2.2 Affordability, subsidy and cost recovery

A primary objective of recent shelter projects has been to reduce unit costs to a level that can be afforded by the project’s target population. This is intended to reduce subsidies, recover a higher proportion of project costs and enable existing budgets to benefit more households. The ability to achieve this objective clearly depends on patterns of income distribution and the costs of land, services, building and finance attributable to the project. These will, in turn, be determined by their market, or opportunity cost and the minimum standards of initial provision considered acceptable by the project agencies concerned.

Despite increasing awareness of the need for shelter projects to be affordable and for an increased proportion of costs to be recovered, the evidence suggests that progress to date has been modest. Keare and Parris (1982: vi) claim that plots in sites-and-services projects financed by the World Bank during the 1970s were affordable down to the twentieth percentile. Such affordability has, however, often been achieved by subsidizing project components such as land, finance, or services to levels that restrict the numbers that can be delivered within available budgets. For this reason, even when it can be achieved, affordability is of little consequence if the supply of affordable units is restricted and households are denied access.

One possible reason for this problem may lie in the concept of affordability itself. Any approach that determines forms and standards of provision based on what households with different levels of income can afford for housing, or other expenditure, is forced into making assumptions that may not be justified. Furthermore, data on incomes are notoriously unreliable and households at a given level of income may have very different expenditure priorities. Finally, the ability to afford a commodity should not necessarily be equated with a willingness to pay for it.

A more reliable indicator of affordability is probably the existing level of expenditure for a given standard and type of shelter. This is also more useful to project planners in that it can be compared to shelter of a particular type, in a particular location, at a given time, providing an insight into the options with which a project will be compared by its target population. The purpose of assessing the amounts that households can afford for housing is to provide a basis for determining total project costs for initial development. A breakdown of such costs is provided in table 3. The reduction or elimination of any of these costs from the project budget will amount to a subsidy on that element. This will, in turn, be reflected in the perceived value of the shelter within the urban housing market and also restrict the number of units that can be provided. Experience shows that it is extremely rare for shelter projects to be planned on the basis of true costs, even during the feasibility stage. The largest single element of subsidy is invariably that of finance, since even a small percentage reduction in the rate of return charged to a project has a dramatic impact on the capital value saved. The next subsidy in terms of value is invariably in the cost of land. In urban areas this can be substantial. Beyond these, the costs of administration, professional fees and project maintenance are almost never attributed to shelter projects in full, and are routinely omitted altogether.

Table 3. Breakdown of project costs

The understandable desire by project planners not to create developments that might be considered “planned slums” frequently results in project standards being higher than low-income groups can afford. This leads to pressure to provide subsidies in order to ensure affordability, but produces projects that are attractive to higher income groups than those intended. Needless to say, many beneficiaries may be tempted to realize the full market value of the housing received by selling their interest. Furthermore, the high demand for housing provides no incentive for project planners to relate standards to levels of affordability. The dependence on subsidies is thus reinforced. This restricts the number of developments that can be undertaken and creates expectations that are difficult to satisfy.

Another important issue when assessing the economic viability of shelter projects for the poor, is the extent to which attributable costs are recovered for reinvestment in order to achieve replicability. The case studies provide valuable information on the issues of affordability, subsidy and cost recovery. In Turkey, for example, costs in squatter-prevention projects are allocated at market value, except that interest rates are only 5 per cent and a 25 per cent deposit is required. Inflation in recent years has been around 60 per cent and commercial rates of interest 70 per cent. The finance subsidy is thus at least 55 per cent per annum. In upgrading projects, no charge is made for the plot of land, even though the value of this may increase dramatically with the granting of tenure and incorporation into the formal land market. Some form of subsidies is the norm in public-sector projects and this has been officially accepted for some years (Tokman, 1990: 34). In the Aktepe project in Ankara, many beneficiaries only paid the initial cost and the percentage of default on credits was considerable (Tokman, 1990:17). At the same time, households in the Tarsus project had average income levels two-and-a-half to three times higher than the specified maximum. Cross-subsidy policies are considered to have helped achieve financial viability and affordability, though the emphasis on commercial development may exclude low-income groups in the future.

Administrative costs can be considerable and are rarely, if ever, estimated in Turkey. When implementing agencies are assessing the eligibility of project beneficiaries this involves a complicated method of allocating points. This assessment adds to total project costs and adversely affects project viability. This problem no doubt increases with the popularity of a project. Prolonged implementation periods and high inflation rates have made housing projects unaffordable to project agencies, since the costs of building have risen at least as rapidly as other costs (Tokman, 1990: 34). Despite subsidies, projects have therefore failed to reach more than a small percentage of the poor.

In Sri Lanka, the proportion of total project costs recovered has also been disappointing, with only 58 per cent of attributable costs recovered in the rural programmes and 47 per cent in urban areas. In three case study projects, cost recovery levels on housing loans were 16 per cent, 16 per cent and 50 per cent respectively (see table 2). Furthermore, out of 81 projects in Colombo city only one pays property taxes intended to recover a proportion of the costs for land acquisition, development and servicing. Although average household incomes among low-income groups have risen to about SLRs. 1600 per month, households can afford only Rs.75-100 a month of this for housing. This is insufficient to obtain a 25 m² plot of land at existing prices and official standards have exacerbated this by imposing minimum plot sizes of 75 m². The inability to afford access to land effectively excludes many low-income households from participating in government projects. The only way of reconciling this anomaly is to provide subsidies. Although the government has recently waived normal building and planning regulations for low-income housing projects, the only way of reconciling this anomaly may be by providing subsidies.

Jayaratne (1990:43) estimates that the Sri Lankan projects received subsidies on attributable costs of infrastructure and house construction loans that amounted to more than 50 per cent of total costs. In addition, land costs and charges were subsidized considerably and only 20-50 per cent of the market price was charged on suburban sites, with even less in Colombo itself. This has been possible because projects for new development have generally been located on Government-owned land. Once this is exhausted and land has to be acquired from private owners at market rates, it will be difficult to sustain the programme. It also appears that some elements of infrastructure works were not costed to the projects. This implies an additional element of subsidy. Yet, even these subsidies have apparently failed to make shelter components affordable to the target groups. Actual costs seem to be five to seven times the levels that the households themselves indicated that they could afford. The ability to sustain the levels of subsidy inherent in the housing loans from the National Housing Development Authority (NHDA) must, therefore, be open to question.

In Zimbabwe, standards of residential development were raised after independence in line with national aspirations. The minimum plot size was established as 300 m², and a four-room dwelling built of breeze blocks was established as the norm (Mutizwa-Mangiza, 1990: 60-61). This was eventually relaxed after pressure from funding agencies such as USAID. It was also hoped that building costs would be reduced through the operation of building brigades working without the profit motive. The achievement of the brigades has, however, been disappointing.

In terms of their internally defined target populations, the solutions offered within the two Zimbabwean projects examined in this report have, generally, been affordable. An assessment of the affordability of these solutions in the context of the entire spectrum of low-income distribution, however, has revealed serious affordability problems. When the Kuwadzana package was tested against the then income distribution in Harare’s low-income areas (assuming, as the Government did that 27.5 per cent of the monthly income is devoted to housing), it was revealed that 16 per cent of the low-income households in Harare could not afford any of the available options and that only 16 per cent could afford the full four-room core house that the Government of Zimbabwe had insisted on as the minimum to be completed in 18 months. Yet, empirical evidence revealed that only 17.6 per cent (instead of the official figure of 27.5 per cent) of monthly income was available for housing. On this basis 42 per cent of the low-income households in Harare could not afford any of the available options while only 7 per cent could afford a four-room core (Mutizwa-Mangiza, 1990: 55).

In the Kuwadzana project in Harare, land costs were subsidized and this is considered to affect the replicability of the project adversely, especially for private sector developers. A further erosion of affordability was due to the time that it took for the project to be completed. Inflation in the costs of building rose at 60 per cent over two years (1986-1988), a much higher level than was budgeted, or that households could afford (Mutizwa-Mangiza, 1990: 31). Similarly, inflation in the construction sector has resulted in a nine-fold increase in the cost of houses within four years (1981-1985), while the general inflation rate was between 20 and 30 per cent per annum (Mutizwa-Mangiza, 1990: 54). There is some evidence that project beneficiaries are selling their plots. It is not clear, however, if this is because of costs being unaffordable, or if it is to realize the capital value of the subsidies provided.

In Indonesia, affordability is assessed by dividing the target population into two groups. The first group is the lowest-income group, which is defined as households earning less than $US250 per month; they receive subsidies on project costs and terms of finance. The second group consists of households with incomes between $250 and $500 a month. These households not only pay the full cost of their housing, they also contribute to cross-subsidies to help the lowest-income groups. This suggests that the true costs of land, services, finance and shelter provided are based on affordability levels of $250 a month, requiring repayments of about $100 a month.

Commercial interest rates in Indonesia are 18 to 19 per cent. These are charged to middle- to high-income groups, though projects for low-income households are only charged at 12 per cent (Herlianto, 1990: 15). This represents a considerable subsidy on finance alone. Cross-subsidies are generated from commercial and other activities, as well as from higher-income group housing.

In the Kampung Improvement Projects financed by the World Bank subsidies of about $US25 per household are allocated by the Government to cover the cost of improving public infrastructure. In other settlement improvement projects, carried out under the National Pioneering Project, subsidies amounting to about $US1700 per hectare are granted (Herlianto, 1990: 18). Considerable use of subsidies was also evident in the Klender project. This project was intended to serve the needs of a wide range of income groups, in the ratio of one high-cost: three-medium cost: six low-cost, thus reflecting the income distribution for Jakarta as a whole (Herlianto, 1990: 26). Herlianto claims that the scheme was self-financing (1990: 27), though it is difficult to see how this has been achieved, given that 75 per cent of all plots received subsidies (it also appears that few of the remainder were paying full commercial costs). Certainly, this would help explain why up to half of the original government employees allocated plots in the project appear to have moved out (Herlianto, 1990: 28) and been replaced by households with higher incomes.

Projects implemented by Yayasan Sosial Sugyapranata (YSS), an NGO in Semarang, for very-low-income households received substantial funding from donations and grants within Indonesia and overseas (Herlianto, 1990: 50). This enabled the projects to be affordable, but it inevitably restricts their replicability. YSS required only token repayments, in order to instil an awareness that households are acquiring their own assets. The repayments are placed in a revolving fund to contribute to future projects. It appears, however, that the project is essentially undertaken on a charity basis.

Affordability in Indonesia, as elsewhere, has been adversely affected by inflation and devaluation. Shelter projects have thus been increasingly unable to reach their target populations (Herlianto, 1990: 62). As a result, credits (and their associated subsidies) are being received by lower-middle- and middle-income households, rather than the low-income groups intended (Herlianto, 1990: 63). Official project assessments of affordability are based on an expenditure of 33 per cent of incomes on housing. As inflation has eroded affordability and interests have been adjusted, adjustments have been made to which households are eligible to participate in the projects. This problem has been most significant in new development projects, where land has to be acquired, than for upgrading projects where people are in place and the emphasis is on physical upgrading. Private-sector schemes seem to be more affected by this than public schemes. Private developers now find it impossible to reach lower income groups.

One reason why affordability issues are less problematic in upgrading projects, is perhaps that unit costs involved are less or that upgrading standards are more flexible than those for new developments. It is also possible that the existence of a local community that can be consulted about project components and costs enables the most appropriate package to be prepared, and that this increases the willingness to pay. In the Kalingalinga project in Zambia, cross-subsidies were applied to reduce unit costs and generate a revolving fund to finance future works. Unfortunately, inflation eroded the value of repayments based upon fixed rates of interest. The funds recovered were thus insufficient for this purpose (Jere, 1991a).

Housing agencies in Colombia recognize that the problem of low-income groups is a priority. In practice, however, government action has concentrated on those groups that can afford the housing provided (Utria, 1990: 18). Low-cost housing has always been subsidized in Colombia. The subsidy commonly amounts to 70 per cent, including land, professional, financial and administrative costs. The definition of target groups contributes to problems of affordability, since an estimated 71.5 per cent of housing supply is targeted to middle- and high-income groups, while 28.5 per cent is directed at low-income groups. The corresponding figures for housing demand are 27.4 per cent and 74.6 per cent respectively (Herlianto, 1990: 29).

Disparities between intentions and outcomes are, of course, common to policies and programmes, as well as to projects, and may be due to a combination of factors. In the Ismailia Demonstration Projects in Egypt, all project components were costed at rates that, at the time of project preparation, were both affordable and replicable. Cross-subsidies were to be generated from the sale by auction of attractive and well located concessions, and commercial and industrial plots, as well as from differential pricing of all other residential plots. This was intended to enable the project to cater for the needs of all income groups on a self-financing basis.

To ensure that this objective was achieved, initial standards of provision in terms of infrastructure and house construction were extremely modest. The emphasis was thus on stimulating a local process of incremental consolidation. In the event, several changes were made to the project that made it more difficult for low-income groups to participate. These included discouraging the construction of houses using temporary materials, increasing the price of land, reducing the period of repayments from up to 30 years to only 5 years, increasing the size of deposit required, installing full public services at subsidized rates, increasing the minimum width of plots and, perhaps most critically, changing the criteria for selecting beneficiaries so that income levels and evidence of need were no longer applied (Davidson, 1984: 142-144). The outcome of these changes was to stimulate considerable investment in the project area. Yet, the degree to which low-income groups were able to participate was less than intended.

In the Bolivar City project in Bogota, affordability was achieved by subsidizing the serviced plots and interest rates on credit. This represented a combined subsidy of 70.6 per cent, or about $US 1208. In addition, the costs of road provision, administration and professional services were not charged to the project, representing an even higher level of subsidy (Utria, 1990: 44). It is hoped that some of these subsidies can be recovered through property taxes, though it is still too early to assess the amounts, since the project has only recently been completed. In another Colombian project, Kennedy City in Bogota, subsidies were provided under the Alliance for Progress programme, making it difficult to replicate the approach with local resources.

The above examples have confirmed that most shelter projects, even those intended to be self-financing, have failed to be affordable to low-income groups. It has also demonstrated that attempts to close this gap through the provision of subsidies have generally been unsuccessful. When the problems of affordability and subsidy are compounded by poor levels of cost recovery, the financial viability of shelter projects is impossible to achieve and the prospects for replicability are severely constrained. Unfortunately, the record has not been good, even for sites-and-services projects. Van der Linden (1986: 67) cites evidence from more than 60 projects carried out with financial support from the World Bank showing that none had achieved full cost recovery. According to Keare and Parris (1982: xii), rates of default in the Zambian World Bank project were generally in excess of 50 per cent and, in some cases, upwards of 80 per cent, with many households making no payments. Yet, it should not be assumed that poor repayment levels automatically reflect a lack of affordability. In Zambia, the poor level of repayments was largely caused by lack of an effective collection system, and also of political will on the part of Zambian Government leaders. Keare and Parris even went so far as to state that government reluctance to make defaulters pay may be a disguised form of subsidy to project participants (1982: 64 and 85).

In Sri Lanka, the situation is no better. Only 13 per cent of the total repayments due had been paid up to 1986, though this gradually increased to 50-60 per cent, excluding the backlog. Levels later varied between 10 and 30 per cent, depending on how defaults are calculated (UNCHS, 1987: 50 and 64). Cooperative thrift societies have been proposed to improve this aspect of the MHP.

Government policy in Zimbabwe has been to recover all project costs except for land. It is not clear, though, if the cost of finance is based on commercial or subsidized rates. Arrears on repayments in the Kwekwe-Gutu project are between 22 per cent and 28 per cent. Quite obviously this has had an adverse effect on replicability. Yet, this is considered as broadly consistent with the record of middle-income groups and not, therefore, a cause of serious concern. It is not clear how this affected the value of the revolving fund established by international grants. What is clear is that some households assumed that because it was a government project, they would not have to make any repayments (Mutizwa-Mangiza, 1990: 46).