ASIAN DEVELOPMENT BANK PPA: FIJ 22181

Transcription

1 ASIAN DEVELOPMENT BANK PPA: FIJ PROJECT PERFORMANCE AUDIT REPORT ON THE L O W-INCOME HOUSING DEVELOPMENT PROJECT (Loan 1005-FIJ) IN THE REPUBLIC OF THE FIJI ISLANDS December 1999

2 CURRENCY EQUIVALENTS Currency Unit Fiji Dollar (F$) At Appraisal At Project Completion At Operations Evaluation (June 1989) (July 1997) (July 1999) F$1.00 = $0.66 $0.69 $0.50 $1.00 = F$1.52 F$1.45 F$1.99 ABBREVIATIONS ADB? Asian Development Bank DTCP? Department of Town and Country Planning EIRR? economic internal rate of return HA? Housing Authority ICR? implementation completion report IDC? interest during construction MIS? management information system MLGHE? Ministry of Local Government, Housing, and Environment NGO? nongovernment organization NLTB? Native Land Trust Board PCR? project completion report PPAR? project performance audit report PRB? Public Rental Board TA? technical assistance UNDP? United Nations Development Programme NOTES (i) (ii) The fiscal year (FY) of the Government ends on 31 December. In this report, $ refers to US dollars and F$ to Fiji dollars.

3 Operations Evaluation Office, PE

4 BASIC DATA Low -Income Housing Development Project (Loan 1005-FIJ) PROJECT PREPARATION/INSTITUTION BUILDING TA No. TA Project Name Type Person-M o n t h s Amount Approval Date 976 Fiji Housing Authority A&O $96,000 9 May Housing Authority A&O 2.5 $202, Dec 1989 Manpower Training 1253 Strengthening the A&O 17 $340, Dec 1989 Department of Town and Country Planning 1254 Housing Sector Resource Mobilization Study A&O 7 $199, Dec 1989 KEY PROJECT DATA ($ million) As per ADB Loan Documents Actual Total Project Cost Foreign Currency Cost Local Currency Cost ADB Loan Amount/Utilization ADB Loan Amount/Cancellation nil Amount of Cofinancing IBRD UNDP Government of Japan KEY DATES E x p e c t e d Actual Appraisal 15 May-3 Jun 1989 Loan Negotiations Nov 1989 Board Approval 21 Dec 1989 Loan Agreement 20 Jun 1990 Loan Effectiveness 18 Sep Sep 1990 Project Completion 31 Dec Sep Loan Closing 30 Jun Mar 1996 Months (effectiveness to completion) A&O = advisory and operational, ADB = Asian Development Bank, IBRD = International Bank for Reconstruction and Development, nc = not calculated, PCR = project completion report, PPAR = project performance audit report, TA = technical assistance, UNDP = United Nations Development Programme. 1 Closing date of the World Bank components. The ADB-financed components were completed by 14 March Based on the completion date for the ADB-financed components.

5 5 KEY PERFORMANCE INDICATORS (%) Appraisal P C R 3 PPAR Economic Internal Rate of Return (Part B) 18.4 nc nc Sites and Services Mortgage Finance New Sites nc Mortgage Finance Existing Sites 22.1 nc nc Rental Rehabilitation Financial Internal Rate of Return Sites and Services nc 18.1 nc BORROWER EXECUTING AGENCIES (i) (ii) (iii) (iv) The Republic of the Fiji Islands Fiji Housing Authority Public Rental Board Ministry of Housing and Urban Development (now Ministry of Local Government, Housing, and Environment) Native Land Trust Board MISSION DATA Type of Mission Missions (no.) Person-Days (no.) Appraisal 1 95 Project Administration Inception 1 4 Review 9 87 Project Completion 1 33 Operations Evaluation At project completion, the economic and financial internal rates of return were calculated only for three ADB-financed and three of the World Bank-financed subdivisions under the Sites and Services Development component. 4 Deleted component.

6 EXECUTIVE SUMMARY Prior to the Project, high land and building costs meant that many lower income families could not afford proper houses. These families squatted on land they did not own and there established informal housing areas with undesirable characteristics. The public sector, primarily through the Housing Authority (HA), was the only significant supplier of building lots for the lower income group, as profit margins were insufficient to attract private developers. However, the HA and other public sector institutions involved in supporting the lower income group were ill equipped to meet the challenge. The Project aimed to address the issues and improve the sector s ability to meet the housing needs of the lower income group. To achieve its aim, the Project was to (i) establish appropriate housing policies and standards to support the lower income group; (ii) strengthen public sector institutions; (iii) improve sector efficiency; and (iv) increase, as a short-term measure, the stocks of affordable house lots and mortgage finance for the lower income group. The lower income group was defined as families between the 13 th and 70 th percentiles of the national income distribution. The Project was packaged as a set of complementary parallel components financed by loans and grants from the Asian Development Bank (ADB), the World Bank, the United Nations Development Programme, and the Government of Japan. ADB financed the improvement of HA s management information system, and a portion of the residential lot development, provision of mortgage finance, and rehabilitation of rental units transferred from the HA to the newly established Public Rental Board. The Executing Agencies were the HA; the Public Rental Board; the Ministry of Local Government, Housing, and Environment; and the Native Land Trust Board. ADB also provided three technical assistance grants to support the loan. The ADB components started on time in late 1990 and were completed in 1996, about one year later than expected. The Project was deemed complete in 1998, some three years later than expected. Delays occurred in the development of the housing lots and because of a loan suspension by the World Bank. The rental unit rehabilitation component was deleted, new sites replaced some of the planned subdivisions, and some of the consultant inputs were not implemented. However, the main project inputs were provided as planned. The Project cost $47.3 million, around 8 percent less than expected. The ADB components cost $17.2 million, which is about 17 percent lower than the appraisal estimate of $20.6 million. The ADB loan of $9.6 million was fully disbursed. The design of the Project correctly addressed the main constraints on the provision of affordable housing to the lower income group. The key element was the lowering of housing standards, supported by policy changes so as to achieve a balance between the cost of housing and the financial capacity of the lower income group. The Project advocated lowering the subdivision standards to make house lots more affordable and promoting an incremental housing approach whereby small shelters could be built and added to later, according to the capacity of the owner. Standards and policies to support lower subdivision costs were established, and some modified standards were incorporated into the Project s subdivisions. However, the standards are no longer applied and other factors negated the cost savings for the project-financed lots. The incremental housing concept was not supported by policy development and was not promoted. Consequently, housing costs were not reduced.

7 7 Apart from the HA s management information system, the large project input into institutional strengthening did not produce any significant benefits. In the case of institutional strengthening for the HA, this was largely due to weak and inappropriate management by the HA. HA s new management team that took over in 1996 offers hope that some of the institutional strengthening might contribute to improved operations in the future, however. The expected improvements in sector efficiencies did not materialize. The HA was able to produce and sell the targeted number of lots, and disburse the Project s mortgage finance. However, it achieved these targets, not by reducing housing costs, but by focusing on better-off families who could afford what was produced. Housing costs exceeded those expected at appraisal. An analysis of the distribution of HA s lot buyers and mortgage loan recipients shows that it is skewed, with few beneficiaries below the 50 th income percentile. While those between the 50 th and 70 th percentiles are within the target group, they may be more accurately described as middle income. A significant proportion of beneficiaries were from groups above the 70 th percentile. In retrospect, the target group as defined was too big, encompassing 57 percent of the entire population, and was inappropriately treated as homogeneous. Project management proved to be particularly weak. The Project s management design relied on existing institutional arrangements. The HA, which was the main entity involved in the Project, was more concerned with its financial condition, and less about the development objectives. The Ministry of Local Government, Housing, and Environment was understaffed and could not correct the deficiency in HA s outlook or promote the adoption of the standards and policies. ADB also did not give adequate attention to the achievement of the development objectives. Despite the satisfactory generation of many of the project outputs, implementation was unsatisfactory, and neither the long-term target of establishing more affordable housing arrangements for lower income households nor the short-term target of increasing the stock of affordable housing for lower income families was achieved. Formal housing remains beyond the reach of the majority of the target group, and over the period of the Project, the number of informally housed families grew rapidly. Overall, the Project is rated as unsuccessful. Two issues rise as key for further development of the sector. The first is that a proper balance between the costs of housing and what the lower income group can afford has still to be found. Various options and combinations of options have potential, but some lowering of standards will most likely be necessary. While sector officials in the Fiji Islands consider such modification of standards undesirable, the alternative of informal housing entails even lower standards. Second, the various public sector efforts directed at lower income group housing are uncoordinated and overlap and gaps occur. Because the full extent of the housing problem has yet to be quantified, responsibilities cannot as yet be assigned; neither can appropriate resource allocations be decided. More information and a central entity with sufficient resources to monitor and support the public effort will be required as a first step in overcoming this weakness. The project experience shows that where substantial policy and institutional development is required to support investment components, a two-phased approach may be more prudent than concurrent implementation. In the management of a project, it is necessary that ADB and the executing agencies maintain a focus on the development objectives. Where executing agencies are also expected to achieve commercial targets, the development

8 8 objectives may not be given sufficient importance. In this regard, the project experience highlights the appropriateness of ADB s newly implemented project supervision arrangements that require specific attention to the development objectives. Care is also required in the definition of target groups to avoid overly large and heterogeneous groupings in which substantial segments may remain unserved despite the achievement of physical targets. ADB is considering technical assistance to prepare an urban sector development strategy, which may help in preparing plans to resolve the policy and institutional issues. However, various recommendations concerning some issues already exist for the Government to act upon. As a first step, the Government could identify an agency or body with sufficient resources to initiate and oversee such actions.

9 I. BACKGROUND A. Rationale 1. The Project was initiated in response to a growing shortage of urban housing for lowincome families in the Fiji Islands during the 1980s. The Government also perceived that the construction involved in correcting the housing shortage could help revitalize the economy and that financially less burdensome arrangements could be found for public sector-supplied housing. B. Formulation 2. Project preparation commenced in 1988 with separate but coordinated studies by the Asian Development Bank (ADB), the United Nations Development Programme (UNDP), and the United Nations Center for Human Settlements, which gave recommendations for development of the housing sector. 5 Project formulation and appraisal were completed in 1989 by various missions from those agencies, the World Bank, and the Government of Japan. The Project was packaged as a set of complementary parallel components financed by loans and grants from ADB, the World Bank, UNDP, and the Government of Japan. C. Objectives and Scope at Appraisal 3. The Project aimed to improve the sector s ability to meet the housing needs within the country, particularly of the lower income group. Although the private sector was involved in developing and financing housing, it served mainly the upper and middle income groups. The public sector, primarily through the Housing Authority (HA), was the main source of housing and finance for the lower income group. To achieve its aim, the Project had four key objectives that addressed the main sector constraints at the time, namely, (i) improved housing and urbanization policies, including policies to establish more affordable housing standards for the lower income group; (ii) strengthened sector institutions; (iii) improved sector efficiency through removal of subsidies, better cost recovery, increased private sector involvement, and increased domestic financing for housing; and (iv) increased stocks of affordable house lots and mortgage finance for the lower income group The project scope was divided into two parts. Part A comprised consultants, training, and equipment under six components for institutional strengthening and policy development. 5 ADB s input was provided through small-scale technical assistance: TA 976-FIJ: Fiji Housing Authority, for $96,000, approved on 9 May Improved sector efficiency was not an objective of the World Bank.

10 10 The six components were (i) improvement of the operational and financial management of the HA; (ii) improvement of HA s management information system (MIS); (iii) improvement in the operations of the newly formed Public Rental Board (PRB); (iv) development of housing and urbanization policies under the Ministry of Local Government, Housing, and Environment (MLGHE); 7 (v) development of operational guidelines for the new policies; and (vi) improvement of land-use planning under the Native Land Trust Board (NLTB). The focus of part B was the development of 3,320 fully serviced residential lots, provision of mortgage finance for the construction of 4,490 houses, and rehabilitation of 764 rental units for sale to existing tenants and other purchasers. The lots and houses to be financed were expected to be affordable to families in the lower income group, corresponding to families between the 13 th and 70 th percentiles of the national income distribution ADB financed the improvement of HA s MIS under part A, and 760 lots, mortgage finance for 1,570 houses, and rehabilitation of 404 rental units under part B. D. Financing Arrangements 6. The appraised cost of the ADB components was $20.6 million for which the ADB, on 21 December 1989, approved a loan of $9.6 million from its ordinary capital resources. The appraised cost of the whole project was $51.3 million. 9 The Republic of the Fiji Islands was the Borrower and the Executing Agencies were the HA, PRB, MLGHE, and NLTB. ADB also approved three technical assistance (TA) grants for the housing sector to support the loan. 10 This was the first externally financed project to specifically address housing in the Fiji Islands. E. Completion 7. Implementation was to take seven years to December 1995, although all the ADB components were to be completed within the first five years. However, the Project was delayed, and the ADB components were completed and the loan closed in March The project completion report (PCR) was circulated to the Board in December At the time that the PCR was done, the World Bank had suspended disbursements on its loan due to the failure of the HA to submit financial statements and address the arrears problem. Therefore, the PCR could not comment on the extent to which the overall project was completed or on the total costs. The implementation and achievements of the UNDP and Government of Japan TAs, and ADB s three TAs also are not discussed in detail in the PCR. Nevertheless, the PCR discusses 7 Formerly the Ministry of Housing and Urban Development. 8 This is a commonly used measure to define population groups in the Fiji Islands. The 13 th percentile comprises the lower 13 percent of the population in terms of average weekly household income, and the 70 th percentile comprises the lower 70 percent of the population according to the same measure. In 1999, the 70 th percentile corresponded to an average weekly household income of up to F$184 and the 13 th percentile income up to about F$55. 9 The International Bank for Reconstruction and Development provided a loan of $16.2 million. The Government of Japan provided a grant of $1.2 million, and UNDP, a grant of $0.3 million. 10 TA 1252-FIJ: Housing Authority Manpower Training, for $202,000, approved on 21 December 1989; TA 1253-FIJ: Strengthening the Department of Town Country Planning, for $340,000, approved on 21 December 1989; and TA 1254-FIJ: Housing Sector Resource Mobilization Study, for $199,000, approved on 21 December 1989.

11 11 the design, scope, implementation, and operational aspects of the ADB components in parts A and B. The PCR found that appropriate policies had been developed, and many of the institutional strengthening components of the Project had been effectively implemented. However, the HA and PRB remained financially weak, and most of the house lots and housing finance benefited families toward the upper end of the lower income group and in the middle income group. Overall, the PCR assessed the Project as partly successful. 8. The World Bank components were completed in late 1998, and the World Bank prepared its implementation completion report (ICR) in June The ICR rated the Project as marginally satisfactory. F. Operations Evaluation 9. This project performance audit report (PPAR) presents an assessment of the Project s effectiveness in achieving its stated aims and objectives, the benefits generated, and the sustainability of its impacts. It discusses issues of current relevance to the sector and presents lessons learned from the experience. While giving particular attention to the ADB-financed components, this PPAR provides coverage of the whole project. The results and discussion are based on the findings of an Operations Evaluation Mission to the Fiji Islands in July 1999, a review of the PCR, the appraisal report, ADB files, Government records, the World Bank s ICR; and discussions with staff of ADB, the Government, nongovernment organizations (NGOs), beneficiaries, and private sector representatives in the Fiji Islands. Copies of the draft PPAR were presented to the Borrower, the Executing Agencies, cofinanciers, and ADB staff for review and comments. Comments received were taken into consideration in finalizing the report.

12 II. IMPLEMENTATION PERFORMANCE A. Design 10. The project design correctly identified and addressed the important constraints within the sector. The key element was the lowering of housing standards so as to bring the cost of housing within the financial capacity of the lower income group. The lower limit to affordability was defined as that within the capacity of the 13 th income percentile. Presumably, this was the lowest limit to which standards could be lowered, and households below the 13 th income percentile would need subsidy to afford even this level of housing. Several agencies existed for this purpose. In addition to lowering subdivision standards to make house lots more affordable to lower income households, the Project advocated an incremental housing approach. Under this approach, households could establish a very small basic house meeting the minimum standards of sanitation and structural strength, and add to it over time according to their financial capacity. Photographs showing relevant aspects of the Project are shown in Appendix 1. However, due to implementation problems, lower standards were not widely applied and incremental housing was not actively promoted. As a result, only the middle income and betteroff households within the lower income group could afford the lots and houses offered by the HA, and informal housing, 11 which was the only affordable alternative for the majority of the lower income group, grew (para. 28). The need to establish more appropriate subdivision standards and facilitate incremental housing remained as relevant at the time of operations evaluation as it did when the Project was formulated, and has only recently started to gain recognition in the country. 11. Most of the Project s inputs were provided as designed. The exceptions were the cancellation of the rental flat rehabilitation component, substitution of new sites for some of the planned subdivisions, increase in the number of lots developed at the ADB-funded Tavakubu II subdivision, and reduction and cancellation of some of the consultant inputs. The rental flat rehabilitation program was inadequately conceived (para. 13) and experienced difficulty. The PRB did not favor sale and reduction in the number of the flats, given the demand for rental accommodation, and because of the reluctance of some tenants to either buy the renovated flats or move out of them in favor of other buyers (para. 42). The consultant to help the NLTB establish a land planning capability did not complete the last 24 months because of personal problems and was not replaced because the Government was reluctant to continue to use loan funds for this purpose. The World Bank-funded consultant inputs to help define the implementing arrangements for the new housing policies also were not implemented due to delays in the approval of the policies. 12. Despite the appropriateness of the Project s general design, results might have been better if a more detailed and comprehensive view had been taken of the lower income group, which was the primary focus of the Project. The lack of detail is apparent in the broad definition 11 Informal housing describes housing created by families squatting on land without a formal title. Where families have a title, they are said to be within the formal housing sector.

13 13 of the target group as households between the 13 th and 70 th income percentiles. The target group so defined included 57 percent of the entire population and was hardly likely to be homogeneous. As it turned out, the demand for housing exceeded the supply, and the HA chose to sell its lots and houses by focusing on households in the upper part of the target income range and in the higher income groups (para. 29) 12 rather than address the need to make housing more affordable. Thus, the Project s physical outputs did not contribute substantially to achieving the Project s aim as shown by the rapid growth in informal housing since 1986 (para. 28). 13. The lack of comprehensiveness in the Project s view of the target group is seen in the lack of analysis to show that the Project was large enough and that the physical outputs were appropriate to meet the housing needs of the lower income group. In addition, the adequacy of resources for households below the 13 th percentile and to address the existing informal housing problem was not assessed, and it was not known how such needs would affect the Project s outcome. One important consequence was that the PRB was caught between a growing need to provide housing for households at the bottom end of the lower income group range and a project-imposed direction to reduce the scale of its operation by renovating and selling flats so as to improve its financial position (para. 42). The conflict in direction contributed to the cancellation of the PRB component. The lack of comprehensiveness also meant that attention was not given to the important role that the PRB could play in providing temporary housing for low-income households who were accumulating savings toward house purchase. Nor was attention given to the upgrading of informal housing as a low-cost solution to meeting some of the housing needs of the lower income group. The need for comprehensiveness was starting to be recognized at the time of operations evaluation, however. 14. The Project included a large amount of consultant input and three TAs to achieve important policy changes and institutional strengthening of four separate entities. However, the majority of the project funds financed a short-term time slice of the country s physical housing needs needs that were partly dependent for success on the results of the concurrent policy changes and institutional strengthening. Despite the interdependence of the components, the project design did not provide formal linkages among components, nor was an appropriate management structure established to ensure that the low investment cost components related to policy changes were accorded appropriate importance. The HA, PRB, MLGHE, and NLTB were the Executing Agencies for their own components, and the monitoring and supervision of the Project was placed under the control of the HA and weighted in favor of the HA s housing lot development component. The result was that the policy changes were not given sufficient importance, contributing to the development of housing outside the affordability range of the majority of the lower income households. In retrospect, results could have been better if the Project had been implemented in two parts, as was discussed during project processing, with policy development undertaken before funding of the physical housing. At the least, a more appropriate management structure was warranted. The weak management structure was partly due to the absence of a strong central sector organization, a situation that remained apparent at the time of operations evaluation (para. 59) In addition, the HA also served households in the 70 th -90 th income percentile range. While MLGHE is an appropriate central organization, it has but one full-time officer to handle housing.

14 14 B. Contracting, Construction, and Commissioning 15. Other than the reductions noted in para. 11, consultant inputs and equipment provided as part of the consultant package were generally provided as planned. 14 There are no particular issues concerned with the appointment of consultants, procurement of equipment, and consultant performance. Institutional memory and clear evidence of specific contributions were available for only some of the consultant packages. The HA s MIS developed under one ADB component is operational, and the HA claims it to be highly useful. The wide range of appropriate information for corporate planning and monitoring produced by the MIS supports this claim. Subsequent to the project inputs, the HA upgraded the hardware and further developed the system; however, the system is based on now obsolete minicomputer technology and cannot easily handle nonstandard data queries. The HA also claimed that the inputs for developing corporate planning were highly beneficial. The consultants for the PRB successfully helped separate the new entity from the HA. The NLTB staff sent for overseas training experienced personal problems and did not complete the training. The consultant for the NLTB completed only 12 months, and although his inputs coincided with the creation of a land-use planning unit within NLTB, the World Bank s ICR assessed this component as having had negligible impact on NLTB s operations, an opinion concurred with by this evaluation. The policy-related consultants produced reports which, along with outputs from the ADB TA concerned with the development of standards, contributed to Government approval of the policy changes. 16. There were no issues related to the performance of contractors in developing the housing subdivisions. However, it was noted that the HA subdivisions are on hilly terrain and the subdivision designs incorporated substantial amounts of earthmoving. Both aspects contributed to the high cost of housing. Flat land is not abundant in the urban areas, but the allocation of such land for low-income housing subdivisions would help to reduce housing lot costs for the lower income group. Outside of the Project, the HA constructed houses on the three ADBfunded subdivisions and on four of the seven World Bank-funded subdivisions under designand-build contracts, each contract involving many houses. Poor construction and weak supervision of this nonproject activity resulted in many houses with defects. This had an impact on the Project since many lots on which the houses were built were slow to sell or still unsold at the time of operations evaluation. Moreover, the HA s financial position was weakened because of additional expenditure to correct the defects and for rehabilitation, and because of loss of profit as many house-and-lot packages had to be sold below cost. C. Organization and Management 17. The Project was implemented through four Executing Agencies, but without any formal structure to link the entities and their activities or to oversee project activities. This is a design weakness that contributed to the failure to reach a balance between housing cost and the financial capacity of the target beneficiaries. 14 In addition, three associated ADB TAs also were to contribute to the development of policies and standards, and to the strengthening of HA and the Department of Town and Country Planning within MLGHE (para. 25).

15 The HA had responsibility for devising and operating a monitoring system. The system it adopted and ADB approved focused on the Project s physical outputs. The main information collected were the income percentile and ethnic race of the purchasers of lots and borrowers of mortgage finance, and the interest rate paid on mortgages. While the data had been collected since 1991, there is no evidence that it was used by any of the executing entities or ADB, and so did not help in ensuring the delivery of the Project s physical outputs to the target beneficiaries. 19. The HA received the bulk of project support. It experienced substantial management problems during project implementation, making decisions that seriously affected the Project s outcome. Examples are the design-and-build contracts for houses on project lots, the failure to correctly reflect bad debt and other provisions in the financial statements over the period , and adoption of high-cost subdivision designs. Generally, it appears that during most of the project implementation period, the HA was not concerned with the development objectives of the Project. Although MLGHE was represented on the HA s board, it could not provide sufficient support to the Project at all times as it was grossly understaffed for such a purpose. ADB supervision missions also provided little assistance in this regard. The supervision reports suggest that the missions focused on disbursement and related aspects of implementation, and not on the development objectives, despite the availability of performance data and the slowness in achieving the important policy changes. At the initiation of the MLGHE, the management of the HA was changed in 1996 and the HA now has a better focus on the development objectives (para. 37). D. Actual Cost and Financing 20. The Government did not do well in recording project expenditure. The World Bank s ICR gives the most complete estimate of project cost at $47.3 million, around 8 percent less than expected (Appendix 2). Although the project components, inclusive of contingencies, cost more than expected, the total project cost was lower than expected because of a large reduction in interest during construction (IDC). Actual IDC was $1.63 million, much less than the appraisal estimate of over $10 million. 15 The cost of the components increased because the larger than expected expenditure on lot development and mortgage finance exceeded savings from lower expenditure on policy development and institutional strengthening, and the cancellation of the PRB rental flat rehabilitation component. 21. The ADB PCR estimates the ADB components to have cost $17.2 million, which was lower than the appraisal estimate by about 17 percent. The pattern of actual expenditure is the same as that reported for the whole project by the World Bank. The cost of land and lot development in the ADB components exceeded the appraisal estimates by about 50 percent, particularly for the Tavakubu II subdivision. However, the increase in cost was commensurate with the increase in the number of lots developed. In contrast, the increase in land and lot development costs in the World Bank components was associated with a decrease in the number of lots developed because of a much higher unit cost for lot development. No 15 ADB s appraisal report gives the expected IDC as $11.25 million; the World Bank s appraisal report gives it as $10.1 million.

16 16 satisfactory explanation was found for the difference between the ADB and World Bank lot development costs. 22. ADB s loan of $9.6 million was fully disbursed. The World Bank contributed $12 million of an expected $16.2 million. The Japanese and UNDP grants were fully disbursed. The HA contributed slightly less than expected, the PRB contribution was cancelled with the cancellation of its component, and the beneficiaries of lots and mortgage finance contributed slightly more than expected because the equity contributions were larger than planned.

17 17 E. Implementation Schedule 23. The ADB components started on time in late 1990 and were completed in 1995, about one year later than expected. The Project had started earlier, in 1989, when consultants financed by the Government of Japan commenced work. Effectiveness of the World Bank loan was delayed due to differences between it and the Government over the formula for determining the relending interest rates. 16 The loan eventually became effective in early 1991, but was suspended in 1997 prior to project completion because of the Government s failure to submit audited accounts and effectively address the arrears problem of the HA. The suspension was eventually lifted and the Project was deemed complete in 1998, some three years later than expected. Generally, the TA and institutional strengthening components were implemented early in the project period; the major delays concerned the subdivision developments. 24. Each of the three ADB-financed subdivision developments was finished about 18 months later than expected. Lengthy procedures for the approval and issue of titles within the Department of Lands and Survey were a significant cause of the delay. Extra staff, eventually allocated, overcame the backlog; subsequent procedural changes eliminated this cause of delay for future subdivisions. The substitution of the Tavakubu VI subdivision in place of the Matavolivoli subdivision, delays in the supply by other Government agencies of service infrastructure such as electricity and water, some unexpected civil work requirements at the Manikoso subdivision, and inclement weather also caused delays. F. Technical Assistance 25. The three associated TAs addressed important sector weaknesses and were completed as planned. TA 1253-FIJ (footnote 6) for support of the Department of Town and Country Planning (DTCP) resulted in the design of an Urban Monitoring Unit and appropriate recommendations on subdivision and house standards for reducing housing costs. The Government did not adopt the Urban Monitoring Unit and only some of the recommended standards have been applied, but not consistently nor in a sustained manner. Urban monitoring, and subdivision and housing standards remain as areas requiring significant improvement. Information generated in the TA, however, contributed to policy development (para. 31). The Housing Sector Resource Mobilization Study (TA 1254-FIJ, footnote 6) was completed, but there is no evidence that it produced any tangible change in the housing sector despite the importance of domestic resource mobilization for long-term funding for the sector. The TA for the HA Training Program (1252-FIJ, footnote 6) was completed generally as expected. About 36 HA staff received short-course overseas training over the period ADB s preparatory TA (976-FIJ, footnote 1) for the Project was a small-scale TA and dealt with the HA MIS, which had a satisfactory outcome. 16 The relending rate formula for World Bank funds differed from that for ADB funds. The World Bank s ICR reported that difference as a cause of the slow drawdown of its loan compared with the ADB loan.

18 18 G. Compliance with Loan Covenants 26. Loan covenants were provided to ensure orderly project implementation management and achievement of certain implementation and operational targets. With a few notable exceptions, the covenants were generally complied with. The exceptions, which were also noted in ADB s PCR, are the failure of the HA to achieve financial performance targets, failure of the PRB to annually submit statements of achievements and future plans, and failure of the Borrower to submit all audited accounts and a PCR.

19 III. PROJECT RESULTS A. Operational Performance 1. Low -Income Housing 27. The Project fared poorly in its overall aim to improve the sector s ability to provide affordable housing, particularly for the lower income group. This conclusion mirrors the opinions of those working in the housing sector and with Fijian poverty, and underlies the growing urgency given to correcting informal housing by several agencies within the Fiji Islands. 28. For middle and lower income families, the HA is the only significant source of housing lots in the formal sector. The PRB and the Housing Assistance and Relief Trust offer small amounts of rental accommodation but have not increased their stock significantly over the past decade. The quantity of housing created by NGOs such as Habitat for Humanity and the Rotary clubs is relatively minor. The private sector continues to focus on higher income housing development where profit margins are attractive. Data show that the HA produced 6,070 house lots and sold 6,165 over the period (Appendix 3). Of that number, 3,189 were provided by the Project. In comparison, the urban population grew much more, with an estimated 16,000 households added over (Appendix 4). Assuming that growth was equal over all income levels, the Project s target group of households between the 13 th and 70 th percentiles would have grown by about 9,000 households. The vast majority of this population growth has occurred in the country s two main urban corridors, Suva-Nasouri and Nadi-Lautoka, where the Project s housing lot development took place. Other data from the HA indicate that the majority of those who purchased houses and/or lots or availed of loans have been from the better-off income groups (para. 29). Less than 30 percent of the HA s outputs are estimated to have benefited households in the bottom 50 percent of the income distribution range because of the high cost of HA s houses and lots. In the absence of affordable products, the informal sector has grown. During , the number of urban households with informal housing grew at a rate more than double that of the whole urban population, i.e., by 75 percent compared with 29 percent (Appendix 4). Over the same period, the number of urban households with informal housing is estimated to have increased by 3,400, a number exceeding the Project s output of housing lots. 29. Information on the income percentile grouping of the recipients of HA lots, houses, and mortgage loans has been collected since However, an aggregate picture of the distribution is not easily obtained owing to the use of different categories in different years, discontinuous records, and failure to periodically update the income percentile definition The same absolute income ranges have been used to define income percentile ranges over the whole period of the Project, despite wage increases. Even though the income percentiles by definition include the same proportion of the population irrespective of actual income levels, if the range definitions are not adjusted as incomes rise,

20 20 Annual distributions of mortgage loans and lot sales are presented in Appendix 5, and show a definite weighting toward the better-off income groups. 18 The PCR Mission undertook a detailed analysis of 628 of 707 ADB-funded mortgages. The analysis attempted to overcome the updating problem by adjusting the income levels for the various percentile ranges over time with the use of the consumer price index and reallocating all transactions according to the updated definitions. The results still show that more than 60 percent of the mortgages went to households with incomes above the 70 th percentile, and less than 15 percent went to households below the 50 th percentile. 30. Clearly, the lower income group in the Fiji Islands is not being adequately served by the formal housing sector, and the current situation appears to be worse than that prior to the Project, given the relatively rapid growth in informal housing. To improve housing for the lowincome group, the Project was to address four key points (para. 3). The project impact in each of these areas is discussed in the following sections where it is argued that, despite achievement of many of the targeted outputs, few meaningful changes of benefit to the lower income group have occurred. 2. Policies and Standards 31. New standards for reducing subdivision costs were developed as planned under the ADB TA. Along with other project inputs, the new standards contributed to the approval by Government of new urban policies in However, the new standards and policies did not help in the short term to bring the cost of project lots down to levels affordable to all of the lower income group, nor are they being used to address this issue over the longer term. With reference to the short-term effects, approval of the new policies came after all the project subdivisions had been designed. Nevertheless, recommendations concerning the new standards were available in 1991, and some of the new standards, such as reducing lot size to 100 square meters, relaxing minimum distances between boundaries and buildings, and reduced road widths and densities were incorporated into the project subdivisions. However, other factors, such as excessive earthworks and expensive sewerage design, negated any benefits from the reduced standards and caused lot costs to remain relatively high. In its most recent subdivisions, due for release in , the HA has adopted a new approach under which maximum lot prices are established first, and the subdivision design is adjusted to meet these target price levels. This new approach has the potential to bring lot costs within the financial capacity of the lower income groups; however, the lower development standards are not being used and, unless the situation improves, lot prices are likely to remain relatively high for low-income households. The main impediments to the use of the lower standards include the lack of details in the approved policy statement and absence of implementing guidelines (para. 32), the existence of overlapping regulations and responsibilities in administering housing regulations, weaknesses within the existing regulations (para. 33), and a general reluctance of government officials to adopt lower standards (para. 34). 32. The new policy statement adopted in 1995 was very general in nature and did not provide any guidance in quantifying targets or defining responsibilities. Partly, this may have been due to the dearth of information on the sector and the inability of the consultants to provide transactions will be recorded for higher percentile groupings than they ought to be. Thus, a bias that reflects negatively on the HA s ability to service the lower income groups occurs in results for the more recent years. 18 Of most interest are the distributions since 1993 when the majority of the Project s lots were offered for sale.

21 21 detailed projections of housing needs and costs of alternative development actions. The implementing guidelines, which could have helped overcome policy vagueness, were not prepared. Consultants for that purpose were not recruited due to the delay in approving the policy and the change in government willingness, in the intervening period, to use the loan s funds for financing consultant inputs. MLGHE had only one staff responsible for housing and was ill equipped to provide further support for the policy initiatives. Consequently, implementation of the policy has languished and standards continued to be applied as in the past. 33. DTCP within MLGHE administers regulations governing subdivisions, while the Ministry of Health administers house construction regulations. The main regulations are contained in the Town Planning Act for subdivisions and the Building Regulations of the Public Health Ordinance for houses. The two sets of regulations overlap on matters such as lot sizes and configurations, and space allowances in subdivisions. Moreover, local authorities are also involved in approving subdivision and building applications and the Department of Public Works exerts control over a wide range of road, drainage, and water supply standards. Each agency can have a different interpretation of the regulations, and some efforts by one agency to lower standards to more affordable levels have, in recent years, been blocked by another agency. Such conflict is exacerbated by the prescriptive nature of the regulations that specify minimum dimensions and controlling formulas and which tend to be inflexible and restrictive of innovative approaches. The alternative is regulations based on performance criteria. The consultants under the ADB TA recommended removal of the overlap and the introduction of performance-based regulations, but no action has been taken. 34. A related concern is the reluctance of government officials to lower standards. While the objective of having high housing standards for Fijian society is commendable, it is resulting in a growing number of informal houses that have very much lower standards. Reduction in lot size is frequently mentioned as particularly undesirable. The HA introduced smaller lot sizes, but built houses on all types of lots. The houses were poorly constructed, and it is not known whether the current dissatisfaction of lower income families for the Project s small lots is due to the size of the lot or the poor quality of the houses that were built. 3. Institutional Strengthening a. Institutional Structure of the Sector 35. In addition to the HA and PRB, several other Government agencies have important responsibilities in the housing sector. The other key agencies are MLGHE, which has overall responsibility for housing in the Fiji Islands; NLTB, which effectively controls all remaining available urban land in the major urban areas; the Department of Lands and Survey, which is attempting to address informal housing as well as controlling Government land and titling; and local governments, which administer some of the regulations. As was the case prior to the Project, the Government s services in housing in 1999 are fragmented and coordination among agencies is limited. It is apparent that the Project had limited achievement in strengthening individual institutions (paras ) and was not designed to build a stronger institutional framework involving these agencies to address housing needs. Increased private sector involvement in low-income housing also has not materialized. Land supply and the upgrading of informal housing areas are emerging as crucial issues in urban Fiji and along with the issue of affordability will need greater cooperation and interaction among agencies (para. 59).

22 22 b. Housing Authority 36. Prior to the Project, HA assessment showed it on the verge of bankruptcy. Its lending operations had a large arrears problem involving 52 percent of accounts comprising 14 percent of its loan portfolio, and its rental operation was incurring losses. A revitalization plan was devised and supported by the Project. The plan included transfer of the rental activities to an independent PRB to be established for the purpose. It was confirmed that the HA would service the low to middle income groups in the 13 th -70 th income percentiles, but it would be financially self-sufficient without Government subsidy. 19 It was to concentrate on producing housing lots, upgrading substandard neighborhoods, and providing mortgage finance for houses and lots. 37. At the time of operations evaluation, the HA was a much more complex institution than it was before the Project. It was offering a range of products and was served by a functioning MIS. Nevertheless, it remains financially weak, not having recorded a profit until 1997; burdened by arrears and an unsold stock of expensive and defective houses; and unable to raise funds without Government guarantee (Appendix 6). At the end of 1998, HA s arrears involved 36 percent of accounts and 42 percent of its loan portfolio. The situation could have been much worse. Following a major review of operations and management, a new management team took over in late 1996 and produced many improvements. The HA s financial situation in 1997 and 1998 was much better than it was before Its operations have been streamlined including a change in approach to lot development with outsourcing of land planning and development, cost and fee reduction, and a campaign against arrears. Apart from lack of financial improvement during the project period, it is also clear that the HA did not perform well in achieving its stated development objectives and has served mostly households above the 50 th percentile, including a large number above the 80 th percentile. The pressure on the HA to improve its financial position is likely to limit its future ability to serve households below the 50 th percentile. 38. Inappropriate management, lack of focus on the development objectives, and some poor decisions between 1992 and 1995 emerge as the most dominant causes of the HA s weak performance. Financial performance was poor despite the high sales margins caused by charging a fixed margin over costs of F$4,000 per lot rather than a margin based on percentage. The HA s large house-building contracts on some percent of the lots created in each subdivision weakened its focus on the production of housing lots. Poor contract supervision meant that many of the houses had defects and some remain unsold, affecting HA s profitability. The HA s house building-activity was not part of the Project s original concept of incremental housing. 39. The large project inputs for institutional strengthening did not enhance HA s performance nor prevent the inappropriate management decisions. Nevertheless, the new board and chief executive, installed in 1996, offer hope that some of the training and systems established under the Project may contribute to more efficiency in the future. The HA management judged the MIS and corporate planning training to have had the most impact on its operations so far, and they are expected to be very useful in the future. However, the computer system on which the MIS is based is outdated and lacks flexibility. The HA will have to address the question of its replacement in the near future. 19 Soon after the start of the Project, the HA s target clientele was redefined as those between the 20 th and 80 th percentiles.