Taking a "synergistic" approach to metropolitan development has strong appeal. As the Bank expanded its urban lending in the late 1970s and early 1980s, it undertook a series of integrated operations in hopes that these would have a greater impact on urban living conditions and incomes than separate projects in subsectors such as housing, small industry, or transport.
Experience has shown that the advantages of integrated projects are very hard to capture. Recent OED audits in three countries illustrate the limitations, problems, and risks associated with integrated urban projects. Such projects depend heavily on a supportive macroeconomic, policy, institutional, and local political environment. They have rarely accomplished their goals of institutional strengthening and better financial management in agencies providing urban services. They require firm commitment by local authorities as well as strong, consistent, and multi-faceted Bank involvement, including very intensive supervision. Recognizing these difficulties, the Bank now again directs most of its urban lending to individual subsectors.
Interventions work better if they are selective and closely focused, but they need to be chosen and planned in the light of an overall plan for urban management and improvements. Dealing with the interface among the discrete interventions is still a major challenge. The review of experience with integrated operations yields some practical recommendations for planning and implementing current projects.
Most of the Bank's urban development lending approved in the 1970s and early 1980s was for low-income housing or urban transport projects, but roughly a third (or a fourth of the number of projects) was for multisectoral, or integrated, projects.
Integrated urban projects view urban and regional problems as those of a single system. Compared with most urban projects supported by the Bank, these are complex projects with ambitious institutional goals. Typically, they contain components for shelter, infrastructure, transport, solid waste management, business support, health, nutrition, and education. When introduced, integrated projects were thought likely to have a greater impact on urban development than discrete projects with a subsectoral focus. Costs of infrastructure and transactions were expected to be lower, and implementation delays shorter, in these projects than in subsectoral projects.
The four projects audited were typical of integrated projects in single large cities or metropolitan areas. All focused on poverty alleviation and had broadly similar goals and approaches.
The Guayaquil Urban Development project, the first of the four to be approved (see table), was the first project for Ecuador's largest city (population then 1 million), and the first urban project in that country. Designed to initiate large-scale programs of housing development and job creation to benefit low-income people, it contained components for sites and services and slum upgrading, support for smallscale enterprises, and technical assistance. As later reformulated, it also helped to rehabilitate flood-damaged urban infrastructure.
The project's main objective, however, was institutional and policy development. This was to be promoted in the course of expanding credit and technical assistance to small enterprises, strengthening of municipal efforts to serve low-income areas, and building up the experience of the national housing bank in financing shelter for low-income people.
The two projects in India consisted mainly of shelter improvements but also contained components for urban transport (Madras), traffic management (Kanpur), basic sanitation (Kanpur), solid waste, municipal maintenance, and institutional strengthening (both).
The Second Madras Urban Development project was designed to help India's fourth largest city (population then 5 million) to further develop and expand the low-cost solutions to urban problems begun under its predecessor. It supported the continued re-orientation of shelter and infrastructure investments in the Madras Metropolitan Area, to make them more responsive to the needs of the urban poor. It sought to strengthen the institutions responsible for shelter, infrastructure, and transport, to encourage greater emphasis on cost recovery, and to support effective metropolitan planning and capital programming and budgeting.
The Kanpur Urban Development project was designed to help India's eighth largest city (1.7 million inhabitants in the early 1980s) to increase the supply of serviced land to the poor by providing low-cost serviced plots for housing and small businesses, and to address the worst environmental conditions in privately-owned slum areas in and around the city's core. Other goals were to strengthen key urban institutions, develop a strategy for state-wide urban development, and obtain full cost recovery from the project's shelter components.
The Recife Metropolitan Region project was the most ambitious of the four. Recife is Brazil's fifth largest city (population 3 million in 1980). This innovative project contained components to:
- improve housing, urban infrastructure, and services within the existing urban fabric;
- provide infrastructure and services to support, complement, and induce desired patterns of metropolitan development;
- enhance income generation, especially for the urban poor; and
- strengthen the institutions responsible for the planning, administration, and financial management associated with urban and metropolitan region development.
Each of the projects relied on several different executing agencies and had complex implementation arrangements involving at least two levels of government. In Recife, the extreme case, the project involved 183 activities and 33 federal, state, and local agencies, including a dozen separate municipalities. In Madras, Kanpur, and Recife, overall project coordination was the responsibility of a metropolitan development authority. In Guayaquil this role was divided between the municipality and two central government housing agencies.
All four projects experienced considerable delays (on average 1.8 years) and other implementation difficulties. Implementation problems were most serious in the two Latin American projects. In both cases, local politics and institutional rivalries, whose importance had been discounted during appraisal, played a major role, as did deteriorating macroeconomic conditions, which adversely affected counterpart funding.
In the Guayaquil project, implementation of almost all the components under municipal responsibility proved problematic. The municipality never reached agreement on cost recovery with the beneficiaries of the upgrading projects. Funding and contract execution problems for a sites and services scheme caused long implementation delays, prompting neighboring families to invade the plots and leading the municipality to abandon the works.
The design of the small-scale enterprise credit program proved too complicated. Rediscount rates set by the central bank were too low, discouraging the participation of private intermediaries. As a result, both the slum upgrading and small business components were eventually dropped. The project was reformulated, reducing the municipality's role, expanding the serviced sites component executed by the national housing institutions, and adding flood reconstruction works. Even so, it achieved few of its policy, institutional, or financial goals, including the strengthening of municipal finances and service delivery and cost recovery from final beneficiaries.
In Recife, macroeconomic instability, shortages of counterpart funds, and political-administrative and institutional changes at all three levels of government adversely affected project implementation. Being both multi-institutional and place-specific, the project proved vulnerable to administrative changes and political rivalries. During the implementation period Recife had a succession of six governors and at least five mayors, as well as seven superintendents of the project coordinating agency; numerous changes also took place among the mayors of other metropolitan municipalities, state and local planning and finance secretaries, and heads of project executing agencies. Each transition, especially when different political parties were involved, brought at least some changes in personnel, procedures, and/or priorities, affecting the pace and sometimes the direction of project execution.
The project was revised twice and several of its original components were either never undertaken or substantially altered during implementation. Other components were only partly executed or proved to be of little lasting economic value, even though some of the infrastructure investments and many of the slum improvement works resulted in tangible benefits to low-income people. The project's financial and institutional accomplishments were negligible.
Ultimately, the project was the victim of its own complexity. Its innovative and ambitious design, difficult to realize in the best of circumstances, could not withstand the economic, political, and institutional turmoil of the implementation period.
In the two Indian cases, implementation proceeded more smoothly. Exceptions were a suspension of disbursements in Madras, because of noncompliance with a covenant requiring periodic tariff increases by the city bus company; initial institutional problems with the sanitation component of the Kanpur project; and land acquisition difficulties in both operations.
In the end, the Madras project substantially exceeded its physical targets for sites and services, slum upgrading, and bus procurement. Operation of the bus company improved considerably. In both cities, shelter investments have noticeably improved the environment in the neighborhoods affected. Both projects had acceptable re-estimated rates of return, confirming the significant potential economic impact of developing land and low-cost housing in large Indian cities, where the supply of both is often severely constrained.
But even these projects had notable shortcomings. Occupation of sites and services areas in both cities has been slow and many plots have "trickled up" to higher-income families. The Kanpur project was largely unable to upgrade slums situated on private lands, and slum improvement in parts of Madras has led to the eviction of some low-income residents. In both cities, project infrastructure is still poorly maintained, and solid waste services in some areas are still poorly run.
Policy, cost recovery, and institutional goals were only partially met. The Madras project clearly benefitted from being a follow-on operation. But it could not significantly alter state resource allocation patterns for housing. Nor did it greatly strengthen the metropolitan planning agency or the Madras municipal government. Similarly, the Kanpur project did not much improve the operation of the local water utility.
On balance, Bank performance was better in Madras and Kanpur than in Guayaquil and Recife, even though the Recife project--the largest and most complex--consumed much the largest amount of staff and consultant resources before approval. The Madras project benefitted from having the same experienced task manager during most of its life cycle. This helped to ease tensions with the borrower when disbursements were suspended. Similarly, the Kanpur project was supervised by the same core team that oversaw its preparation and appraisal. Both Latin American operations, in contrast, experienced several changes in task managers, while supervision missions often had insufficient personnel to assess performance adequately or provide the assistance needed for all project components.
Experience in these projects shows the pitfalls of trying to deal simultaneously with a broad range of subsectors, problems, and institutions in operations targeted on individual cities or metropolitan areas. The normally complex design of such projects makes them particularly susceptible to macroeconomic instability and political/institutional change. It also calls for above-average Bank involvement during project implementation as well as preparation.
In view of the experience with integrated urban development, more recent Bank urban projects have tended to focus more on individual subsectors such as housing or transport. Others have provided lines of credit for municipal development, covering a range of possible investments but not as part of a predefined package for any particular town or city. Greater attention has been given to the administrative and financial strengthening of urban--particularly municipal-- institutions, increasingly as an eligibility criterion for receiving additional investment funds. The initial experience suggests that these approaches are likely to yield better results than many of the integrated projects supported during the first two decades of Bank urban lending. (See box.)
- In urban project design, appraisal, and supervision, pay close attention to political factors. Borrower commitment to project goals and a viable implementation set-up are key requirements. Political factors will always affect project performance, but their potential adverse impact can be reduced either by taking account of such factors in tailoring project implementation periods, or by designing operations flexibly enough to accommodate shifts in priorities when administrations change.
- Take the legal, regulatory, and institutional framework clearly into account in the design of urban projects. Even the comparatively successful Madras and Kanpur projects highlight the difficulties of trying to influence national or state urban policy through interventions targeted on a single metropolitan area. The potential impact of both these projects was constrained by national and state legislation regulating urban land development, public land acquisition, and shelter.
- Keep project design as simple as possible; including multiple components and institutions disperses efforts and increases risks.
- Consider the advantages of multi-location or program-type urban operations over single-location integrated projects. The former are less vulnerable to changes in local political conditions. They reduce the risks of poor performance by decentralizing implementation and spreading project resources over a larger number of cities. They can stimulate competition among different localities, with likely efficiency gains. They encourage "learning by doing" collectively, due to the larger number of project participants.
Table: Project Facts ____________________________________________________________________________ Project Approved Closed Loan Amount ($ million) ____________________________________________________________________________
Guayaquil Urban Development 1979 1987 31
Madras Second Urban 1980 1988 42 Development
Kanpur Urban Development 1981 1987 25
Recife Metropolitan 1982 1989 124 Region Development _______________________________________________________________________________
Box: Comparisons from Brazil and Ecuador
OED evaluated the Guayaquil and Recife projects jointly with other urban projects in Ecuador and Brazil. These cluster audits permitted an assessment of the effectiveness of "sectoral" versus integrated approaches to urban development in the same countries over roughly the same periods. Operations of the former type were more straightforward to implement and they generally performed much better.
Ecuador's National Low-income Housing project (1982) followed directly from the successful housing component of the Guayaquil operation. Carried out in 15 large and medium-sized cities, it supported a shift in national housing policy toward provision of affordable shelter for low-income families. Implementation was highly satisfactory due to the project's comparatively simple design and good performance by the executing agency. Physical targets for both low-cost housing and home improvement loans were exceeded. The project successfully introduced policy changes in housing finance and strengthened the participating institutions.
Brazil's Third Urban Transport project (1981) likewise exceeded its physical targets: more intermediate cities than planned benefitted from local transport and traffic management improvements, and low-income neighborhoods in more urban areas than planned benefitted from the paving of bus routes. The project also enhanced the operational efficiency of the national urban transport agency.
The Northeast Urban Flood Reconstruction project (1985) was an "emergency" program to help rebuild housing and infrastructure damaged by severe flooding in Brazil's poorest region. Though start-up delays led project resources to be reallocated from housing to infrastructure, the project successfully installed storm drainage and other flood control works throughout the Northeast, affording protection to many low-income urban neighborhoods.
The Parana Market Towns Improvement project (1983), also in Brazil, involved a combination of loans and grants to local governments. It too surpassed most of its physical targets and achieved most of its institutional and financial goals. Among the reasons for success:
- The project was based on an existing local initiative that was able to maintain strong political support even when control of the state government shifted from one party to another.
- Most of the subprojects were small, with relatively simple designs that facilitated implementation.
- The project was able to meet the priority demands of a large number of towns, providing it with additional political support.
- Because it covered several towns, risks of delayed execution or non-execution were spread over a large number of subborrowers.
- The project was coordinated by a competent and motivated team in a state with good public administration and comparatively strong local institutions.