IDA staff maintained informal contact with project agencies, which generally understood the reasons for the Bank Group's decision. 3.04 The PCR (paras. 32-37) provides an excellent account of both the "success" and "constraining" factors and the risks associated with project implementation, especially execution delays. Because of the lessons they contain for on-going and future operations, these features are summarily listed below: (a) Factors of Success: the experience of the implementing agencies in MUDP I; 4 the continuity of staff and technical management in the executing agencies; and focusing responsibility for project administration and coordination in MMDA, which had performed a similar role in the first operation (and had benefitted from its institution building activities). (b) Constraining Factors: insufficient planning and advance action with respect to land acquisition; inadequate focus during project preparation on the detailed factors (eg. bid documents, contract slicing and packaging) which affect procurement efficiency; lack of state government support for the financial objectives of PTC and the need to reorient TNSCB away from slum clearance/tenement building and to develop a citywide shelter strategy; price and factor increases which were beyond the control of either the central or state governments; and IDA's not having insisted on the inclusion of private slums in the slum improvement component of the project. (c) Anticipated Risks (all of which, in fact, turned out to be problems during implementation): poor collections performance and cost recovery for slum improvement investments due to delays in providing tenure; weakness in the implementing capacity of the agencies responsible for the project's shelter components; and poor administration of estates transaction by TNHB which affected the rate of occupation of serviced sites by project beneficiaries. 4 The PCR observes (para. 30) that "the least successful aspects of the project were implemented by agencies who (sic] were newly-involved in MUDP II, whose involvement in MUDP I had resulted in little improvement in performance, or whose role in MUDP II was peripheral." An example was the arterial drains component, which constituted a very small part of the state's overall drainage program and, thus, was given "relatively low priority" by the executing agency. 18 (d) Inadequately Foreseen or Unexpected Risks: the impact of delays and inadequacies in fare increases on PTC's financial situation; land acquisition difficulties; construction and estates management problems for sites and services; the impact of oil price increases on PTC's costs and financial condition; and the single procurement of buses, which, while cost- effective in the short run, contributed to the company's financial problems described above. 3.05 The PCR for KUDP provides a much sketchier account of project implementation. Nonetheless, it indicates the principal causes of execution delays, which were shorter than those experienced in MUDP II, but occurred for some of the same reasons. Thus, while, like the second project in Madras, KUDP required an extension of the original closing date, this involved only one year, from June 1986 to June 1987. Unlike MUDP II also, implementation started much more slowly in Kanpur due to a combination of insufficient staff to prepare and award construction contracts and land acquisition problems. The PCR indicates (para. 13) that the water agency (KJS), in particular, was understaffed at the time project execution commenced. It also became necessary to substitute one of the three original sites and services areas, which was lost due to litigation, with additional land acquired at one of the other locations. Implementation of the sites and services component was further hampered by poor coordination with the local electric utility, KESA. This reportedly resulted in a significant delay in providing the connections required for operation of the sewage pumping stations at both areas. As anticipated in the SAR, finally, the purchase of "ahatas" also proved to be very problematic. 3.06 A review of project files reveals that there were a number of other persisting concerns and problems during KUDP's execution. Foremost among these were financial and institutional weaknesses both in the water agency (KJS) and the municipality (KNM), slow occupancy of sites and services plots, 5 and cost recovery difficulties, particularly for water and sewer connections, home improvement loans, and in slum upgrading areas generally. KJS' problems included its inability to comply with credit covenants 6 and to properly execute the sewer connections program. The latter difficulty led KDA to take over execution A May 1984 supervision mission reported that plots were not being occupied because of: (a) the cumbersome plot allocation process; (b) the lack of security at the project sites; (c) the need for time and financial resources on the part of beneficiaries to build or finish a minimum shelter; and (d) the lack of economic pressure on plotholders since they were not paying KDA. 6 Section 3.08 of the Project Agreement required KJS to introduce quarterly water meter reading and semi-annual billing starting in April 1981 and quarterly billing starting in April 1982, to develop its water meter repair and replacement capability, to introduce an effective collections program, and to generate sufficient revenues through operational improvements and, if necessary, tariff revisions to cover 100% of the operation and maintenance costs and debt service for its water supply and sewerage operations commencing April 1, 1984. KJS was not in compliance with the latter part of the covenant at project closing. 19 of the subproject in 1983, after which its performance improved considerably. KNM, in turn, suffered from problems in the areas of revenue generation, solid waste management, and infrastructure maintenance. However, rapid progress was reportedly made on a number of fronts, including the implementation of physical works and cost recovery from slum improvement beneficiaries, once supervision missions indicated that processing of the proposed state-wide follow-on project might be delayed in the absence of appropriate government action. B. Cost Variations 3.07 Actual costs of MUDP II at completion were 19% higher in local currency terms and 24% lower in US dollar equivalent terms than anticipated at appraisal. The higher local currency costs reflect the operation's longer-than- expected implementation period and the associated expansion of project physical works and acquisitions, particularly serviced plots, improved slum areas, and the number of buses procured. Higher-than-anticipated domestic inflation during the project execution period also played a role. Although the PCR does not specifically address this point, the difference between the local currency and US dollar equivalent costs on completion reflects devaluation of the rupee in relation to the dollar between 1981 and 1988. As a result of the underrun in US dollar terms, the fully-disbursed IDA credit actually accounted for 63% of total project costs, as opposed to the 48% expected at appraisal. 3.08 Comparing the participation of the main components in total project costs at completion with appraisal estimates, the actual share dedicated to transport and sites and services increased from 24% to 32% and from 32% to 37%, respectively, while that destined to slum improvement decreased from 38% to 25%. The increased share of total costs absorbed by the transport component is a direct reflection of the procurement of an additional 335 buses, or 55% more than the appraisal target, during the extended implementation period. The number of serviced plots developed also increased by some 22% over appraisal estimates, to about 18,300. The sharp decrease in the relative total cost share of the slum upgrading component, in turn, reflects the much lower expenditures than initially expected for home improvement loans and community facilities in slum areas, despite the increase in the number of beneficiaries by 20%, to some 60,000. 3.09 Total costs of KUDP at completion were 33% higher in local currency terms than anticipated at appraisal. They were 18% lower in US dollar terms, 9 for the same reasons indicated above in the case of MUDP II. Unlike MUDP II, however, the IDA credit's share of total project costs on completion was not ' KDA also assumed responsibility for implementation of the draina&e component on behalf of KNM. 8 The same tactic produced similarly positive results in Madras, as substantial improvement in collections from slum improvement beneficiaries under MUDP II was made a condition of negotiations for TNUDP. 9 The PCR does not present an actual cost figure in US dollar terms. The estimate used is the result of dividing total cost in rupees by the rupee/US dollar exchange rate in effect at the time of project closing (12.96:1.00). 20 significantly different from that anticipated at appraisal (52% as opposed to 48%). More than US$ 3.3 million was cancelled at the time of last disbursement, despite the state government's request that additional investments be financed under the project, because the follow-on UPUDP had already been approved. 3.10 As occurred with MUDP II, the relative participation of different components in total project costs on completion was substantially different from that expected at appraisal. While the share of both sites and services and alum improvement decreased, from approximately 47% of the total to 43% and from 24% to 10%, respectively, those of sewerage (13% at appraisal to 21% at completion), traffic management (from 2% to 8%), and drainage (from 4% to 8%) increased significantly. Not surprisingly given these figures, even though the number of sites and services plots increased marginally in relation to the appraisal target (14,892 versus 14.777), the number of slum households benefitted by KUDP fell sharply, from a projected 20,000 to roughly 10,400. This shortfall was primarily due to the legal difficulties in acquiring "ahatas" previously mentioned. As a result, only 69 -- as compared with an appraisal estimate of 87 -- "ahatas" were actually acquired, of which just 41 were improved under the project. On the other hand, despite a very slow start, the number of sewer connections made outside slum areas reportedly more than doubled, from the appraisal target of 10,000 to over 20,050. However, far fewer water connections were made (50,660 through July 1986) than initially anticipated (80,500). 3.11 In summary, the two projects share several important similarities. Both were considerably less successful carrying out slum upgrading activities than developing sites and services, while implementation was hindered by land acquisition problems in both cases. Efforts to improve slums on private land were particularly ill-fated. In general, non-shelter components fared better in terms of physical implementation (as reflected in actual cost shares) than shelter improvement, especially bus procurement in MUDP II and sewer connections, road and drainage works, and traffic management in KUDP. This is particularly evident in Kanpur, where shelter investments accounted for only 52% of total project costs on completion, as compared with the 70% anticipated at appraisal, while environmental sanitation, maintenance, and traffic management absorbed 45% of the total, rather than the 27% initially projected. Despite PTC's persistent financial problems and KJS's institutional weaknesses, in short, bus procurement and the installation of sewer connections proved easier to implement than slum upgrading, especially in privately-owned areas. C. Proiect Results 1. Physical Achievements and Environmental Impacts 3.12 MUDP II exceeded its targets for sites and services (22%), slum upgrading (20%), and bus procurement (55%). This was possible due to a two year extension of the project implementation period, made necessary by land acquisition delays and the suspension of disbursements, and "savings" of part of 1 Difficulties in ensuring adequate solid waste collection and municipal maintenance of the infrastructure and community facilities provided in slum areas further complicated this situation in both Madras and Kanpur. 21 the IDA credit due devaluation of the rupee in relation to the US dollar. However, while the number of serviced plots installed and slum households benefitted " under the project increased relative to appraisal estimates, the number of sheds for small industries in sites and service areas and of lease-cum- sale tenure agreements and home improvement loans in upgraded slum neighborhoods were well below appraisal projections. 3.13 Most road and drainage works were implemented as planned, albeit in many instances with sigrificant delays. Feeder storm drains built by the Madras Corporation, moreover, were reported (PCR, para. 64) to be "functioning inefficiently" as the result of inadequate maintenance, while delays in the drainage program implemented by the Public Works Department produced large cost overruns. In some cases, drainage works that were to be carried out under MUDP II still were unfinished at the time of the Audit mission (April 1990) as the result of persisting problems with squatters. Equipment for solid waste management, maintenance, and other municipal services was acquired -- again often with considerable delays -- but has reportedly not always been used effectively, while many of the findings and recommendations of consultant studies were disregarded, leading the PCR to conclude (para. 67) that "little long term benefit to maintenance and service appears to have been obtained from the project's support for the Madras Corporation." 13 3.14 Sites and services and slum upgrading investments have clearly introduced substantial physical environmental improvements in the neighborhoods immediately affected. Infrastructure improvements, together with the provision of land tenure, have also contributed to increased private housing investments in numerous slum areas, as residents have sought to replace precarious temporary dwellings with more permanent structures. Because of its uneven implementation record, however, the environmental impact of the drainage component has been mixed. Annual monsoon-related and other flooding has been sharply reduced, if not eliminated altogether, in those areas where planned drainage improvements have been introduced, but the delays and impediments mentioned above have postponed the realization of these potential benefits in other areas, adversely affecting perhaps as many as 20,000 families. Unresolved maintenance problems place future drainage benefits in jeopardy even in those places already served. 3.15 Even though the Audit mission was able to visit only a handful of upgraded slum areas (whose representativity in relation to the much larger n1 The actual number of slum areas improved under the project was 250, compared with the 221 expected at appraisal. 12 Only about 23,500 lease-cum-tenure agreements and 17,000 home improvement loans were finalized during the project implementation period, while the number of industrial sheds provided was just half that originally programmed. 1 In addition, as in MUDP I, both MC and the Madras Metropolitan Water Supply and Sanitation Board (Metrowater) reportedly "balked...at taking over the operation and maintenance of project facilities because of lack of funds, disagreements over engineering standards -- and in the case of Metrowater -- project areas being outside their existing operational area." 22 universe of such locations is unknown) and many of MUDP II's sites and services plots remained unoccupied, visual inspection suggested that sanitation and shelter improvements at these locations have been substantial and were being reasonably well maintained. The mission also visited both areas where drainage improvements (eg. the widening, realignment, and cleaning of drainage canals) had been completed and others -- often along the same channels -- where such works were still pending, with si3nificant differences in local environmental quality being clearly evident. In gene.al, however, it was apparent that, together with poor drainage and inadequate water supply, environmental sanitation -- especially solid waste and sewage collection and treatment and the reduction of pollution in the main rivers and canals that cross the metropolitan area -- remains one of the most critical problems facing Madras. 3.16 While KUDP substantially underran its targets for slum upgrading and water connections and only slightly exceeded that for sites and ser-ices, it more than doubled the appraisal target for sewer connections outside sites and services and slum areas. However, when the shortfall of slum sewer connections (roughly 9,300) is considered together with the increased number of non-slum connections (less than 10,100), the net increment is not nearly as impressive. In those slum areas which were upgraded under the project, the shelter and environmental benefits seem to have been equally great as those observed in Madras. According to the PCR (para. 17), "the slums have been transformed by the provision of paved roads and paths, small public squares, lighting, drainage, and water supply. The ahata dwellers have responded by steadily improving their properties by adding upper floors to increase accommodation, and by painting and decorating their homes." As in Madras, the Audit mission's visit to several "ahatas" confirmed that such benefits have, in fact, accompanied slum upgrading, although it was impossible to determine the extent to which the neighborhoods visited were typical of those affected by the component more generally. 3.17 The provision of "pay and use" community toilets and bathing facilities -- which in some cases replaced poorly maintained existing municipal toilets -- at various localities around Kanpur was similarly reported (PCR, para. 22) to have been a "very popular and successful" outcome of the project. These facilities are operated and maintained by a private entity, which is also responsible for cost recovery, thereby limiting the municipality's (ie. KMN's) role to electricity and water provision. The project also allegedly (PCR, para. 19) resulted in "marked improvements" with respect to solid waste management, although this could not be independently verified by the Audit mission. Perhaps even more importantly, KUDP reportedly improved methods of collection and disposal of night soil -- including the transfer of such wastes to a sewage farm rather than dumping them in local watercourses -- and reduced the number of households forced to rely on this system, with evident health and environmental benefits. Finally, the construction of new drains and the remodeling and desiltation of existing ones have reportedly had a significant impact in terms of reducing or eliminating flooding in various low-lying and predominantly low- income residential areas. As in Madras, however, environmental sanitation remains a Lerious problem both in Kanpur and in the Ganges valley more generally, where pollution control is one of the main objectives of the Uttar Pradesh Urban Development Project. 23 3.18 KUDP's traffic management component, finally, was credited (PCR, para. 23) with "being very successful despite the tremendous difficulties in catering to the variety of road users in Kanpur: pedestrians, handcarts, bicycles, rickshaws, motorcycles, tempos, taxis, buses, and trucks -- not to mention the large number of wandering cattle." " Although numerous traffic mana&ement improvements were reportedly introduced, lacking an ex-ante view, their effect on traffic flows was not apparent to the Audit mission. As further discussed below (para. 4.04), project road investments, however, have had a significant impact in terms of facilitating the city's recent expansion, as well as in improving physical access to KUDP's sites and services and other KDA housing projects, presumably also increasing land values, in the southern part of Kanpur, a consequence not mentioned by the PCR. 2. Rates of Return and Poverty Impact 3.19 Both projects appear to have been highly successful in terms of their re-estimated economic rates of return (ERR). While precise figures are not presented, the PCR for MUDP II argues (para. 83) that the appraisal 9RRs for both sites and services (28%) and slum improvement (13%) " were "easily achieved" since unit investment costs were not much greater (sites and services) or considerably lower (slum upgrading) than anticipated, while market rents for serviced land increased more rapidly than costs in a "sellers scarcity market." While no ex-post ERR is presented for the roads component, the PCR observes that the growth in vehicular traffic on sections of the Inner Ring Road built under MUDP I was greater than earlier projections and that this is likely also to be the case for those segments constructed under the second project. Visits by the Audit mission confirmed the heavy vehicular use of this road, the central part of which is presently being expanded from two to four lanes under TNUDP. Again, even though no specific figure is cited, the PCR affirms (para. 85) that the rates of return projected for MUDP II's bus component "are being realized and are likely to be maintained in the middle run." Rising labor costs, however, were identified as "a persistent threat" to the financial and economic rates of return of the bus component over the long run, despite PTC's continuing efforts to reduce its operating costs. 3.20 Similarly, re-estimated ERRs for both the sites and services (24%) and slum upgrading (30%) components in KUDP exceeded appraisal projections (18% and 26%, respectively). As in MUDP II, shelter benefits were estimated on the basis of imputed rental values of serviced sites and shelter improvements similar to those introduced under the project. No ex-post ERRs are presented for other components. " About 80% of all trips are made by bicycles and rickshaws, while motorized vehicles of all sizes (including scooters and motorcycles) account for only about 17% of the total and animal-drawn vehicles for the remaining 3%. 5 Section 4 in Part III of the PCR does cite a re-estimated ERR in the range of 25-30% for the slum improvement component, which is probably a direct reflection of the lower-than-expected actual unit costs for these investments. 24 3.21 The high re-estimated rates of return for the shelter components of both projects confirm the positive economic impact of land and low-cost housing development in rapidly growing large Indian cities, where the supply of both serviced land and adequate shelter is often severely constrained. This is further illustrated by the fact that demand for sites and services in both cities greatly exceeds supply -- reportedly by ten to one in the case of Madras -- and that, thus far, cost recovery performance on project-provided urbanized land seems to have been very good in both cases, even though many plots remain unoccupied for considerable periods of time after their initial allocation. In upgraded slum areas, the returns on public and -- where tenure is effectively transferred -- private investment also appear to be very high, reflecting the central location of the squatter neighborhoods affected. The high potential land values of improved slum areas situated on private land also helps to explain the difficulties faced by public sector housing agencies in attempting to acquire such areas both in Kanpur and Madras, since existing legislation limits the amounts which can be paid for such parcels to well below their market value, thus generating understandable reluctance to sell on the part of their owners. 3.22 According to the PCR for MUDP II (para. 43), "70% of the component beneficiaries allotted plots were in the 'Economically Weaker Sector' income groups, at or below the poverty line." During the Audit mission, however, OED was given a lower figure (55-60%) by MMDA. Visual inspection of MUDP I and II sites and services locations also left the impression that, in many cases, plot occupants were considerably above the poverty level, especially when compared with residents of both upgraded and, particularly, unimproved slum neighborhoods. Contacts with local NGOs suggested that many beneficiaries of project sites and services schemes were not from the EWS group, with one such organization indicating that various ways existed for interested parties to "get around" the Housing Board's income requirements (mainly by underreporting actual income). It was also observed that just because a serviced plot was "allotted" to an EWS family did not necessarily mean that this household would ultimately occupy the lot. Indeed, an evaluation study of one of the main sites and services developments under MUDP I reveals that a considerable amount of plot turnover and renting out of serviced sites and shelter units occurs over time. 1 These findings will be further explored below (see paras. 4.11-4.17). 3.23 The PCR for KUDP does not specifically address the operation's poverty impact, not does it comment on the income level of sites and services and alum upgrading beneficiaries. One of the reasons for this appears to be that, as in MUDP II, many of the serviced plots remained unoccupied at the time the PCR was written, even though most of these sites had been allocated and payments were being received by KDA. This continued to be the case at the time of the Audit mission. In the case of Madras, both MMDA and TNHB informed the Audit mission that plot occupnncy rates were lower for MUDP II than MUDP I at the 16 Kirloskar Consultants Limited, Report on the Study of the Arumbakkam Sites and Services Project, Pune, December 1987. This study was financed under the institutional strengthening component of MUDP II. The PCR for MUDP II (para. 81) acknowledges that "some selling out to.. .and infiltration of higher income groups into the plots reserved for lower income groups has occurred." 25 same stage of development, 1 a fact attributed to the greater distance from the city center of the sites developed under the second operation in relation to those installed under the first. Reported reasons for low plot occupancy in both Kanpur and Madras included problems of physical access, an initial lack of community (including commercial) facilities and on-site security, insufficient resources on the part of beneficiaries to proceed with construction -- despite the availability of materials loans 1 -- and possible speculation. 3.24 Despite these problems, based on the experience in both Madras and Kanpur. it is apparent that sites and services and (particularly) slum upgrading represent much more cost-effective ways of improving the shelter conditions of the urban poor than the approach traditionally followed in India involving the construction of often highly subsidized finished housing, whether for middle and upper-income beneficiaries or to accommodate slum dwellers who are forced to resettle. This lesson appears to have been well-learned in both project cities, even though KDA and TNHB continue to provide more costly solutions for non-low- income groups, while TNSCB argues that, because of their often precarious location along river banks, roadways, and rail lines, not all slums are suitable for upgrading. The relative importance that should be given to non-low-cost shelter solutions by public sector agencies, in short, remains a point of contention between IDA and local officials in Madras. " 3.25 Despite their contribution in terms of increasing the supply of improved shelter to low-income families in a cost-effective way, neither project was able to significantly reduce the existing housing deficit in a Quantitative sense. This is particularly evident in Madras, where the slum population is estimated by MMDA and TNSCB to be roughly 250,000 families 20 and new housing demand to be in the range of 30-40,000 units a year. In contrast, MUDP I and II together provided a total of some 30,000 serviced plots -- including roughly 20,000 for EWS -- between 1977 and 1989, or roughly 2,500 a year. While this represents a substantial increment of low-cost shelter relative to that supplied 17 At the time of the Audit mission, occupancy rates at the two oldest MUDP II sites, Mogappair East and West which were completed in 1988, were reported to be 40-50% and 30% respectively, while occupation of the newest developments at Manali I and II, was just beginning. 1 These were reportedly much in demand in sites and services areas, but considerably less so in improved slum neighborhoods, both in Kanpur and Madras. 1 NGO representatives interviewed by the Audit mission affirmed that local building contractors constitute a powerful lobby in Madras and that, in addition to a persisting strong "engineering bias," public agencies such as TNSCB prefer tenement construction because of the much larger contracts involved. 20 A profile of the slum population carried out in 1986 identified some 171,000 slum families located in 1,413 different areas and involving a total population of nearly 880,000 persons. According to local NGOs, however, when the large number of "pavement dwellers" is taken into account, in addition to the continued growth of the slum population over the past several years, the actual figure is considerably higher. 26 in the years prior to MUDP I, it still falls well short of meeting the projected need for adequate shelter in the MMA, which is conservatively estimated at some 15,000 new units a year. " Furthermore, even though slum upgrading contributes both directly (though tenure transfer and improved infrastructure, community facilities, and municipal services) and indirectly (by inducing slum dwellers to invest in home improvements) to enhancing the quality of the existing housing stock, neither KUDP nor MUDP I and II were able to effectively intervene in slums located on private lands, nor does this increase the quantity of housing made available per se. 3. Financial Performance 3.26 The financial performance of both operations was mixed. The problems encountered by PTC and JS in complying with IDA financial covenants for the bus component of MUDP II and the water supply component of KUDp 22 have already been touched on above. The two projects had similar experiences with respect to cost recovery from sites and services and slum upgrading investments. While cost recovery performance appears, thus far, to have been very good for the serviced sites components, 23 recovery of investment costs from slum improvement beneficiaries has been considerably more problematic. 24 The cost recovery rate for the sewer connection loans under KUDP was also reported to be comparatively low (59% in March 1988). 3.27 Attempts to enhance municipal revenues have also fallen short of appraisal expectations both in Kanpur and Madras. In the latter city, the PCR reports (para. 70) that, while the Madras Corporation adhered to the financial covenants of MUDP II, it was constantly short of resources for maintenance and services and, thus, was reluctant to assume such responsibilities in connection 21 The 30,000 figure was drawn from a MMDA pamphlet entitled Shelter Prozrammes in Madras prepared for the International Year of Shelter for the Homeless in 1987. 22 The PCR for. KUDP (para. 25) identifies "the willingness of KJS to increase its tariffs consistent with costs and to collect its bills" as one of the two main areas of "risk" with respect to project sustainability. It further observes that the agency's tariff structure had been unchanged since 1985 and that "KJS has not been able to collect arrears forcefully from nationalized industries and in turn owes (funds] to the power authority." Local officials informed the Audit Mission that this situation was the result of a court stay that impeded JS from collecting the payments due. 23 The Borrower's PCR for KUDP reports a cost recovery rate of 84% in 1988- 89 for the sites and services loans, while the corresponding figures for four of MUDP II's sites and services schemes varied from 78% to 90% as of February 1990, according to TNHB. 24 In Kanpur, slightly over a third of the roughly 2,700 slum improvement beneficiaries who had received tenure were in default in March 1988, according to KDA. In Madras, the amount recovered from project beneficiary slum dwellers was reported to be 55% of the total outstanding as of February 1990. 27 with the project. Furthermore, state government transfers of funds generated through the state sales tax surcharge were limited in amount and impact, while "there was little movement during MUDP II on property revaluation or tax increases in the Corporation." Largely as a result, and despite the purchase of additional equipment and the provision of technical assistance, inadequate coverage and quality of municipal maintenance and other services !1.O.ains a problem in both cities. 4. Institutional Development 3.28 Achievement of project institutional strengthening objectives was only partial in both cases. MUDP II succeeded in upgrading munio4pal accounting and management information systems, as well as in improving the administrative operations and financial planning of the bus company. New accounting and management information systems were also installed by TNHB and TNSCB as a result of the project. However, other goals, such as adoption of the recommendations of a route rationalization study by PTC and efforts to enhance MMDA's overall planning, capital programming and budgeting, and performance evaluation activities were largely unattained. The project's inability to reinforce MMDA's potential regional coordination role is attributed by the PCR (para. 77) to the relatively low level of "government interest in urban problems, policies and objectives." Despite these shortcomings, MUDP II sponsored useful studies on a range of topics in support of all major project agencies, especially MMDA and the Madras Corporation. 3.29 KUDP's institutional development goals were similarly ambitious and also only partially achieved. While a Traffic Engineering and Management Cell (TEMC) was successfully established in KNM and cost accounting, management information, and auditing improvements were undertaken within KJS, efforts to improve its metering, billing, and leak detection activities met with only relative success. Project-related recommendations to enhance municipal revenue generation by simplifying the Octroi tax and modifying administration of the property tax were adopted in part. 5. Policy Impact 3.30 The main areas where MUDP II and KUDP were expected to have an impact in terms of local development policy were: increased public sector provision of and cost recovery from basic urban infrastructure (eg. water and sewerage in Kanpur), transport services (Madras), and affordable shelter solutions (both cities); enhanced municipal revenue generation, maintenance, and service delivery; more efficient operation of local agencies (including the water utility in Kanpur and the bus company in Madras); and better urban planning and management generally. Both projects also sought to extend the approach followed to the state level through the preparation of multi-city follow-on operations. Since the experience with most of these concerns has been summarily covered in the preceding paragraphs, only one or two points will be further elaborated here. 3.31 As in many other Bank-supported urban development operations, while considerable progress was made with respect to achievement of project policy goals, in neither MUDP II nor KUDP were these objectives fully realized. The principal areas which proved problematic included the targeting of public sector 28 shelter investments, the strengthening of local government functions, and, for some components, cost recovery. As concerns shelter investments, RUDP had only limited success and MUDP II none at all in upgrading slums located on private land. Furthermore, while successfully introducing (KUDP) or consolidating (MUDP II) low-cost alternatives to public shelter provision/improvement, in both cities the relevant agencies continued to provide more costly solutions to higher income populations and/or destined to support slum relocation efforts. Thus, while the projects demonstrated that sites and services and slum upgrading constitute viable alternatives to dealing with urban shelter problems, IDA's effort to promote a citywide shelter strategy in Madras was largely frustrated. As a result, two of the main lessons drawn from the project experience by the PCR (para. 1.01), with which the Audit fully concurs, were: (a) once it is accepted that Bank-supported projects have passed the demonstration stage (as in Madras] and that the Bank's strategy is to support more effective urban programs as a whole.. .the whole effort of public shelter agencies should be directed towards meeting massive city shelter needs with the limited resources available; and, (b) it is also evident that the public sector, no matter how efficient, cannot by itself meet the shelter needs of even the lower income population in fast-growing cities; constraints on management capacity, financial resources and land availability dictate that the private sector must be enlisted in the interests of enhancing the supply of affordable serviced land, 25 particularly to lower income groups. 3.32 Strengthening the capacity of local governments to improve the financing and delivery of municipal services constitutes the second major area where MUDP II and KUDP attempted to have a policy impact at the metropolitan level. As in the case of the targeting of public housing investments on low- income beneficiaries, they were only partially successful. Two factors help to explain the limited ability of these operations to improve municipal service provision and administration more generally. One is the inherent financial, 25 The PCR affirms that "future projects should not contain a Bank-funded shelter component with acceptable objectives implemented by a public agency, while contrary objectives are being pursued by the same agency with its own funds." While the Audit agrees in principle with this restriction, it also recognizes that public housing agencies often have multiple sources of funding (eg. HUDCO and the "minimum needs program" in the Indian case) and potential clienteles and that, therefore, the most effective way to influence their investment allocation decisions is likely to be through appropriate policy changes at the state or national level. Such policy changes should focus primarily on modifying legislation and regulations which restrict access by the poor to serviced land and/or which inhibit a more adequate supply response by the private sector to the rapidly growing demand for urban shelter. 29 institutional and -- especially in the case of Madras n -- political weakness of local governments Per se. The second is that the municipal role in project planning and implementation was, from the very beginning, a secondary one, despite the conscious effort to reinforce municipal capabilities in key areas. 3.33 Considering the crucial role of local governments in assuring the sustainability of project benefits through the adequate maintenance of project- provided infrastructure in sites and services and slum upgrading areas, greater attention should have been given to these functions from the outset. Advance planning and agreement between sectoral executing agencies and municipal governments with respect to engineering standards and the services to be provided in project neighborhoods, together with their financial requirements and sources of funding, could have avoided some of the problems later experienced. In addition, as suggested by the PCR for MUDP II (para. 1.01), recognizing their inherent weaknesses, future urban projects in India should "address the whole system of municipal administration and finance in a state and avoid the piecemeal financing of projects in individual municipalities." 3.34 In short, while both MUDP II and KUDP were able to make important policy advances, neither fully achieved its objectives in this regard. Although MUDP II's policy goals were more ambitious than those of KUDP, the former had the advantage of being able to build on the substantially positive record of MUDP I and, thus, to attempt to consolidate gains already made and extend the agenda to such areas as citywide investment planning, shelter strategy formulation, and improvement of the general operations and allocation decisions of agencies such as TNHB, TNSCB, and PTC. As concerns these broader goals, however, its effectiveness appears to have been limited, especially among the two shelter agencies. KUDP ran into similar difficulties with respect to KJS and KNM. IV. POINTS OF SPECIAL INTEREST A. Proiect Impact on City Development 4.01 Both MUDP II and KUDP were concerned with increasing the physical and financial access of low-income families to urban services and improved shelter. A second major objective of both operations was institutional development. Neither explicitly aimed at guiding or redirecting the physical growth of the cities involved. Nevertheless, both projects appear to have had a significant impact in this respect. This is especially true in the case of Madras when MUDP I and II are considered together. 4.02 MUDP I and II both involved construction of parts of the Inner Ring Road (11.7 kilometers under the first project and another 5.8 kilometers under the second), together with other road improvements. These investments were not primarily aimed at improving the situation of the urban poor, but rather at 26 Although municipal councils are directly elected in much of India, Madras has been administered by a state-appointed special officer since 1974. 30 improving traffic circulation in and around the metropolitan area more generally. They appear to have been largely successful in this regard, and further improvements to the Inner Ring Road (IRR), -!.ncluding the widening of key central sections, are being carried out under TNUDP. In addition to facilitating traffic flows, these investments have helped to structure the physical expansion of MMA and provided improved access both to the various sites and services schemes installed under MUDP I and II and other major developments currently being developed by MMDA, especially the Koyambedu Wholesale (Food and Flower) Market Complex and the Sathangadu Iron and Steel Market. 4.03 Both of these markets are expected to help alleviate congestion in the center of Madras, as well as to provide more appropriate facilities for wholesalers and their clients. From the standpoint of the urban poor, however, both the trunk road improvements and the relocation of existing markets are likely to have an appreciating impact on land values on the metropolitan periphery, making access by low-income families more difficult and increasing pressures on less affluent residents or allocatees of nearby sites and services areas to sell or resell plots to higher income households. 4.04 Similarly, in Kanpur, the IDA project included road investments, which, according to KDA, have greatly facilitated the city's expansion to the south. Access was improved both to KUDP's sites and services schemes and to KDA's non-project financed conventional housing projects. These transport improvements have reportedly stimulated increased private housing investment in the southern part of the city. 4.05 In both Kanpur and Madras, therefore, project interventions appear to have had an impact -- perhaps a significant one -- on the direction, if not the rate, of urban growth, even though this outcome was not explicitly highlighted in the respective SARs. To some extent, this probably reflects differing priorities on the part of IDA (poverty alleviation), on the one hand, and KDA and MMDA (metropolitan development more generally), on the other, not to mention distinct institutional agendas of the various sectoral agencies involved. From this standpoint, both KUDP and the urban projects in Madras ultimately involved a combination of social and physical development goals at the metropolitan level, with different weights given to each set of objectives by different actors. The Bank Group's interest in combatting urban poverty notwithstanding, the two experiences suggest that the physical development aspects of its urban operations merit greater attention than they often receive and that this is all the more important when the Bank expects to finance a series of projects involving the same city or metropolitan area over time. ' From a different perspective, a somewhat similar conclusion was reached by Webber, Haines, and Vaidya, op. cit, in their review of the Bank's urban development operations in Madras. They affirm that "the Bank has focused on task-specific projects with bounded aims as the p,imary instrument of urban betterment in Madras...As an alternative to project-focused lending, we suggest that strategic Programs aimed at macro-development might yield greater long-term results... .These would call for multisectoral investments that promise to induce both mutually reinforcing outputs and autonomous enterprises." 31 B. The Role and Advantages of Follow-on Proiects 4.06 MUDP II was a follow-on project and both it and KUDP gave origin to ambitious statewide urban development operations, which are presently attempting to extend many of the features of the earlier projects to a larger number of citics in Tamil Nadu and Uttar Pradesh, respectively. For the most part, MUDP II was able to build on the lessons learned from the initial urban project in Madras, even though it was unable to overcome all of the institutional, political, and other obstacles that impeded smoother implementation and unambiguously positive institutional and policy results. Nevertheless, MUDP II was able to consolidate a number of innovations introduced in the first project, especially in the shelter sector, and to move forward on other fronts, as, for example, with respect cost recovery in slum improvement areas. 4.07 While their success is by no means guaranteed, follow-on operations such as MUDP II frequently possess a number of advantages over first projects in a particular sector and locality. One obvious advantage is the greater familiarity of both the Bank and the borrower with the other's priorities, procedures, and constraints. A second advantage is that undertaking two or more operations extends the time of Bank/borrower involvement in the sector and/or locality in question, often making it possible to more effectively address issues and problems, especially those involving institutional and/or policy change or sensitive political matters, which may require a longer time horizon to be effectively dealt with than is normally available during a single operation. This is particularly important in the case of urban projects where much of the emphasis needs to directed toward on-going management processes, including the operation and maintenance, as well as initial planning and financing, of basic services. 4.08 For both reasons, continuity of borrower agency decision makers and technical staff is important. This is clearly illustrated by the experience in Madras where one of the main reasons for the comparatively successful outcome of MUDP II was the existence of knowledgeable and highly motivated officials and staff in key agencies who had previously participated in the planning and implementation of MUDP 1. This factor, together with good IDA-Borrower relations at the technical and operational level, were also highlighted in a recent independent review of the Bank's urban projects in Madras. 2 4.09 Of particular importance in this context has been the presence of a relatively small number of Indian Administrative Service (LAS) officers, who have served as chief executives of the main project agencies, and of the higher level technical staff within these agencies, who have provided much of the continuity between MUDP I and II and the on-going TNUDP. This suggests that, beyond the 2 Ibid. Webber, et. al. conclude (paras. 1.1-1.2) that "the Madras and Tamil Nadu urban development projects have reflected a remarkable congruence of interests and styles shared by the World Bank and by the numerous public officials within the Madras urban region... .Further, the numerous Madras and Tamil Nadu governmental bodies have themselves joined into partnerships of shared programs, shared obligations, and mutual support, partnerships that might never have congealed had the Bank not intervened as broker." 32 possibility of implementing a series of operations in the same sector and locality, prospects for project success, especially in policy, institutional, and financial terms, are likely to be strongly influenced by the nature of the evolving relations among the different local agencies (and their staff) involved 3 as well as between these agencies and the Bank. 4.10 The only potential drawback to follow-on operations in the same sector and localities is their possible tendency to focus excessively on those components which are most successful, either not addressing or leaving by the wayside subprojects that are less manageable in a physical, financial, or institutional sense or which appear to generate less tangible or less immediate results. This has, in fact, been attributed to the Bank's urban operations in Madras by one recent outside assessment, which observes that "the array of product types were narrowed progressively from the first project initiated in 1977 to the third in 1988. Shelter projects were progressively expanded, while sectoral subprojects that were more directly related to economic development and human welfare received lessened attention and resources." ' Although essentially correct, this statement ignores the extent to which IDA has supported non-shelter interventions in Madras through other operations and that MMDA and other agencies are actively promoting local economic development through the market projects mentioned in paras. 4.02-4.03 above and other initiatives. C. The Filtering-Up of Shelter Improvement Benefits 4.11 The tendency for some share of project sites and services and slum upgrading benefits to "filter up" to families possessing higher income levels than the target population was mentioned in paras. 3.21-3.22 above. While this appears to have occurred in both of the operations under review, it is more readily apparent in Madras, where the earliest sites and services projects developed urder MUDP I have now been almost fully occupied for several years and provide a "preview" of what the situation in the MUDP II schemes is likely to be some five to ten years hence. It is, thus, of direct relevance to summarize some of the main findings of a MUDP 11-sponsored evaluation of the oldest MUDP I sites and services scheme, Arumbakkam. This project was also visited by the Audit mission, which found it to be a vibrant urban community possessing abundant commercial activity and seemingly well-integrated into the rest of the city. 3 The less happy experience with the Bank's urban development projects in Calcutta, as reported in Sanyal and Tewari (op. cit.), illustrates the kinds of problems that can occur when such relations are unstable over time and/or more antagonistic than collegial. The situation in Lucknow/Kanpur appears to have been much closer to that in Madras than to that in Calcutta, for the most part with similarly positive results. ' Webber, et. al., op. cit., para. 1.7 The authors also indicate that "some Madrasians we spoke with were especially critical of the large emphasis placed on shelter projects, contending that other sectors deserved more attention than they received." 5 Kirloskar Consultants, op. cit. According to MDA, 2,334 serviced plots were provided at this location, of which 1,721, or 74%, were reserved for EWS. 33 its residences appeared well-finished, suggesting that theit owners or ccupants had comparatively good incomes and/or were able to mobilize rable savings in order to complete their dwellings. The Arumbakkam evaluation study was based on a survey of plot ts undertaken in October 1986, at which time 100% of the serviced lots had Located and 83% were occupied. Among its general findings were: . EWS households were allotted 67% of the lots, as compared with a target of 74%; however, income of the majority of the EWS beneficiaries was concentrated near the upper end of the permissible income range; close to 5% of the total had incomes exceeding 1,500 rupees per month, or more than twice that of the EWS families, while 28% had between 700 and 1,500 rupees. . Monthly payments for serviced lots represented some 10-20% of household income in most cases; direct cost recovery was at about 95% of the target level, but indirect cost recovery (ie. water tariff and property tax payments) was unsatisfactory. . Slightly over 20% of the lots had been transferred, including 25% of the sites allocated to the lowest income families, but only 13% of those allocated to the highest income group. . There was an improvement in income levels on the part of all groups; 28% of the participants had supplemented their incomes by rentals, of which roughly one-third were living in the project area and two-thirds elsewhere; about 570 project participants were employed in industries and informal sector businesses in the area. . The quality of housing was high, with 90% of the dwellings estimated to have a life of over 15 years; most of the houses have walls built with burnt brick and concrete flooring, while there were only 38 huts in the project area. . Although a substantial majority of the EWS beneficiaries reported improved access to water supply (77%) and employment (63%), less than half (42%) indicated improvement in environmental conditions after coming into the project area; while health conditions reportedly improved for most residents, educational, social and recreational facilities had not. Another result of the study was that plot allocation procedures were Dlproof," as there was a clear indication that some plots had gone to income families whose declared incomes had not been adequately cross- at the time of allotment. There was also evidence that subsidized EWS 3re being sold to households who were "far better off than the target Lon and deserving no subsidy from any source." It was likewise found ren though project planners expected most beneficiaries to use self-help :tion techniquea, the majority of the occupants preferred to hire labor, 34 in many cases because household noads could not afford to take time away from their jobs or possessed insufficie-t skills. 4.14 These findings suggest that possession of a sites and services plot may be an important potential source of upward mobility for many families due to better access to employment, the use of part of the dwelling for income generating activity, the possibility of renting out part or all of the associated housing unit, and, in some cases, transfer of the (frequently subsidized) lot to another family, presumably at some profit. However, they also suggest that a tendency indeed existed for many plots originally allocated to low-income households to end up in the hands of higher income families either through rentals or transfers which, by the time of the evaluation study, had affected nearly half of the units developed at Arumbakkam. 4.15 As noted in para. 3.21 above, a tendency also exists for the expulsion of some low-income residents from alum lots once the areas where they are located have been upgraded and land tenure transferred to their occupants. To quote the PCR for MUDP II (para. 61): "there is evidence that particularly on larger plots, where there were rented houses, a significant proportion of.. .low- income inhabitants were evicted by original squatter occupiers...(many of whom were also low-income). The original occupiers were enriched by the enhancement of values on newly-tenured serviced plots at the expense of their evicted tenants, who presumably were left worse off than before the program was implemented." Discussions with TNSCB, TNHB, and MMDA during the Audit mission confirmed the existence of this tendency to an unknown extent, as well as the complexity of land occupancy in many slum areas, due to the subdivision and subletting of plots and dwellings -- and in some cases both -- among two or more related or unrelated households. ' This situation is further complicated by the fact that many plots are "owned" by former squatter families who no longer reside on the lots themselves, but have rented them out ro other households, who, in turn, may have sublet part of the area or structure they presently occupy. 4.16 Under these circumstances, establishing which family or families should receive tenure and/or how existing plots should be legally subdivided among two or more resident households is by no means an easy matter. According to TNSCB, this difficulty is one of the main reasons for the cost recovery delays and collection problems experienced in connection with this component in both MUDP I and II. 7 The same types of problems, although perhaps on a less dramatic scale, also appear to have affected the slum upgrading component of KUDP. In any event, just as in the case of sites and services, slum upgrading cum tenure provision does clearly result in increased land valuet and improved 6 The 1986 survey of slum dwellers in the MMA mentioned above found that each shelter unit contained an average of 1.3 families, meaning that one in every three dwellings was shared by more than one family. In Madras city, however, where two-thirds of all metropolitan slum dwellers are located, this ratio was even higber: 1.4 families per slum shelter unit. 7 An alternative and at least equally plausible explanation put forward by the PCR (para. 58), however, was the lack of political and administrative support for an adequate collection effort. 35 environmental and shelter conditions in alum neighborhoods, in many cases generating tangible benefits for those who are ultimately able to own and/or reside on these plots. a Due to an upward "filtering" process in terms of income levels, in short, not all of the ultimate beneficiaries are among the groups originally targeted by the projects, while some intended beneficiaries in slum areas may, in fact, have been displaced as a result of these operations. 4.17 One likely explanation for this phenomenon is the general scarcity of urbanized land and adequate shelter in Madras -- as in other large Indian cities, including Kanpur -- which makes sites and services solutions and newly- tenured upgraded slum plots attractive to higher income groups despite their generally small size. This scarcity, in turn, appears to be due to a combination of inadequate provision of basic urban infrastructure and services and an insufficient public and private sector supply of affordable housing. These deficiencies are directly associated with insufficient public sector resource mobilization, on the one hand, and an existing legislative and regulatory framework which introduces significant distortions into local urban land and housing markets, on the other. As the experience in Madras illustrates, such distortions have the effect of impeding sites and services and slum improvement investments from fully benefitting their intended target populations and/or, in the case of sites and services, limiting access within the lower-income part of the target group to families who are comparatively better off. D. Sustainability 4.18 The sites and services components of both operations are likely to be sustainable, given their good cost recovery record and visual evidence that occupants have mobilized additional resources to expand and improve their housing units. The findings of the Arumbakkam evaluation reported above also suggest both that the solutions are affordable to their present occupants and that residents' incomes have increased over time. The only question mark concerns municipal maintenance of project infrastructure and the provision of community services. The former appears to be much less problematic in sites and services areas than in upgraded slum neighborhoods. Even in the latter, based on the situation in the areas visited by the Audit mission in Madras and Kanpur, many newly-tenured residents seem to have further improved the condition of their dwellings, and it appears likely that the benefits of project investments will be largely sustainable unless there is a total breakdown in service delivery and maintenance. Additional community development work and increased cost recovery would nevertheless enhance the likelihood of proper maintenance and service provision in improved alum neighborhoods, as would better coordination between executing agencies and the respective municipal corporations during the planning and implementation stages of future projects. 4.19 The bus transport component of MUDP II is also likely to be sustainable in the short run, given the good operating performance of PTC and its awareness of the need to adjust tariffs in order to cover increasing operating and other costs. However, its longer-term sustainability is less certain, 6 TNSCB reported to the Audit mission that, as of March 1990. at least 1,858 slum lots upgraded under MUDP I and II had subsequently changed ownership. 36 especially if and when the Bank Group eventually decides to withdraw its support (which, for the moment, is continuing under TNUDP). The PCR (para. 86) highlights the persisting pressures of labor costs on the company's operating budget and its longer-run financial and economic rate of return. PTC and HMDA officials indicated to the Audit mission, however, that, given the political influence of the labor unions with which most of its employees were affiliated, attainment of further reforms, as well a3 the adoption of more frequent tariff revisions, was not likely. This view was reinforced by the observation that, in the absence of IDA's pressure on the state government, it is highly unlikely that bus fares would have been raised to the extent they already have and, thus, that PTC's financial situation would be as good as it is when compared with that of similar companies in other major Indian cities. 4.20 In the same vein, sustainability of project investments under the water and, especially sewer, components of KUDP is uncertain. As noted in para. 3.26 above, the willingness of KJS to increase its tariffs for sanitation services and to improve the collection of outstanding bills, so as to strengthen its financial performance more generally, is questionable, despite IDA's insistence regarding the need for such measures in connection with both KUDP and UPUDP. The other problem identified by the PCR (para. 24) as affecting the sustainability of KUDP's sewer component consisted of possible blockages due to siltation in the trunk severs into which effluents from the newly-connected residences were to be discharged. However, the PCR expressed confidence that increased flows into the system from the r.ew connections and adequate future maintenance made possible by "new sewer cleaning equipment and an ongoing program of sever cleaning" would reduce this risk to an "acceptable level." The Audit mission was not able to determine whether this, in fact, was the case, although it did find that cost recovery for sewer connection loans was poor because of the apparent unwillingness of beneficiaries to pay and that KJS's accounting system was unable to quickly flag defaulters. V. THE ROLE OF THE BANK 5.01 The Bank Group's role in both operations was largely positive. IDA's concern with and insistence on the importance of project institutional goals and financial covenants appears to have been crucial both to the results achieved with respect to the former and eventual borrower compliance with most of the latter. Despite considerable reluctance of executing agencies and/or state and municipal governments in some instances to undertake needed administrative reforms and/or adjust or enforce tariffs or other charges, IDA supervision missions -- with the added leverage provided by the suspension of disbursements for MUDP II -- were able to induce project entities to improve their financial and institutional performance over time. The strong incentive of funding for proposed follow-on operations (ie. TNUDP and UPUDP) appears to have been a key 9 See, for example, the PPAR for the Calcutta Urban Transport Project (Credit 1033-IN), OED Report No. 7912, June 29. 1989, and the PCR for the Bombay Urban Transport Project (Loan 1335-IN), OED Report No. 6239, June 4, 1986. 37 element in this, together with consistent IDA staff attention to these areas during project preparation, appiaisal, and supervision. 5.02 Despite these contributions, both PCRs touch on certain inadequacies with respect to Bank supervision which should be briefly mentioned. In the case of MUDP II, the PCR affirms (para. 88) that "it was difficult to supervise certain aspects of the project with the IDA budget and the state government management arrangements and information systems available." Several of the smaller components received insufficient attention during supervision, especially small-scale enterprise support, education, health and nutrition programs that were being undertaken in project-funded community facilities, and the home improvement loans part of the slum improvement component. Crucial community organization and development work seems to have been given inadequate emphasis in the execution and supervision of both operations. In addition, the PCR for MUDP II observes that both the quality and the implement&'ion of project- sponsored civil works would have benefitted from a greater input on the part of IDA engineering staff during supervision, especially in the area of sanitation. One reason why these areas received less attention than would have been desirable appears to have been the large amount of IDA staff time that had to be spent in connection with the bus fare revision problem and the eventual suspension of disbursements. 5.03 In the case of Kanpur, the PCR suggests that IDA supervision activity should have been more intensive during the first two years, rather than in the middle, of the project execution period, as, in fact, occurred. The increased supervision effort during 1983-84 was attributed by the PCR (para. 28) to IDA's "recognition of the slow progress being made in the early phase of the project and the strength brought to bear to effect a successful correction." While KDA and other project agencies appear to have appreciated IDA's supervision visits, there seems to have been a feeling that such missions or extra-mission contact with local project staff could have been more frequent. IDA's "helpful visits" are nevertheless specifically credited by KDA with improving allocation procedures for sites and services plots. On the other hand, KDA felt that project works were defined too specifically in the credit's legal documents, making changes difficult and leading to the suggestion that a more flexible approach should be adopted in future operations. 5.04 One interesting and important feature of IDA's involvement in both operations, but particularly MUDP II, was the continuity of the technical staff involved. KUDP was prepared, appraised, and largely supervised by a municipal engineer, who worked together with the same institutional-financial analyst during most of the implementation phase. However, different urban development specialists participated in the pre-approval (architect/urban planner) and mid- implementation (engineer/urban planner) stages of the project, respectively. 5.05 MUDP II was identified, prepared, appraised, and supervised throughout by the same task manager (an economist), who also identified, appraised, and continues to supervise TNUDP, covering a period of well over twelve years. This truly remarkable continuity in terms of the involvement of an individual staff member with operations in a particular sector in the same city is probably unique in the Bank Group's experience and appears to OED to have been a key contributing factor in MUDP II's comparatively successful performance, 38 despite -- and partially because of -- the suspension of disbursements. This aspect was highlighted in many of the discussions held by the Audit mission with project agencies in Madras, especially MMDA, in which it was affirmed that the presence of the same project officer greatly facilitated relations with IDA even during the tense period when both disbursements and supervision missions were suspended. 5.06 Especially given the institutional complexity and political sensitivity of most "integrated" urban development projects such as those in Kanpur and Madras, maximizing the continuity and consistency of Bank staff orientation over time is essential. Both in terms of building and consolidating relations between local agencies, officials, and professionals and the Bank Group and in terms of truly understanding the context in which Bank/IDA urban operations are being undertaken, efforts should be made to avoid dramatic shifts in task managers and other project personnel. KUDP similarly illustrates how a stable team of two or three IDA staff members or consultants possessing complementary skills can enhance the effectiveness of project supervision. On the more negative side, hr,ever, MUDP II and KUDP confirm the tendency witnessed in other Bank urban projects to focus largely on the "big ticket" components, particularly shelter and physical infrastructure (including their important institutional and financial aspects), but to give inadequate attention to smaller components, particularly those of a social service or directly productive nature. 5.07 A second shortcoming of the Bank's approach to both MUDP II and KUDP was the tendency to underestimate the extent to which the existing policy, regulatory, institutional, and political environment in which these projects were undertaken would restrict the effectiveness of the proposed investments in terms of their basic poverty alleviation goals -- as, for example, by hindering the improvement of slums located on private lands in both cities and the possibility of achieving a significant reorientation of TNHB's and TNSCB's expenditures in Madras -- and contribute to the eventual filtering of part of their benefits to non-target groups, as in the case of some sites and services and slum improvement plots. More concretely, the Bank appears to have failed to perceive the extent to which national and state laws and regulations affecting urban land development and housing supply, such as the land ceiling, slum improvement, and rent control acts, have themselves contributed to the situation of scarcity of serviced land and adequate shelter in cities such as Madras and Kanpur. This suggests the need for the Bank to address such policy and other distortions more directly in connection with its future lending in the sector. VI. CONCLUSIONS AND LESSONS LEARNED A. Conclusions 6.01 Despite -- and, in fact, largely because of -- significant execution delays, MUDP II exceeded most of its physical targets, particularly for sites and services, slum upgrading, and bus acquisitions. The project also achieved many, if not all, of its financial targets, although a one-year suspension of 39 disbursements was required in order to bring about needed bus fare increases. However, small-scale business support and municipal strengthening activities did not fulfill their aims. The drainage component likewise remained unfinished at the time of the Audit mission, while some of the drains installed under the project were functioning poorly due to lack of maintenance. 6.02 Following closely in the wake of the first urban project in Madras, MUDP II was able to further demonstrate the socio-economic and financial viability of the sites and services and slum upgrading approaches to shelter improvement. Like its predecessor, however, the second project was unable to convince state officials to reorient public shelter investments exclusively to the benefit of the urban poor. On the other hand, MUDP II did contribute to continued human resource development in many of the participating agencies, helping them to garner a sufficiently large contingent of experienced personnel to permit the state to undertake the multi-city Tamil Nadu Urban Development Project, which is currently under implementation. 6.03 Other positive aspects of MUDP II include physical environmental betterment in sites and services and slum upgrading areas, major road improvements, and the fact that a large majority -- perhaps as many as two-thirds -- of the direct beneficiaries of the shelter components were at or below the poverty level. Many of these families appear to have experienced substantial increases in income and wealth subsequent to taking possession of their urbanized lots or receiving tenure in slum areas. Some benefits from project shelter improvements have, nevertheless, "filtered up" to higher income groups. Finally, even though MUDP II's sector policy and institutional development results were mixed, various agencies appear to have been considerably strengthened as the result of their participation in the project. On the other hand, IDA's attempts to induce local agencies to take a citywide approach to shelter supply, enhance the role of municipal authorities, and improve metropolitan planning and investment programming were largely frustrated due to insufficient political support at the state government level. 6.04 In short, this was a generally satisfactory project, most of whose benefits are likely to be sustainable, even though staffing and wage problems represent a continuing threat to the longer-term financial health of the bus company and a need persists to improve the maintenance of upgraded slum areas and drainage canals. On the whole, both the Borrower's and IDA's performance were good despite the need to suspend disbursements in mid-stream and the fact that implementation and supervision of some project activities were adversely affected by poor agency and state government management and IDA budget constraints, respectively. The continuity of technical and managerial staff in the principal project agencies appears to have permitted the state government to react reasonably quickly and effectively to most problems -- bus fare revisions being the principal exception -- and to introduce needed changes in execution arrangements. The presence of the same IDA task manager from identification through completion also provided a rare opportunity for a single staff member to influence project design and monitor implementation from start to finish, with positive consequences for IDA-borrower relations in general. 6.05 KUDP was also, for the most part, a successful operation. The target for the project's largest component (sites and services) was slightly exceeded, 40 although that for slum upgrading was not achieved due to difficulties -- which had been foreseen at appraisal -- in purchasing land from private "ahata" owners. As a result, the slum improvement component benefitted less than 60% of the original target number of households and its share of actual project cost fell from an expected 24% to less than 102 of the total. Those slums which were upgraded have been provided with paved roads and paths, small public squares, lighting, drainage and water supply, and "ahata" dwellers benefitting from the project have responded by further improving their own dwellings, in the process helping to stimulate demand for slum upgrading elsewhere in the city. Public acquisition of slums on private lands nevertheless remains problematic and, thus, places a constraint on future slum improvement activity in Kanpur and other Indian cities. 6.06 With the exception of slum upgrading and residential water connections, physical targets for project components were largely met and substantial improvements have reportedly occurred with respect to storm drainage and solid waste management, producing environmental and public health benefits. The sewer connections program was implemented rapidly following a slow start and a change in the responsible executing agency. The provision of privately- operated community toilets and bathing facilities, the creation of an effective traffic engineering and management cell in the municipal corporation, and the parallel execution of a UNICEF-sponsored nutrition and health program in slum areas all contributed to the project's positive social and environmental impact. As in MUDP II. it appears likely that benefit flows will be sustained for most components, although, as in Madras, municipal maintenance remains a problem, as does cost recovery for the sewer connections program and on the part of the water utility more generally. 6.07 While project policy and institutional development objectives in Kanpur were less ambitious than in Madras, KUDP did give special attention to strengthening the operations and financial management of the local water utility, KJS, achieving mixed results. Although there were some start-up difficulties and inter-institutional coordination failures in relation to the provision of electricity to new sewage pumping stations and tube wells at two of the sites and services areas, KDA, the state government steering committee, and the "project monitoring cell" in Lucknow seem to have been capable of handing most implementation problems as they arose, as well as taking a leading role in preparation of the follow-on Uttar Pradesh Urban Development Project, which is now under execution. IDA-Borrower relations were reported by all concerned to have been good, although, in retrospect, IDA supervision should have been more intensive during the first two years of project implementation when land acquisition, agency staffing, and procurement problems were the most serious. B. Lessons Learned 6.08 As the objectives, basic approach, implementation experience (aside from the suspension of disbursements for MUDP II), and initial results of the two operations under review were largely similar, it is possible to generalize many of the lessons that can be drawn from them. Aside from those contained in the respective PCRs, the main lessons from these projects are the following: 41 (a) It is necessary to take the legal, regulatory and institutional frameworks and their implications for the supply of urbanized land and housing more clearly into account in designing future urban development operations in India and elsewhere. In addition, such projects should directly seek to reduce the distortions generated by these non-physical constraints on the provision of urban services and shelter. In this connection, it is likely that statewide programs such as TNUDP and UPUDP can be more effective than projects targeted on individual cities, even very large ones such as Madras and Kanpur. (b) Both operations reveal the need for a much better understanding of the intricacies and dynamics of land and shelter occupancy, informal "tenure" relationships, and social organization more generally in slum areas. This suggests the need for greater ex-ante familiarity with the particular slum areas to be upgraded. There is also a need to develop more agile (and market-sensitive) mechanisms for public sector land acquisition and improvement-cum-tenure transfer to occupants of privately-owned slum areas. (c) Similarly, there is a need to have better knowledge as to what happens to the initial occupants of sites and service plots and slum properties once they have been tenured and improved. In particular, it is important to gather information with respect to the lowest income beneficiaries/occupants of such areas and the extent to which, over time, they may be displaced -- voluntarily or otherwise -- by more affluent households. (d) Both projects demonstrate that there is often a considerable time lag between the physical completion of sites and services plots and their allocation to specific families, on the one hand, and home construction and the actual occupation of such lots by these households, on the other. The reasons for this need to be better understood and measures taken to accelerate the process in order to minimize the risk that project infrastructure investments will be underutilized. (e) The experience with both operations likewise suggests that interventions that are mainly concerned with shelter improvements (and bus transport in the case of MUDP II) are not the most appropriate vehicle to bring about major changes in the organization, financial management, and service quality of municipal corporations such as MC and KNM. Such objectives are likely to be more effectively achieved through specific municipal development, rather than multi-component "integrated" urban development, projects. When municipal authorities are to participate in the latter type of operations, however, they should be involved in an active way in project preparation and implementation from the very beginning. 42 (f) Independently of the type or degree of involvement of local governments in future operations, the planning of urban projects should give greater attention to institutional coordination and, especially, post-construction maintenance of project-related infrastructure and facilities. Where necessary, specific institutional and financial provisions should be made to ensure that such coordination and maintenance will be forthcoming. Beneficiaries should also have a more direct and active role in operating and maintaining project-provided services and infrastructure, and community development activities in such areas should be undertaken to this end. (g) Despite the fact that it was not fully successful with respect to its "software" objectives, MUDP II clearly illustrates the advantages of a follow-on project in terms of consolidating gains made under an earlier operation and attempting to move forward on the often difficult sector policy and institutional development fronts. The experiences in Madras and Kanpur reveal that the time horizon required to achieve project "software" objectives may be considerably longer than that needed to attain physical targets and, thus, that it is both appropriate to set such goals in relation to a series of investment projects and probably unrealistic to do otherwise. (h) Both operations likewise demonstrate that the prospect of Bank Group financing for future projects in the same sectors and localities and/or involving the same executing agencies can provide considerable leverage with borrowers in terms of inducing their compliance with financial and other covenants, at least in the short run. (i) MUDP II and KUDP also illustrate the importance of continuity of qualified and motivated technical and managerial staff among coordinating and executing agencies for successful project performance. Such continuity facilitates the creation of working "alliances" among key agencies and their personnel in project planning and execution. (k) Finally, MUDP II reveals the advantages of strong continuity of project task management within the Bank. This is especially important for "integrated" urban development operations, which have multiple objectives and components, involve politically sensitive institutional and policy issues, and focus on complex social environments such as urban slums.