4.2 Replicability of housing projects

Many governments in developing countries have been undertaking innovative projects in the housing sector for several years. The concept of replicability as a major consideration of project design, dates from the time when international agencies, especially the World Bank, became involved in the housing sector in the early 1970s (Cohen, 1983). It was envisaged that if unit costs could be reduced to levels which low-income households could afford, the burden of subsidy would be reduced and it would be possible to repeat projects to the level required, within existing budgetary resources. Yet, as Rakodi has observed, “the achievement of replicability is dependent on the extent to which the technical, financial and social measures adopted in the projects to improve the supply of land, utilities and services, and housing finance, and facilitate the construction and improvement of housing, are appropriate and sustainable” (1989: 2). Replicability depends not only on political commitment, but the institutional capacity to design and deliver these components at the scale required.

The evidence suggests that replicability has been more difficult to achieve in shelter provision projects than in upgrading projects. This is not to deny that some projects have been substantial in themselves and have been expanded into long term programmes; the Dandora project in Nairobi provided 6000 new serviced plots and formed the basis for further projects funded by the World Bank and other agencies. Yet, it took more than seven years from project inception to plot allocation. This reduced the rate of supply to an average of about 850 plots a year at a time when the city needed about 14,500 a year (Chana, 1984: 20-21).

In Colombia, the project approach is held, not only to have failed to address the strategic constraints restricting housing supply for low-income groups, but to have diverted attention away from them (Utria, 1990: 93-4), Furthermore, projects have been designed according to norms and standards more appropriate to the needs of middle-income, rather than low-income households. It is thus the middle-income households that have benefitted most from housing projects in Colombia (Utria, 1990:95). In Turkey, squatter-prevention projects like Aktepe were replicated throughout the country. Yet, they were eventually abandoned as ineffective, since there was no cost control or effective project management (Tokman, 1990: 19). If the approach adopted in the Tarsus project proves successful, it is envisaged that it would be expanded to other areas of Turkey. However, projects tend to be restricted to situations where public land is available and/or where land acquisition is easy and cheap (Tokman, 1990: 27).

In Zimbabwe, the Kuwadzana project met 36 per cent of Harare’s total official housing waiting list for 1981/82 (Mutizwa-Mangiza, 1990: 52). This, however, overlooks the total period during which the project was being prepared and implemented. On this basis, its contribution is more modest. Even this achievement was short-lived and on balance the project approach “has dismally failed” to cope with low-income housing demand (Mutizwa-Mangiza, 1990: 3). One reason for this failure is that public-sector allocations for low-income housing are declining and that private-sector finance institutions, such as building societies, have not yet succeeded in expanding their operations sufficiently to fill the gap. Although the idea of partnerships between public- and private-sector agencies has been accepted, this has yet to be translated into practical strategies.

Another factor inhibiting replicability in Zimbabwe is that standards of infrastructure provision are very high. This has pushed the cost of housing above levels that the intended beneficiaries can afford (Mutizwa-Mangiza, 1990: 68). Increasing subsidies in order to improve affordability would only reduce the prospects for achieving replicability still further, given the declining resources available to meet ever increasing shelter needs.

If the replicability of the project approach in Zimbabwe is to be improved, much will depend on the ability and willingness of the building societies to increase lending to low-income households. Although no building society has so far replicated the approach of the Kwekwe-Gutu project, the experience gained in lending for self-help housing projects has encouraged them to extend their lending operations (on an individual basis) to low-income housing. The largest building society has now financed a total of 2292 housing units for low-income households (Mutizwa-Mangiza, 1990: 47). The fact that several independent evaluations have been made of shelter projects is a positive development. Furthermore, it appears that these evaluations have been taken seriously by the administrators and professional planners in Zimbabwe, to the extent that projects and strategies have been modified as a result. Replicability has not been made easier, however, by poor cost-recovery levels. The main constraint on achieving replicability in Zimbabwe is the inability to produce serviced and affordable land. One way of overcoming this constraint is to expand sites-and-services projects, with a percentage of plots allocated for rental housing. This would, however, give the public sector a major role in the provision of serviced land, despite general experience in other countries that such projects have failed to ensure replicability.

Such problems are not, of course, restricted to Zimbabwe. In a private-sector development in Bandung, Indonesia, replicability was adversely affected by inflation, and later phases of the projects thus drifted up-market. Nonetheless, by providing cross-subsidies from more profitable projects for higher-income households, the approach was still replicable. In fact, the same developer has carried out many similar projects since, and has now built nearly 17,000 units in other parts of Bandung and other cities (Herlianto, 1990: 41). The ultimate test of a successful housing project is that it is repeated or copied by private developers or community groups. Replicability is thus best ensured when other developers want to undertake similar types of project. On this basis, the project in Bandung was definitely successful.

In another Indonesian project, in Semarang, undertaken by an NGO, about 1800 low-income households were able to obtain their own dwellings. Yet, reliance on funds from a large number of private donations and grants restricted the replicability of the approach adopted here. An even bigger constraint on replicability, however, may lie in the limited number of dedicated and energetic leaders available to initiate and supervise such projects.

One of the difficulties in learning from projects is that evaluations are frequently not undertaken and that even when they are, the information gained is not incorporated into subsequent project proposals. In Colombia, evaluations are normally not carried out at all. Furthermore, documentation is often inadequate or unavailable, making it difficult for others to undertake the task. The lack of stability in policies and personnel further impedes the scope for evaluating past experience and learning from it (Utria, 1990: i). The lack of channels for evaluating projects discourages the development of an iterative process in project development and management. This in turn is a major impediment in developing projects relevant to the scale and nature of demand. Even when projects are evaluated, it is comparatively rare for the results to be widely disseminated, so that professionals in other agencies can benefit from the experience. Presumably, one reason for this is the reluctance of public-sector personnel to expose themselves unnecessarily to what may be considered public criticism. Overcoming such reservations will not be an easy task, given that it permeates many government departments. If the objectives of support or enabling policies are to make a major contribution to the provision of affordable and acceptable shelter for all, however, it is vital that the principle of learning from experience is accepted.

In Colombia, it is claimed (Utria, 1990: 59) that the Bolivar City project is replicable, even though the level of subsidy was well in excess of 70 per cent and total budgetary resources for the housing sector are declining rather than increasing. In assessing replicability, it is clearly important to distinguish between acceptability of an approach and the capability of realizing it. In this case, the two would not seem to coincide.

Upgrading projects appear to have fared better in achieving replicability than new development projects. In Jakarta, the Kampung Improvement Programme enabled virtually all low-income households to obtain services and security of tenure within a period of about 12 years. Likewise, in Manila it was expected that the upgrading programme would achieve replicability within 15 years (Walton, 1984: 177). In Lusaka, nearly 20,000 existing houses were provided with basic infrastructure and 5500 plots provided for those affected by plot regularization. This enabled the projects to reach 70 per cent of the residents in unauthorized areas of the city (Rakodi, 1989: 3). One reason for this distinction between new developments and upgrading projects could be that in the latter, residents are already occupying the land at densities that would be difficult to achieve in new development projects. Another factor is that projects for supplying new plots have not been used to enhance institutional capacity to develop land-delivery systems capable of meeting the demand from each income group (Rakodi, 1989: 5).