India - Madras Urban Development Project

D_mt G The WorldBank USEONLY FOROFFICIAL Report No. 6330 PROJECT PERFORMANCE AUDIT REPORT INDIA FIRST MADRAS URBAN DEVELOPMENT PROJECT (CREDIT 687-IN) June 30, 1986 Operations Evaluation Department Thisdocument hasa restricted only in theperformance distributionand maybe usedby recipients of their officialduties.Its contentsmaynototherwisebedisclosed withoutWorldBankauthorization.  Fon OFCIL US OfILY PROJECT PERFORMANCE AUDIT REPORT PROJECT INDIA FIRST MADRASURBANDEVELOPMENT (CREDIT 687-IN) TABLE OF CONTENTS Page No. Preface * * * * @ @ * @ @ @ * @ * * @ @ @ @ * * v @ * * J i** Basic Data Sheet .. .. . . . . .... ii ................. Evaluatio Summary * Iv PROJECT PERFORMANCE AUDIT MEMORANDUM I. ...... ,..,,..., BACKGROUND ............. 1 A. Urbanization in India and Madras.,.................... 1 BD The Project...e................... .. *...... 4 II. IMPLEMENTATIO .............. 6 A. Gen e r a l 6 B. Features of Project Design........................... 7 C. Physical Accomplishments and Their Impact.ac.. .9.. 0 00 9 D. Institutional Development..... ....... , ,,,,...... ..... 18 III. POINTS OF SPECIAL INTEREST.................. 19 A. Relation of Project to Madras' Shelter Needs5........ 19 B. Policy Imat........@,,,P,. ,,., C. Problems of Land Acquisition......................... 24 D. Metropolitan Management: the Role of Special Agencies and Traditional Authorities..0.,,6...... 26 E. The Madras Experience with Cost Recoveryoossee6...... 27 F. Maintenance and Sustainability.o.,,o,,....... 28 ANNEXES 1. Project Description. ........................ .. 31 2. Borrower Comments .................... . 35 PROJECT COMPLETION REPORT I.o Int r odducctl a0&000000000 onn.o0e0000* 37 II. Project Identification, Preparation and Appraisal .... ,.6 38 III. Project Implementationand Cost .......................... 43 IV. Institutional Performance and Development................ 48 V. Economic Re-evaluation. oo o o o 51 VI. Conclusions ........... 52 Thisdocumenthis a restricted only in the performance distributionand maybe usedby recipients of their oMcialduties.Itscontentsmay not otherwise without World Bankauthorization. be disclosed Table of Contents(cont'd.;, ,, Page No. Annexes A. Principal Report Sources............... ....... ...... 55 B. AppraisalEstimates 57 and iinal C>sts...................... C. e........... Schedule of Disbursemeonts................. o* 59 Map IBRD 12401  - i - AUDIT REPORT PROJECTPERFORMANCE PROJECT INDIA FIRST MADRAS URBAN DEVELOPMENT (CREDIT 687-IN) PREFACE This report presentsa performance audit of the above project for which a credit of US$24 million was made by the International Development Association(IDA) on April 1, 1977. The credit was closed on June 20, 1982 althoughthe physicalcomponentswere not completeduntil June 1984. The report consists of a Project PerformanceAudit Memorandum (PPAM) prepared by the Operations Evaluation Deparmtment(OED) and a Project CompletionReport (PCR) prepared by IDA's South Asia Regional Office. The PCR is based on a project completionreport prepared on behalf of the Government of India and on informationobtainedfrom the Government and vari- ous executingagenciesand from IDA files. The PPAM is based upon reviewof the Appraisal and President's Reports,IDA files,the PCR and a transcript of the ExecutiveDirectors'meeting at which the credit was approved. An OED mission visited Madras in October/November 1984 to visit the facilities included in the project and to discuss the projectwith the Indianauthori- ties whose cooperation is gratefullyacknowledged. The PPAR was sent to the Governmentof India and the project authoritiesfor their vieus. Commentsreceivedare reflectedin the teAt as well as reproducedas Annex 2 to the PPAM.  - li - PROJECTPERFORMANCE AUDIT REPORT URBAN INDIA FIRST MADRAS PROJECT DEVELOPMENT (CREDIT 687-IN) KEY PROJECT DATA Appraisal Actual or Item Estimate Reestimate Total Project Cost (US$ million) 52.0 62.3 Overrun (%) 0 19.8 CreditAmount (US$ million) 24.0 24.0 Disbursed ) 24.0 24.0 Cancelled ) as of June 30, 1985 0 0 Repaid ) 0 0 Outstanding) 24.0 24.0 Date PhysicalCcmponents Completed 3/81 6/84 ProportionComploted by OriginalCompletion Date (X) 100 80 Proportion of Time uverrun (X) 0 81 Economic Rate of Return '2) 2!% Above 21X CumulativeEstimatedand Actual Disbursements (US$ million) FY77 FY78 FY79 FY80 FY81 FY82 Estimated 0 .5 7.5 18.0 24.0 24.0 Actual 0 2.2 6.7 11.2 19.4 24.0 Actual/Est.(X) 0 440.0 83.3 62.2 80.8 100.0 OTHER PROJECTDATA Item Original Actual First Mention in Files 04/72 Government's Application 11/73 - Negotiations 02/77 02/77 Board Approval 03/77 03/77 Credit AgreementDate 03/77 03/77 Effectiveness Date 04/01/77 04/01/77 ClosingDate 09/30/81 12/31/82 Date of Final Disbursement 06/29/82 Borrower Government of India ExecutingAgencies MadrasMetropolitan Development Authority Tamil Nadu HousingBoard Tamil Nadu Slum ClearanceFoard Madras Corporation MadrasMetropolitan Water Supplyand SewerageBoard Pallavan TransportCorporation Department of Highwaysand RuralWorks Small Industries Development Corporation Department of SocialWelfare FiscalYear of Borrower April 1 - March 31 Follow-on Project: SecondMadras Urban Development Project (Cr. 1082-IN)  - tii - MISSIONDATA Mbnth/ No. of No. of Person- Date of Item Year Weeks Persons weeks Report Reconnaisance 10/73 1.0 4 4.0 01/74 IdentificationI 02/74 3.0 4 12.0 08/74 IdentificationII 11/74 0.5 3 2.0 - PreparationI 04/75 2.5 3 7.5 - PreparationII 11/75 0.5 4 2.0 12/75 Pre-Appraisal 04/76 3.0 6 18.0 06/76 SubTotal/a 11.0 24 46.0 Supervision I 03/77 2.0 4 8.0 04/77 Post Appraisal 04/77 3.0 3 9.0 05/77 Supervision II 06/77 0.5 1 0.5 08/77 Supervision III 08/77 1.0 1 1.0 09/77 Supervision IV 11/77 2.0 2 4.0 12/77 Supervision V 02/78 0.5 1 0.5 03/78 Supervision VI /b 06/78 2.0 2 4.0 07/78 Supervision VII 10/78 2.0 2 4.0 11/78 Supervision VIII 01/79 1.0 2 2.0 - Supervision IX 03/79 2.0 2 4.0 06/79 Supervision X 03/80 1.5 2 3.0 04/24 Supervision XI 08/80 2.0 2 4.0 12/80 Supervision XII 05/81 2.0 2 4.0 07/81 Supervision XIII 10/81 2.0 2 4.0 12/81 Supervision XIV 02/82 0.5 1 0.5 06/82 Supervision XV 06/82 2.0 2 4.0 - SubTotal 26.0 31 56.5 TOTAL 37.0 55 102.5 CURRENCY RATES EXCHANGE Name of Currency (abbreviation): Rupees (Rs) World Bank Average Exchange Rate Fiscal Year US$1 - Rs 1976 8.86 1977 8.42 1978 8.06 1979 8.03 1980 8.17 1981 9.28 1982 9.90 1983 10.39 /a The mission data for the appraisalwas not providedin the PCR. 7b Work on the SecondMadras Projectstartedfrom Supervision VI. Personweeks have been proratedbetweenthe two projects.  - iv - AUDIT REPORT PROJECT PERFORMANCE INDIA FIRST MADRAS PROJECT URBANDEVELOPMENT (CREDIT687-IN) EVALUATIONSUMMARY Introduction The First 'JrbanDevelopment Project of Madras (MUDPI),one of the first of the Bank Group's integratedprojects in the urban sector (PPAM, para. 7), was an ausplcious start to the Association's support for the development of Madras,the fourthurban agglomeration in Indiaand one of the world's major metropolises (PPAM, para. l). With its broad scope that covered some six sectors the principalones being shelter, transportand water supply (PPAM, Annex A), MUDPI involved some ten differentagencies (PCR, para. 2.02), with overallcoordination entrustedto MMDA (PPAM,paras. 15 and 45-47). The ambitious scope of MUDPI was justified for at least three reasons; firs..ly and most importantly, the enormity of the urban problems confrontingMadras (para. 1), to which the project componentswere well related (para. 13), called for a substantial intervention, seLondly, the executingagenciesof the government of Tamil Nadu (GTN) were generallysound institutions with demonstrated capacityfor the kind of large scale project implementation required (PPAM, paras. 1 and 30), and thirdly, the project componentsbuilt mostly upon the agencies'previousexperience by helping to perfect solutions that were well understood and for which political commitment could be secured (para.14). Objectives The generalobjectivesat appraisalwere: (a) to develop and promote low-cost solutionsof Madras' problems in the sectors of shelter, employment,water supply, sewerage and transport, and particularlyto make investmentsresponsiveto the needs of the urban poor; by introducing (b) to ensure replicability, full cost recoveryfor key investmentsfor which costs have traditionally not been fully recovered; and (c) to strengthenmetropolitanplanning and capital programmingand budgeting. Accordingto the PCR (para. 2.07), objective(a) was reformulated toward redirectingand substantially increasingpublic investmentfor low income people (PPAM, para. 8). The specific objectivesas detailed in the projectdocumentswere as follows:  (a) Shelter - provisionof sites and serviceswith sanitary cores at three separate locations in Madras and the improvement (viz. upgrading) of 83 slum areas (later reduced to 56) by the introductionof basic infrastructure services. (b) Water supplyand Sewerage- a programof remedialworks pendingthe necessarymajor rehabilitationof the system. (c) Transport- a series of components that includedthe completionof the Inner Ring Road, variouscentralarea trafficimprovements,and the provisionof buses and servicingfacilities to improvethe bus transportation system. (d) Employment- provisionof worksheds,trainingand credit for small scale and cottageindustryin the sheltersites. Impementation Experience Although there is some uncertaintyregardingthe details of what was accomplishedunder the project and its costs (PPAM, paras. 21 and 49), the project was generally well executed to a high physical standard of completion although with considerabledelays, particularlyto the shelter components, caused principallyby the prolongedlegal procedures requiredfor public sector land acquisition(PPAM,paras. 3 and 42). Results The project resulted in significant additionsand improvements to the housing stock of Madras (PPAM, paras. 16 and 20), in some additional employment(PPAM, para. 25), in significant improvementof Madras'municipal bus service (PPAM,para. 26), and in a number of importanttrafficmanagement improvements (PPAM,para. 28). More limitedaccomplishments were recordedin the constructionof the Inner Ring Road (PPAM para. 28) and the water supply component (PPAM, para. 29). The PCR estimated the ERR at over 21% since higher than exDectedrental ,alues more than offset the negative effect on this rate that would have resulted from the delayed benefitstreams. There have been significantaccomplishments in cost recovery; however, for some componentsthe audit has been unable to determinetheir extent (PPAM,paras. 48-49). The audit regards the project as having appropriately built on previous experience thereby facilitatingeffective implementation(PPAM, para. 14) and as having consolidatedthe positive aspects of existing policies, particularly on slum improvement,by making their conception sounderand their executionmore efficient. The Region,on the other hand, considersthat the existingpolicyobjectives were quite contraryto thoseof MUDPI, and that the project, therefore, brought about radical changes. Various aRpects of the policy impact issue are discussed as a point of special interestlater in this report (PPAM, paras. 38-41). The audit does recognize, nevertheless,important innovationsthat can be attributed to  - vi - MUDPI, such as the incroductionof the concept of cost recovery (PPAM, paras. 48-50), the notion of security of tenure in slums (PPAM, para. 44), and the idea of using a sites and servicesapproach for low income housing (PPAM, paras. 40-41). The local authorities in Madras are now in the commendableposition of being able to define finite time horizonsfor the fulfilment of sore long term planningobjectives(in slum improvement, for example). There can be no doubt that MUDPI helped them to attain the rare aclievement of being able to urban problemsof a major developing keep some of the most impo.,rtant country metropolisunder control. Sustainability The inadequateprovisionfor maintenancein the projectmay affect its sustainability. Other sustainability issues related to the project cannot be fully analyzed. However,reliance on an establishedinstitutional model augurs well for sustainabilityas does the introductionof cost recovery concepts. However, full cost will need to be established and general collectionwill need to be achieved for there to be sustainability (PPAM,para. 54). Findingsand Lessons The major findingsand lessonsof the projectare: (a) Sites and servicesand slum improvement related to Madras housing needs in quite different ways. While MULDPI sites and services provided for a net increase in the quantity of Madras' total housing stock by providing additional units, slum improvement raised the average quality of the existingstock by increasing the proportion of it that was "environmentally acceptable" (PPAM, paras. 33-37). that the project achieveda significant (b) The audit cornsiders shift in emphasisin existingshelterpoliciesby furtherencouraging the more cost effectiveand replicablesolutionsaimed at low income groups, expecially through sites and services for example. Nevertheless,the pre-MUDPI political commitment by the local authorities to slum improvement,for example, and their ready understandingof the technicalsolutions proposedunder the project were essential Ingredients of this achievement (PPAM, paras. 38-41). (c) The problemsof land acquisitionare acute; it is a very expensive approach for the public sector, especially in view of private sectormonopolyholdingsof urban land (PPAM,paras.42-44). (d) MUDPI demonstrateda successfulmodel of metropolitanmanagement through the effectivenegotiationof sectoral and local interests throughthe MMDA (PPAM,para. 45-47).  - vii - (e) The idea of cost recovery was successfullyintroduced ty the project and achieved some concrete results in a number of cases (PPAM,paras. 48-49). (f) The lack of maintenance is a serious problem needing urgent attention lest the physical assets of the project be lost altogether(PPAM,paras. 51-53). (g) ivotwithstandingthe effort mad( .o develop them under the project, reliable and simple managementinformation sytems still need to be used more effectively to improveproject implementation and control both by the executingagencies and the Association. The lack of complete and consistent informationon physical accomplishments precludedthoroughevaluation of projectresults (PPAM,para. 21).  AUDIT MEMORANDUM PROJECTPERFOiMANCE URBAN INDIA FIRST MADRAS PROJECT DEVELOPMENT (CREDIT 687-IN) I. BACKGROUND A. Urban'zation in India and Madras 1. It is a well known fact that.,in spite of its low level and relatively slo'f rates of urbanization, the sheer size of India's population places the country among the world'smost importanturban nations. (Mohan, Rakesh and Pant, Chandrashekhar, Morphology of UrbanizationIn India in "Economicand PoliticalWeekly", Bombay, Vol.17,no. 38/39 (Sept. 82i7Tpage 1, World Bank ReprintNo. 295.1 Madras, itself, with more than 5.2 million inhabitants, is the fourthurban agglomeration in India after Calcutta(pop. 10.5 million),Bombay (pop. 9.6 million) and Delhi (pop. 7.8 million),and must be countedamong the world'smajor metropolises. (Ali figuresare taken from the 1981 Censusand refer to the city and its hinterlandin each case.) Of the metropolitan areas of India,Madras is widelyregardedas being one of the best organizedand administered, a reputationderived from a relatively long period of sound and competentadministration by the Government of Tamil Nadu (GTN), but even so, the city presented as formidable a list of urban problems as almost any other major developingcountry metropolis. Among those derived from inadequate infrastructure, for example,one might mention Madras' inadequate water supply system which is able to serve many areas of the city for only two hours a day, the drainage problems in low lying neighborhoods leaving them prone to frequent flooding and associated environmentalproblems, the generally inadequate housing stock and the inefficient use of the road network as pedestrians, cyclists,animal drawn vclOicles and motorizedtraffic as well as animals all compete for the same limited3pace. Solutions to these kinds of problems have to be sought in the contextof what is a predominantly low incomeurban economywith a per capita incomeless than most other metropolitan areas of India. Such a situation as this poses an enormouschallengeto Madras'urban plannersand policymakers, and makes it paramountthat the prioritiee determined be supported by a full analysis of the economic opportunities available something which has not traditionally been emphasized in Madras'planning._! 1/ Attentionhas been drawn, for example,to the lack of specificinstitu- tional arrangementsfor advisingGTN on the requirements for the econo- mic developmentof the f,fA(see StructurePlan for Madras Metropolitan Area, Vol. 1 - Existing Situation, Madras Metropolitan Development Alan Turner and Associates, AutU,ority/wnosultants June 1980,page 43).  - 2 - 2. Importantplanning initiatives have, nevertheless, been taken in Madras in an attempt to at least order the approach to some of che problems mentioned. As far as the shelter sector was concerned,for example, GTN began to tackle the city's slums under the Slum Areas (Improvementand Clea:ance) Act of 1956, which gave the Governmentpowers of compulsory acquisitionof land for the purpose of either clearing them and relocating their low income inhabitants, the so-calledEconomically Weaker Sectionsin speciallybuilt tenements,or undertakingimprovements in situ through the installation of basic infrastructure.Early on, most emphasis in India had been upon clearance, but the latter,more cost effective approachwas gaining favor so that major slum improvement (called environmentalimprovement) schemeswere alreadyunderwayin Calcutta by 1971 and also in Madras,itself, by the tfllowing year. From FY1972/73to FY1976/77,48,954 families in Madras had benefited from such (environmental improvement)schemes (data taken fr(m StructurePlan for Madras op.cit, p. 47), almost twice the number of beneficiaries under MUDPI. Nevertheless, over the same period, 17,484 families had been accommodated in expensive tenements following slum clearanceschemes. Throughhelping to curtail the latter and perfecting the slum improvement programprincipally by introducing conceptsof cost recovery and securityat tenure,MUDPI can be creditedwith having helped consolidate a major shift in policy on slums. The MUDPI experience has been observedby politicians and administrators from most other states and has had widespread implications for shelter policiesin Indian cities as a whole. In the case of the provisionof new low cost housingunits, the projectwas similarlyto have an importantinfluence over sector policies(see PPAM, paras. 38-41). 3. The direct effects of policy shifts such as these (and of MUDPI, itself)might have been even greaterhad it not been for the distortionsin urban markets in which they had to be implemented. Considerable areas of urban land in Indian cities are Government-owned, while that which is in private hands is dominatedby monopoly holdings. (See: Gupta, DevendraB., Urban Housing in India, World Bank Staff Working Paper No. 730, 1985, pages 119-120). In a situationwhere the solutionof urban problemsis held to be exclusivelythe responsibility of Government,the land requirements of the latter's urban development agencies can, especially in the face of a restricted monopolistic supply,invitespeculative dealings,force up prices, and make public sector land acquisitionat market rates a very expensive businessindeed. Partly in recognition of this, an unsuccessful attemptwas made to dismantleurban land monopoliesthrough the 1976 Urban Land (Ceiling and Regulation) Act, but insteadof increasing the availability of urban land to the public sector as intended,the Act effectively led to the withdrawal of the privatesector from the market. In the specificcase of Madras,the persistentproblemsof land acquisition and the continuation of the poor and unimproved condition of slums on privateland (whichhousedmore than half of all slum dwellers,accordingto the 1971 Census)are but two consequences of the land ownershipproblem in the metropolitan area (see PPAM, paras. 42-44 for discussionof these aspects as a point of special interest). MMDA are exploring to include alternatives to land acquisition including Guided Development and Land Readjustment. Acquisition of land from private owners would not be required, as owners would iTi effect become partners in the development(or redevelopmenc in the case of slums) of specialized areas.  -3- 4. As in the case of any growing metropolis,the deficiencyof the urban infrastructure of Madras will remain at existing levels unless new capacityis providedat a fasterrate than that of urban growth. With a pro- jected populationgrowth rate of 4% per annum, each year the equivalentof about 40,000 new households(using 1984 base and 6 persons per household) seek to satisfy their shelter needs (see PPAM, para. 36). However, the supply of extra units cannot meet all this potentialdemand and the infra- structure deficit is in fact increasing. As a result many families (especiallythose with low incomes) are obliged to share homes and the resultingovercrowding is perhaps as meaningfulan indicatorof the housing deficit as homelessness itself is. Lack of both up-to-date data and also a generallyrecognizedset of standards do not permit a precisequantification of Madras' housing deficit. However, recent estimates of the physical conditionof the housing stock (as opposed to the housing deficit)indicate that some 45% of it is deficient,30% of the total consisting of slum huts while another 15% is made up of dilapidated"permanent" structures.2/ 1Uow this housing stock was affectedby MUDPI is discussedlater in this reportas a point of special interest which draws attention to the impact of the projectupon the housingneeds of Madras (PPAM,paras. 33-37). 5. The populationgrowth of the Madras metropolitanarea (MMA) has also led to a significant increase in the demand for transport. In a city where some 40% of all trips are made on foot or by bicycle,the principal constraint upon the growth of these modes is the increasingcongestion in the central areas of the city and their lack of rights-of-way, urgent problems that were both addressedby MUDPI. Another 40% of trips in Madras are made by bus, using extensive services provided by the Pallavan Transport Corporation, but with predominantly over-agedand overcrowded vehicles prior to MUDPI. The three local electrifiedrailway lines that serve Madras account for some 10% of all trips, many of which are integrated with the bus network. The remaining 10% of trips are carried out by private cars (car ownership rates are falling), motor cycles,taxis and auto-rickshaws, all but the last one showing little potential for growth. By concentrating upon pedestrian,bicycle and bus traffic,the MUDPI's transportcomponentswere well harnessedto the principalmodes of transportused by the majority of the population of Madras,particularly those with low incomes,the demand foe which can be expected to continue to increasein the foreseeable future as the city itself grows. 2/ Structure Plan for Madras MetropolitanArea, Vol. I, Ch. 4. The dilapidated"permanent" structurescategory would itself be classified as slum housing in many countries,but the StructurePlan follows the statutory definition in India by using the term slum to refer exclusively to hutment areas without proper access, water supply and drainage.  - 4- 6. One sector in particular that has been left far behind as Madras' population has increased is that of water supply. Not only is the distribu- tion network inadequate, but the existing sources would not, in any case, be sufficient to supply an improved network. The PCR makes clear reference to the important problems of water supply in the metropolitan area of Madras (PCR, paras. 1.03 and 2.06) and the seriousness of the situation was well understood by the Association at the time of appraisal. As a major study, sponsored by UNDP/WHO, was then underway into the water problems of Madras, a duplication of efforts might have been involved if MUDPIwere to have fully addressed what perhaps was MMA's most serious urban problem. Major policy decisions affecting the sector have been taken but water shortages continue to affect Madras, and even if all the necessary works were begun immediately, little relief could be expected in the short run due to the long lead time of the major investments required. B, The Project 7. MUDPIwas the third urban project in India to be supported by the Bank Group. [The first one was the First Calcutta Urban Development Project (Credit 427-IN); for the audit of this project see PPAMNo. 4023. The second was the Bombay Urban Transport Project (Loan 1335-IN) which is still under implementation.] At the time that the Credit for the first Calcutta project became effective, the Association was examining the possibility of supporting an urban development project in Madras. In contrast to Calcutta, Madras was not felt to be in a state of crisis nor was it thought to be as congested as Bombay. Furthermore, the Association noted Madras' active pursuit of urban policies and planning in the city with approval, even if it did not entirely agree with the priorities adopted. Thus, an appropriate context had been identified in which to implement one of the Bank Group's first of a then new type of integrated urban development project, recently described as follows: "These projects were distinguished from (earlier) shelter projects by the scale of individual components, the number of agencies involved within a citywide framework, and the effort to integrate investments. They were designed from a broad conceptual perspec- tive that attempted to achieve an overall urban view - rather than a subsectoral one - at the outset."3/ 8. At appraisal (SAR, para. 3.01) the main objectives of the project were: (a) to develop and promote low cost solutions to Madras' problems in the sectors of shelter, employment, water supply, sewerage and transport, and particularly to make investments responsive to the needs of the urban poor; 3/ Learning by Doing: World Bank Lending for Urban Development,1972-1982. World Bank, Washington, D.C., 1983, p.20.  by introducing (b) to ensure replicability, full cost recoveryfor key investmentsfor which costs have traditionally not been fully recovered; and (c) to strengthenmetropolitanplanning and capital programmingand budgeting. Some revision of the substanceof project objective (a) in particularwas made in the PCR (para. 2.07) where, instead of the pursuitof the ambitious new low-cost solutions to Madras' problems implied at appraisal,specific emphasis was placed upon the more realistic aim of redirecting and substantially increasing public investment in favor of the low income population. The PCR's restatementmore accurately reflectsthe true goals of a project that succeededin perfectingand giving a new emphasisto solutions (such as slum improvement) that had, in large measure, already been tried before (albeitwith little success,according to the Region). 9. To achievethese objectives,a complexprojectdesign was conceived which covered some six different sectors and involved ten executing agencies. The fact that all these agencies were already in place and operatingsuccessfully to a greateror lesser extent,justifiedthe ambitious scope of the projectand reducedthe risk that its scale and complexitymight overwhelmthe capacityof the GTN institutions to successfully implementit. The principal componentsof the project (detailed in PPAM, Annex A) as appraisedand implemented can be summarizedas follows: (a) Shelter - provisionof sites and services with sanitarycores at three separate locations in Madras and the improvement (viz. upgrading) of 83 slum areas (later reduced to 56) by the introduction services. of basic infrastructure (b) Transport- a seriesof componentsthat includedthe completion of and the Inner Ring Road, various centralarea trafficimprovements, the provisionof buses and servicingfacilitiesto improvethe bus transportation system. (c) Water Supplyand Sewerage- a programof remedialworks pendingthe necessarymajor rehabilitationof the system. (d) Employment- provision of worksheds,trainingand credit for small scale and cottage industryin the sheltersites. In addition,the projectprovidedfor a small health componentand technical assistanceto the Madras Metropolitan Authority(MMDA)and Madras Development Corporation. The PCR lists the agencies responsiblefor the executionof each oL the components(PCR,para. 2.02). 10. The project addressedsome of the most importantissues of urban development in Madras and adopteda somewhatconservativedesign inasmuchas it mostly called upon executingagencieseither to alter and to expand some  - 6 - of the programs with which they were already familiaror otherwisecurtail other, less effectiveones on their agenda. With the except'onof the Inner Ring Road and one of the sites and servicesschemes,all the project compo- nents were implemented much accordingto the appraisalex.pectations. Those components which dependedupon land acquisition sufferedslow start ups; pro- cessing took three years in some cases, much longer than anticipated(PPAM, paras. 42-44). Delayswere also experienced in grantinglease-cum-sale (LCS) agreementsto low incomeresidents,but this is to be expectedwith what was an innovativeaspect of MUDPI. These delays, combined with the unusually high inflation rate (by the standardsof India)during implementation, led to cost overruns of some 19.8%. Nevertheless, when eventuallyfinished,the physicalcompletion of the componentswas generallyof a high standard. The overall coordination of the project developmentwas entrustedto the MMDA without depriving the executing agencies of the direct control of the management of the components'physical implementation. This was a wise provision given the scope and complexityof the project. The MMDA was particularly suited to the task as it had already accummulated some experiencein coordAnating metropolitanaffairs without having stifled the actions of the traditional local authorities. II. IMPLEMENTATION A. General 11. While there is some uncertaintyin documentation as to the exact physical accomplishments and cost experiences of the project, it appears to have been implemented along the broad lines set out at appraisal. Maintaining the integrity of the originalprojectdesign can be attributed to the fact that, from the outset, the projectaddressedMadras' most pressing urban needs, and also to the fact that GTN consistently afforded such needs project implementation.4/ However,there high politicalprioritythroughout were significantdelays in execution primarily due to land acquisition problems. Land acqi.isition prior to Board approval would, of course, have speeded implementat.on although it would also inevitably have delayed Board presentation.As _ar as the urban development needs at Madras are concerned, such a change, although it might have improvedthe apparent performanceof the Association's project itself, would have been little more than cosmetic since the actual time from identification to completion would be unchanged. One way or another,land acquisition and metropolitan real estate management 4/ The audit considersthat elementsof such a government commitment, such as existed toward slum improvementprior to MUDPI should be a sine qua non of any development project. The nearly 50,000 families who benefitedfrom slum improvement before the Bank project (even if these early programs were less effectivethan those under MUDPI) is taken as an indicatorof GTN commitmentto low cost shelter solutionsfor the urban poor. have to be faced by the local authorities as major urban planning issuesin their own right, irrespectiveof their being or not being formallypart of a specific developmentproject (PPAM, paras. 42-44). Cost overruns were also caused by unexpectedly high inflationduring implementation, somethingthat could not have reasonablybeen foreseenat appraisal. The immediateImpact of MUDPI on Madras' urban problemsas a whole was obviouslylimited by the sheer scale of the latter, but even so, project componentswere to account for about one-halfof public sector investments in the relevant sectorsover the seven year implementation period. This did not justify, however, the claim made at appraisal (in 1977) that all slum areas in Madras would be improved by 1985 (SAR, Summaryand Conclusions, para. xi). As well as prema- ture, this affirmation did not fully take into account the dynamicnature of Madras' urbanization as new slums were being built while the existingones were being improved (PPAM, para. 22, footnote 7). Nevertheless, through their experiencewith MUDPI, the authoritiesin Madras have probably come nearer than many others in similar situationsto not letting their city's urban problemsget out of control. The very fact that it is now possiblein Madras to consider target years for the fulfilmentof realistic long-term urban policy objectivesis itself a significantachievementfor which the projectcan take a good deal of credit. B. Featuresof ProjectDesign 12. Project identificationin this case was complicated by considera- tions about the degree to which MUDPI should be involved in Madras' water supply problems,one of the most crucial urban developmentissues in the metropolitanarea at the time of appraisal(PCR, para. 2.01). Faced with such an issue, the Associationopted for only limiL:d remedial works for water, and wisely left the general water supply issue (whichhad state-wide implications beyond the MMA) to the ongoing UNDP/WH0-ss-insored study. Althoughsuch an option incurred the risks that other components(especially shelter)might have their performance impaired by the shortageof water (as in fact occurred),to have included major investmentsin the water supply system would have been premature without the outcome of the study being known. If the project had been delayed to await the results of the study, significantimprovementsin other sectors would also have been postponed needlessly. 13. The projectcomponentswere well chosen for being closelyrelated, as has alreadybeen mentioned,to the pressingneeds of the low income popu- lation of Madras. Thus, slum upgradingaimed to providesome relief to the shelter problemsof some of the lowest income inhabitants of the MMA, while sites and services were intended to supply extra housing units to families that had somewhat higher incomes but still had been unable to afford the rents in erstwhilepublic housing schemes. MUDPI also achieveda low income orientationby concentrating on the transportmodes used mostly by those strata of society (i.e., buses, bicycles and pedestrians). As these three modes accountedfor the bulk of the trips in Madras,MUDPI'stransport compo- nents also aimed directly at improvingthe efficiencyof the system as a  -8- whole. Althoughon a strictlylimited scale, income and employment genera- tion throughthe provision of factorysheds for small industriesin the sites and servicesareas and the financial and technical supportfor cottageindus- tries in the slum areas contributedto the economicneeds of the population. 14. As indicated in para. 10, an essential ingredient for the successful implementationof the project was the inclusion mostly of componentswith whose methods of implementation GTN agencies were familiar, if not in every detail, throughtheir experience prior to MUDPI. The project succeeded in consolidatingthose more cost-effectiveapproaches and in improvingthe managementof them, while at the same time, discouraging the applicationof other less appropriate solutionsthat still found themselves on GTN's agenda. An example of the former is slum improvement, where MUDPI built upon and perfected the previous experienceof the Tamil Nadu Slum ClearanceBoard (TNSCB)with Environmental Improvement Schemes(EIS) in slums which had benefited 48,594 householdsprior to the Associationproject. 5/ In the case of the latter, MUDPI, through its credit covenant,succeeded Tn curtailingthe very expensiveslum clearance/purpose built tenementscheme. The project sites and services schemes, however, represented a greater departurefrom the p;ior emphasisof the Tamil Nadu HousingBoard (TNHB)upon the provisionof completedunits for higher income groups, but the component presented no technical difficulties for an agency that had previously executed a program of "open developedplots" (sites and servicesfor higher incomes) and that had thirty years' experience in awarding credits for buildingmaterials. The project'sbus transport componenthad a similarkind of impact by further improvingthe efficiencyof PTC, a bus company with a proven track record prior to the Association's assistance. The approachof buildingupon and perfectingthe more effectiveparts of established programs that characterized MUDPI not only reduced project risks but also helped to promote the institutional developmentof GTN agencies that were already in place (see PPAM, paras. 45-47). The audit considersit to be much to the creditof the projectdesign and preparation that MUDPI was able to introduce changes throughthe adaptation of solutions(such as slum improvement) which were well understoodand supportedby GTN as opposedto unsuccessful attempts in some other countries to introduceuntriedsolutionsfor which there was no political commitment. 5/ Accordingto the Region,however,such schemeswere not successful since the facilitiessoon ceased to operate through lack of maintenanceand little or no home improvementtook place because of lack of tenure. While the audit considersit too early to draw definitiveconclusions regardingprogresswith maintenance (PPAM,para. 52) in MUDPI'sown slum improvement,the project did introduce new concepts such as cost recoveryand securityof tenure for the beneficiaries.The latterwas a particularlyimportant innovationsince, among other things, it can contribute to the permanenceof the .mprovementsgiven the legal rights to them that it confersupon the beneficiaries.  - 9 - 15. The project relied upon the coordinatingcapacity of the MMDA (strengthened by technical assistance)to oversee implementation and the physicalworks undertakenby the executingagencies. This was a wise design feature in view of the scope and complexityof the project, enabling it to take advantageof the MMDA's previousexperienceof coordinatingmetropolitan affairs as MMA's planningauthority. In this way, the projectalso reduced the administrative burden on the Associationleaving it to concentrate on major issues and problems, with responsibility assigned to MMDA for many day-to-day matters. Becauseof its implicationsfor the designof other Bank Group urban projectsthis important experiencein metropolitanmanagementis discussedas a point of special interestlater in this report (PPAM,paras. 45-47). and Their Impact C. PhysicalAccomplishments 16. MUDPI's di-ect contributionto increasingthe housing supply in Madras was through the sites and services component which was to have provided 13,500residentialplots at the three separatelocationsplannedat appraisal. Accordingto MMDA/TNHBdata suppliedfor the audit, however, only 8,068 lots were actually prepared,some 40% short of the total intendedat appraisal. Cost overruns at these sites (interpretedfrom PCR, Annex B, p. 1) would appear to have been nearer 60% than the 35% referred to in the text of the PCR (para. 3.04). The combinationof the reduced physical accomplishment with increased total costs would imply a more than twofold increasein the unit cost of the lots. The reasons for the increased costs are explainedin the PCR (para. 3.03). Of the total of 8,068 plots prepared,  - 10 - however, the audit mission was informed that up to September19856/ only 2,684 (33.3%)were in fact physically occupiedby the beneficieries, although construction of the houses had begun on another 2,860 (35.47). Between the three sites there were considerable differencesin the degree of progress achieved. Visual inspectionof the Kodungaiyursite especiallyand also Villivakkamconfirmedthe slow advaAicement of the componenteven after the plots had been fully assigned. The main reason given for the fact that 31.3% of the plots remained undevelopedand unoccupiedwas the lack of funding availablefor the allotees to build their homes. Other reasons advancedby the Region for the slow rate of plot occupationincludethe lack of water and seweragedisposal infrastructure, securityand access problems on Kodungaiyur and slow issuanceof allotmentorders due to TNHB staff constraints. Apart 6/ The Region reports that considerable progress has been made since that date but the latest figures are not comparablewith the data in this section which was suppliedby MMDA/TNHB. For a site to be regardedas occupiedthe audit considersit essentialthat there should le physical occupation by the beneficiary,even if the house is no. completely built, and not simply with construction well under way as TNHB statisticsallow. Without the physical occupation of the lot (even if the house were complete) there would be no economic benefit flows related o shelter services as intended by the project, although admitted I there could be financialbenefits from what could amount to nothing re than a speculative holding of the asset. The point is an importan. ne to make in Madras since the differencebetween rates of physical upation (adopted by the audit) and those from TNHB statistics referredto by the Region are considerable,as follows: Rate of Physical TNHB September1984 Occupation/a Occupation/b Site: Arumbakkam (3) 71.7% 97.1% Villivakkam 26.4% 75.3% Kodungaiyur Phase I (4) 2.0% 23.9% TOTAL 33.3% 68.7% Source: MMDA/TNHB occupied. Notes: /a Definedas lots with houses physically 7" Definedas lots assignedwith constructionunderway(whether physically occupiedor not). /c Data for Arumbakkamrefers to July 1984 (others September 1984). /d KodungaiyurPhase II excluded from this analysis as it is part of followon projectMUDPII.  - 11 - from its more strategic location in relation to centers of employment, housebuildingfinance from commercialbanks was the key factor in ensuring the success of the Arumbakkanscheme, probably one of the more successful projectsof its kind, which is already some 72% occupied (in contrastto the 2% occupationrate at Kodungaiyurand 26% at Villivakkam). 17. Field visits to the Arumbakkamsites and servicesscheme confirmed that extensivebuildinghad already taken place and, wisely, the autho:ities had adapted the intendedland use and height regulations to suit the needs of the residentsand of the shopkeepers in particular. Thus, for example,the main commercialcenter was located in a differentstreet from that intended in the original plan. Also, building regulationswere applied flexiblyin order to give maximum encouragement to the owner/builders. Perhaps the flexibility in building codes went too far in one respect,however,where the clearancebetween the house and the sanitary core required for maintenance was often built over, meaning that part of the house might have to be demolishedif major repairsto the drains ever becamenecessary. But so far, seriousmaintenanceproblemsdo not seem to have arisen in Arumbakkam. The streets and footpathsare in good conditionand attentionhas been paid to urban design through tree planting. One serious problem was, however, the shortageof water whose supplyto the site appearedto be sporadic(see PPAM, para. 12). In fact, during the site visit by the audit mission a number of women of Arumbakkam were protesting about the lack of water. While reflecting a failure of the Madras water supply system as a whole, the demonstration ironically revealed the successfuluse of the communitycenter providedunder MUDPI, as it served as the meetingpoint for the discussion of the problem. Conversations with a number of residentsimpressedthe audit mission by the genuine desire on their part to attain home ownership;the status that they thought this conferredupon them was cbviouslycomplemented by the fact that many were paying only one-half in repaymentinstallments of what they had previouslypaid in rent. There can be no doubt that the Arumbakkam scheme, becauseof its generallysuccessful performance, would be worthy of further detailed study in order to determinethe lessons that can be learnt from it and appliedto other schemesin Madrasand elsewhere. 18. The other two sites, as has already been mentioned,were much less developed. At Villivakkam, the drainageand reclaiming of what was a former irrigationtank led to considerable delays and increasedland development CostS. At Kodungaiyur, the delays were due to land acquisitionproblems. In spite of the difficulties, the audit mission was informedthat all the plots at both sites had been allocdted,although (by September1984) construction had begun on only 24%, while a mere 2% were in fact physicallyoccupiedby the beneficiaries. The lack of occupationof both sites has led to a deterioration in the physicalconditions of the infrastructure(e.g.,manhole covers missing, drainage culverts broken) which will require considerable expenditure to repair. The effect of the sanctionsupon those alloteeswho do not develop their plots seems to have been limited to encouraging them at best to undertakethe minimum foundationwork necessary in order to avoid being evicted under those rules. The lack of finance for housebuilding was given as the main reason for the lack of progress at these sites, although  - 12 - access and securityproblemsdo not make this the most soughtafter location in the MMA. 19. In spite of their varying physicalperformance, all three sites had some limited success from a financial point of view by generatingfunds through cost recovery. In the case of the Arumbakkamscheme the recalcula- tions indicatea low but positive financialrate of return. Effectivecost recoverymechanisms have been set up for the three sites and servicesschemes and enabled a significant fund to be accumulated, althoughso far it has yet to be applied in an ongoing program. The fund has, therefore, yet to acquire a truly "revolving" nature and a permanentprogram of sites and serviceshas yet to get underway. Nevertheless, the introduction of the cost recovery concept to sites and services schemes in Madras can be considered an impor- tant achievement of MUDPI (see PPAM, paras. 48-50). As far as the economic evaluationis concerned,the audit considersthat the 30% economicrate of return reportedin the PCR (para.5.04) for the Arumbakkam sites and services (cf. the 20% ERR estimatedin MMDA's PCR) is an unduly optimisticstatement of that component'sperformance given that 28.3% of the lotE had still not been occupied at the time of the reevaluation. While methodologically correct and well elaborated,the reevaluationexercise,however, assumes a full economic benefit flow of imputed rental values from the first year of operationof the project,irrespective of whether the site is occupied. The audit considersthat imputed rents, as a proxy for the economicbenefitsof housing,ought to be counted only when a lot is physically occupied,thereby effectivelydemonstrating the resident's reaping of the intendedbenefitsin the form of shelterservices. The audit considersthat the recalculation of the ERR should take into account the actual experienceof the rate of physical occupationon the site which is still incompleteat Arumbakkam. Notwithstanding the higher levels of imputed rents reported by the Region, counting the economic benefits only from the year in which they actually occurred would bring the rate down significantly. The ERR on the two othersites and servicesschemes would be even more affectedgiven their much lower occupancyrates. All this may have been foreseenat appraisalbut the appraisal report containsno projections on this aspect of the project. 20. As far as the slum improvement componentwas concerned, MMDA's PCR reportedthat, insteadof the 85 intendedat appraisal, only 53 slum areas in Madras had been improvedunder MUDPI, but that 24,862 householdsbenefited; more than the 23,000 targeted at appraisal. Althoughthe projectdescription also specifiedthe number of slums and the land area to be improved (see PPAM, Annex A, Section B), TNSCB evidentlysaw the target exclusivelyin terms of the number of beneficiaries. The question raised by fewer slums having been improved,however, is whether there was in fact a reductionin the component's physicalscope (without which even more households could have benefited),or whether the 53 slums improved,through coveringas extensive an area, would have requiredthe same physicalworks for their improvement as the original85 would have done. Physicalspace is an important variablein determining the design of the kind of infrastructural works installed during slum improvement whose costs depend, to a large extent, upon the area to be serviced.  - 13- 21. Unfortunately, the audit mission was unable to obtain data on the total area eventuallyimprovedby the component, nor was it possibleto make estimatesfrom field visits to the (still)numerous sites spread throughout the MMA, With inconsistent nomenclature, incomplete data and differingpopu- lation estimates of the same slums, it is difficultto quantify,on the basis of the information made available,the component'sphysicalaccomplishments. Similarly,it is difficult to determine exactly what improvements were to have been provided to each householdor their estimatedcosts and how the improvements finallymade compared in extent an; cost to the originalinten- tions, This also raises issues regardingthe reestimatedERR for the slum imptovementprogram which the PCR reckons to be 23%; it is difficultto see how such a number can be calculatedin the absence of detailed information. There is also the more general issue of monitoringand control of projects with numerousand physically dispersedinvestments,and highlights the need to make full use of reliable(but simple)management information systems,which, in the case of MUDPI, could have kept MMDA (and consequently the Association) more fully informedon progressand the final achievements of each component, in particularslum improvement. Similar unansweredquestionsremain with regard to the details of executionand costs of the three sites and services schemeswhich precludethe making of a completeand effectiveevaluation. As it is, the PCR does not contain information which could be readilyrelated to the Project Description(PPAM, Annex A) on the various elements,including off-siteworks of the sites and servicescomponentnor on the number of lots sold, built upon (and by what method and with what option)and occupiedunder this element nor on the exact improvements which were made in the various slum areas and for other projectelements. 22. In any event, that nearly 25,000 householdsin Madras'slums bene- fited from MUDPI is a significantcontribution representingas it does an improved living environment for some 10% of the total slum population. Nevertheless, as establishedslums were being improved, other new unes were being formed. Even if the intensiverates cf slum improvement under MUDPI were to be maintained,considerably more time, by rough calculationat least  - 14 - 30 years,7 / would have to elapse before Madras' slums would all be improved; this contrastswith the appraisalprognostication of completingthe task by 1985 that has already been commented on (PPAM,para. 11). While 30 years may appear to be a long time, the very fact that it is possibleto make any esti- mate at all is in itselfa tribute to the effortsand commitment of GTN agen- cies, which with the Association's support,have approached the slum phenome- non in a realistic way, Only in exceptionalcases like this (Jakarta, Indonesia is anotherwith its Kampung Improvement program. See PPAR No. 4620 for an audit of that project),is it possibleto contemplate any finite time horizonat all for such a long-term goal achievement, since the growth of new slums and squattersettlements commonly outstripsthe capacityand willing- ness of local authoritiesto deal with them. In M&dras, MUDPI (and its follow-on project,MUDPII)can be creditedwith having given a major stimulus to this important achievement. 23. The several improvement works inspectedby the audit mission were generallycompletedto a high standard. In some cases, however, inadequate attention had been paid to land contours and this might lead to serious drainage problemsand perhapseven floodingduring the monsoons (as the site visits were made during the dry season there was then no problemwith drain- age). The success of the sanitary installations in the slums was only limited. While the audit mission was informed that the design of communal W.C. blocks for use by a number of familiesmight be incompatible with cul- tural traditions, it was in any case impossibleto put them to the test since there was iisufficient water to sustain their proper use (see PPAM, para. 12). The road and footpath infrastructure, which was generallycompletedto 7/ The estimateis based upon the followingparameters. The existingslum populationis taken to be 250,000 households. About 175,000 of these have already benefitedfrcm improvement or redevelopment (25,000under MUDPI + 30,000 under MUDPII + 100,000under TNSCB schemes not financed by the Association) so the current backlogis 75,000 slum households in need of improvement. In addition,a program destined to improve all slums would have to take into account the formation of new slum households,estimated at 10,000 householdsper annum (i.e., the same rate as the annualgrowth of MMA's population as a whole). Having taken 14 years to attain the significant achievement of improving the conditions of 175,000slum households, one may concludethat the overall programbenefited, on average, 12,500per annum. If this high level of improvement "output"were to be sustainedin a continuingprogram,only 2,500 units of it could, in zeality, be considered as tackling the existingdeficit (of 75,000unimprovedhouseholds), since the majority, 10,000units (12,500minus 2,500),would be simplykeepingpace with the formation of new slums. Accordingto this simple calculation, it would take at least a period of 30 years (75,000dividedby 2,500) to improve all of Madras' slums, at the end of which a permanentprogram would be needed just to handle the continuinggrowth of the phenomenon. The calculationhere is to illustratethe process in hand rather than to make preciseestimates.  - 15 - a high standard of finish, on the other hand, appeared to have better use made of it. Now that the infrastructure Is in place, however, maintenance becomes an Important issue, especially given the reported reluctanceof Madras Corporation to accept the responsibilityfor maintainingmany of these areas through its own lack of finance to do so. During site visits,regular maintenancedeficienciesdue to worn pavements,for example, were evident, althoughno case of major structuralfailureitself was observed. Since the maintenanceissue affected other components,it is treated as a point of special intereatlater in this report (PPAM,paras. 51-53). 24. Notwithstanding the large scale of MUDPI's successful slum improve- ment program, this kind of scheme does not increase the overall supply of housing able to meet the extra shelter needs in a growingdeveloping country metropolisas some Bank documentsimply. The PCR (para. 6.03) refees to an increase in the supply of "environmeatally acceptable shelter", which includesa slum improvement componentof 15,000 units, as going some way to satisfyingthe potentialdemand for shelter,but this representg essentially a reclarsification of that part of the existing housing stock that was improved,and would inevitablybe matched by a corresponding drop in the supply of "environimentallyunacceptable shelter". Withoutaiming to diminish the importanceof slum improvementand the successfulresults obtained it. this case, the audit wishes to stress that slum improvementis aimed at improvingthe environmental qualityof existingslums, itselfa most laudible objective,to providebetter living conditions for the urban poor who already have some kind of shelter, albeit in appallingconditions. After all, any improvement has, by definition, to be made to somethingthat already exists; in this case, an area with existing slum hutments which, notwithstanding their poor condition,do make up part of the housing stock and do provide housing services,however precarious,to the families that already live in them. How, therefore, can such slums (which already exist and are occupied to the point of overcrowding) be regardedas an additional supplyof shelter capable of meeting the potentialdemand for much needed extra units? The increasedsupply of the latter is forthcoming from the self-helpeffortsof the poor themselveswho, against all odds, still provide at least an extra 7,500units a year in Madras. Only insofaras these slum dwellersthemselves expand their houses, subdivideor build extra rooms, can a housing supply increasebe said to have occurred. This result, however,has not been fully monitored;this is somethingthe audit considersshould be pursued. In the meantime, it should be emphasizedthat it is not correct to count all the hutment units affected by improvementto constitutean increasedsupply of shelter. Because of its importanzain bringina out the relationship of the slum improvementprograms to the housing stock of Madras, this aspect is dealt with in more detail later in this report as a point of specialinterest (see PPAM, paras. 33-37). 25. Although the PCR is correct irnaffirmingthat MUDPI's employment componentswere small and uninfluential in quantitativeterms (PCR, para. 2.05), the experience them was nevertheless of implementing an important one in that the provisionof factorysheds in the sites and servicesschemesand the financialand technical support to cottage indus:ry in the imp.oved  - 16 - slums, if replicatedon a large scale, could lead to significant employment creation. The Small IndustriesDevelopment Corporation of Tamil Nadu (SIDCO) was effective in the provisionof these sheds and was able to bring its experienceto bear upon the project. The PCR reported the failure of the cottage industrycomponent(PCR, para. 2.05),althoughthe audit mission was able to visit a number of successfulbusiness operationsin some improved slum areas. But, as the PCR points out (ibid.),the employment component as a whole has not been truly evb..uated, so that reliable data is not forth- coming on the number of businessesserved, how many of these were new ven- tures, the number of jobs created,capitalcost per workplace and the general performance of the businesses estahlished. The audit mission was unable to explore these matters in the contextof Madras' employmentneeds owing to the reluctance of the local authoritiesto allow the necessarycontactsto be made. In spite of this, it was evident that an important constraint to the implementation of such employmentcomponentsin Madras would be the limited local capacityto fully appraiseindustrial projects,something that was con- sidered to be, so the audit mission was informed, a function of commercial banks. Given that income and employmentare such important aspectsof urban development, and yet the attainmentof these "elusive"goals through Bank Group supportedurban projectshas proved so difficult,(Learningby Doing, op.cit.,page 8), specialattentioncould be given to the successful aspects of the limited MUDPI experienceand the constraints upon the scope of its impact. 26. Of all the projectinterventions, the bus transportcomponentwhich 'provided buses and servicefacilities to PTC, is generally reported to have been one of the most effectiveand successful. The speed of disbursements and the rapid implementation of the componentreflect the efficiencyof PTC as a key MUDPI executingagency. With the benefitof the component, PTC was able to further improve its operationsand to raise the utilization of its fleet of 2,100 buses to 87.6x. With a more modern and bettermanaged fleet, latest figures on breakdowns show them being responsiblefor only 48,000 trips lost per year in contrastto 79,000 only four years ago when the fleet was actuallysmaller. (Datataken from PTC Annual Reports.) In spite of the considerableimprovementsin its operating efficiencyand consequentcost savings,PTC was still unable to meet the covenanted ratio of 95 (revenueto operatingcosts)without the necessaryfare increases againstwhose enactment there were severe politicalconstraints. Only in May 1985, almost a year after the formal suspension of disbursements on MUDPII for nonobservance of the relevantcredit covenant, did GTN increasefares by 33X, therebyenabling PTC to achieve the agreed rAtio. Although ostensiblyconcernedwith the issue of cost recovery,the Association's resort to legal remediesto ensure compliancewith the (MUDPII) project agreement also served to strengthen its credibility as a serious interlocutor in urban policy. At the same time, it provided an opportunityfor the Governmentto take a politicallyunpopular decisionand therebyhelped to avoid the decapitalization of one of its most important agencies. To judge from press reports on this matter in India, these achievements may have been at the cost of some publicrelationsdamage to the Association in that country. The issue itself was quite beyond the controlof PTC which, it should be stressed,performedthe role expectedof it by the projectmost effectively.  - 17 - 27. A complete assessment of the physical accomplishments of the remaining transport componentswould have required more informationabout their implementationthan was readilyavailableto the audit mission. As in the case of the numerous slum improvementschemes, the many and physically dispersed transport investmentscall attention to the issue of project control and the need to use an effectivemanag..anentinformation system to allow the progress of the works to be monitored (PPAM, para. 16). In the absence of such a system, the audit mission was able to follow up the physical accomplishmentsof these componentsonly through a series of site visits. 28. As far as the Inner Ring Road was concerned,progress has been somewhatdisappointiag. The road is not completeand formidable and perhaps insurmountable land acquisition problems hinder the formationof a complete link. All the road intersection improvements,however,have been completed to a high standardand are well used, with the only major difficultynoted being with the grade separations that replaced the railway level crossings, where animal drawn vehicles and bicycles move up the steep gradients so slowly as to cause considerable traffic congestion. The assessmentof the physical accomplishment of the cycle tracks and footpaths componentwas hampered by the difficulties of locating the works. In general,cycle ways in Madras are very poorly marked and little respected by road users in general, since they consist of short stretcheswhich are frequently blocked by all kinds of obstacles. In order to avoid the difficulties of walking along footpaths which in many instancesalso are obstructed,pedestrians themselves are forced to used the cycle tracks,while cyclistsmove over to occupy the main road carriageways. The result is intensivecompetition for limitedroad space between all kinds of vehicles and pedestrians and only a fractionof the costly,heavy duty roadwayfinds itself in fact being used by motorized traffic. The cycle tracks and footpaths provided under MUDPI seemed to have had very little effect upon this situation. The failureof this component,the audit mission was informed,was due to the general lack of maintenance and controlof this kind of infrastructure in Madras. In this specific case, the possibilityof the lack of maintenancehaving led to a prematureand total loss of some of the project'sphysicalassets is dis- cussed as a point of special interest later in this report (PPAM, paras. 50-51). 29. The continuingmajor water supplyproblemsin Madrasdo not make it possible to judge the impact of the water supply and sewerage componentof the project. As has already been mentioned(PPAM, paras. 17 and 23) water shortagesprejudicedthe performance of some of the project'ssheltercompo- nents. The solution of this problem, however, lies beyond the scope of MUDPI, given the inadequacyof the present source of supply from the Red Hills Lake and the age of the limiteddistribution system, 40% of which is over 50 years old. It will not be until the appropriate major measures are fully implementedin the sector, that other urban investments will cease to be the victimsof this very serioussituation.  - 18 - D. Institutional Development 30. The PCR deals with the main institutional developments arising out of the project as they affected the principalproject agencies (PCR, paras. 4.01-4.07). The audit concurs that the institutional buildingachievements of the project, in strengthening these agencies, were considerable. Technical assistance provided under MUDPI, especially to MMDA and the Madras Corporation (MC), helped reinforcethe on-the-job experienceafforded by the project. In all cases, the efforts were concentrated not upon the creationof new agencies,but upon improvingthe performance of the existing ones. It is worth recalling, in this respect, the reputationfor sound administration enjoyed by GTN and its specializedagencies (PPAM, para. 1). Such was the case of PTC, for exa;mple, a well run bus company prior to the project ?CR, para. 4.05), whose management and operatiunalsystems MUDPI helped t' perfect. MC itself, was successfullyassisted by the project through the introduction of new managementtools to improve its accounting systems even though, along with many traditionallocal authoritiessome organizational problems remain. Both TNHB and TNSCB were also given the opportunity by the project to strengthen their planning and management activities, althoughthe audit considersthat the PCR somewhatoverstates the degree of change that affected the latter agency's managementand policy (PCR, para. 4.07). The audit mission was informed,for example, that TNSCB at times found its capacity overstretched in order to meet current targets. This suggeststhe desirability of an assessmentof the entity'sstrengthsand weaknessesand of the need to develop it further. At the same time its work could be complemented by expandingMC's role in slum improvement, given that the Corporationitself undertookworks in 12 of the 53 slums improvedunder MUDPI. The performance of SIDCO, which was in charge of the factory sheds program,especiallyin relation to its capacityto fully appraise industrial projects,is also worthy of furtherstudy,given the important though limited experienceof employmentcreation through the MUDPI component (see PPAM, para. 25). 31. A particularlysignificantinstitutional experienceof MUDPI was the role of MMDA as the agency responsiblefor overallprojectcoordination. The managementof urban projects,whether, it should be entrusted to the traditional local authorities,or whether a specialagency should be created to oversee it, is a constantissue in their design. The Madras experience was especiallyinterestingin this respect as it steered a middle course betweenthese two extremes. Becauseof its implications for other Bank Group urban projectb, the institutional developmentof the MMDA and MC in the metropolitan managementof Madras is discussed later in this report as a point of specialinterest(PPAM,paras. 45-47).  - 19 - III. POINTS OF SPECIALINTEREST 32. Because of the issues they raise and the possible relevance of their lessons for future Bank Group projects,attentionis drawn here to a series of points of special interest. They range from policy and general sector issues to specificexperiencearisingout of the project. A. Relationof Projectto Madras'ShelterNeeds 33. The two MUDPI shelter components,sites and services and slum improvement and sites and services,were more than just differenttechniques; the two approachesaffected the housing needs of Madras in quite different ways, an importantdistinctionnot always brought out in the project docu- ments. Basically, the sites and services component contributed to an increase in the quantity of shelter supplied through the provisionof ser- viced plots,whereas the slum improvement component raised the qualityof the supply, but not the total supply itself since the housing units (i.e. slum huts) in the areas improved already existedprior to the program. 34. A global analysisof Madras' housingstock can help demonstrate the differentimpact of the MUDPI componentsupon the shelter needs of the MMA. A starting point for such an analysisis the housingstock which consistsof the shelter sector'sphysicalassets (i.e. houses, apartments and huts etc.) able to supply shelter services to the populationof Madras. The latter is constant in the short run, since as well as being fully occupied (to the point of overcrowding in the case of low incomeresidents), these assetsare themselves fixed in the short run. To increasethis supply, it would there- fore be necessary to provIde additional units of shelter either through building new ones or subdividingthose that already exist. On the other hand, shelter supply can fall when ageing (or simply inadeq'atemaintenance) of the existing stock can make units effectivelyuninhabitable. Urban redevelopment schemes can also reduce supply insofar as they do not replace houses that are demolished. Although the demand for shelter is in reality split up between several parallel markets (segregatedby income and other factors), in its simplest form, the potentialdemand for shelter services consists of the total number of householdsrequiringshelter. The simplest way to measurean increasein this demand is throughthe growthof new house- holds seeking shelter in Madras either as newly formed familiesor migrants from elsewhere. In Madras,as in general,demand exceedssupplyand leads to what is termed a housingdeficit in the form of homelessnessor overcrowding, the latter being numericallymore important in the MMA.  - 20 - 35. In this respect, MUDPI's slum improvement componentcan be under- stood as having improvedthe qualityof the city's existinghousing stock by providingbettrr conditionsfor a good number of existing shelter units, 8/ while MUDPI's sltes and services componentsincreasedthe quantity of the housing stock as a whole throughthe supply of additionalunits. It would be quite incorrectto consider that slum improvement is capable of increasing the supply of shelter and thereby meeting the potentialdemand for extra shelter (see PCR, para. 6.03) if there is no increase in the number of units in these already overcrowdedareas. Even if qualifiedas an increase in the supply of "environmentally acceptable"shelter, it could still not satisfy the need for extra units of shelter since for each extra unit of "environmentally acceptable"shelter there would be one corresponding unit less of "environmentally unacceptable"shelter in a processof what is effec- tively a reclassification of the housing stock (see PPAM, para. 24). This does not mean that sites and services,for increasingthe supplynecessarily makes a more important contributionto solving shelter needs than slum improvementdoes, since each approach serves a differentpurpose and both would be needed if shelter policy is to aim correctlyat both increasing the quantityand improvingthe quality of shelter available,especiallyfor low incomeg.oups. 36. As far as the need for additionalshelter is concerned,the 4% annual growth of Madras' population(using 1981 figuresas the base) implies a potential demand for some 40,000extra units of sheltereach year (assuming six persons per householdand one unit per household). This would be on top of the extra housingneeded to overcome the existingdeficit. Bank staff's own estimatesprojectthe annual supplyof formalhousing(publicand private sector)at 13,500units, only one-thirdof the potential demand. Some 5,500 of this would be supplied by the private sector and the remaining8,000 by TNHB (of which 3,000are in the form of sites and services). The latterout- put would be higher than the annual provisionof sites and services under MUDPI which averaged 2,250 units over six years. However,due to the lag between site preparationand the constructionand actual occupationof a habitablehouse, only 33.3% of MUDPI sites are in fact supplying shelter services (PPAM, para. 16). In the absenceof specificdata upon the actual output achievedby TNHB (instead of the target figurespresentedto the audit missionand reportedin the PCR, Highlights, 4th paragraph and subsequentGTN orders requiringtheir being met), one must reserve judgmentupon the degree to which TNHB can further increaseits supply as a result of the introduction of the more cost effectivesites and services. But what is clear, neverthe- less, is that, even at the levels projected,almost half of Madras' extra potentialdemand for shelter each year will still have to be met by the 8/ When demolitionand displacementof existing householdsis involved, slum improvementcan actually lead to a reductionof the total housing stock. While data is not available on this point, MUDPI slum improvementsdo not appear to have involvedmany demolitions, which in any case were limited to 5% of the houses in a neighborhood by project agreements.  - 21 - Informal sector, either through self-helpbuilding of more slums or by the increasingsubdivisionof the existing ones leading to more overcrowding. Given the importance of the informal sector in satisfyingthe shelterneeds of Madras' low income population,it would be appropriateif future urban project designs consider the orientation of this spontaneous settlement through the provisionof partiallyserviced plots (at even lower cost than coventionalsites and services) such-as is done through the "progressive urbanization" policiesin some Latin Americancountriessuch as Colombiaand Nicaragua, for example. Such an approachwould be an essential complementto the sites and servicesapproach initiatedunder MUDPI which cannot by itself providean answer to the shelterneeds of Madras' low income population. 37. In order to fully assess the contributionof MUDPI'sslum upgrading to the improvement of Madras' existing housing stock, it would be necessary to have full detailsof the pre and post projectsituations as they relate to the physical conditionof the shelter units - their immediateenvironment, the servicesprovidedand most importantly, the socio-economic conditions of the population. Unfortunately, precise data are not available on either situation. Such information would be necessaryto gauge the cost-effective- ness of various slum improvement options, for example, to judge their effi- ciency in execution and to evaluate them againstother means of improvingthe quality of housing. Officialcensus figures for 1971 reported that 58% of the stock was permanent(pucca)housing in good condition, while the remain- der was inadequate. Given that there has been an annual shortfall of housing supply against increasesin population, there is every reason to expect that the overall conditionof the housing stock will have tended to worsen over time. Nevertheless, if it is assumed that the current slum populationof Madras consistsof 250,000households, various improvementschemesundertaken since the early 1970s or currentlyunder way will have affectednearly three quarters of the slum households in the city (10% under MUDPI, 20% under MUDPII and 40% under prior schemes not financedby the Association). These results speak for themselves, and must compare very favorablywith what has been achievedin other developing countries' metropolises. B. Policy Impact 38. Madras is probably endowed with a more appropriateand realistic set of shelter policies, with their priority for low income groups using cost-effectiveand replicable solutions, than many other cities. The influenceof MUDPI on the formationof these policiesis an important aspect to examine,firstly, because the policy impactbeing perhapsthe broadest and longest term consequence of this as of any development projectand, secondly, since much attentionhas been drawn to it in the PCR (Highlights, p. v). Identifying a change in policy in any countryand even more the causes of the change cannot fail to be a matter of debate when it may not even be possible to agree on what parameters should be used; should they be policy statements and plans or are these nothingmore than empty declarations of intent;budget profilescould be :evealing but are they amenable to analysis; the physical output of particular programscould be meaningfulbut do they contradict the agency's ty-ueintent? In order to assess the policy impact of MUDPI, it is  - 22 - necessaryto give a full consideration to the shelter policiesthat were in force prior to the project and also to be quite clear about what has been achieved in the current set. In this respect,the audit considersthat the PCR does not provide a complete picture of the Madras policy towards slums that was in force prior to MUDPI. 39. In the five year period before MUDPI (1973-77),TNSCB's policy towards slums (measuredby the physical output) was to clear and rehouse 3,497 households a year and improve slums to benefit another 9,791 (figures taken from Madras StructurePlan, Vol. I, p. 47). Becauseof the much higher unit cost of slum clearanceover improvement, financialcommitments heavily favored the former (prior to MUDPI), notwithstanding the smaller number of beneficiaries. Recent (post-project) financialdata were unfortunately not providedto the audit mission. In the latest year for which data were avail- able for the audit (FY1983/4), TNSCB's policy was to clear and rehouse 1,010 households and to benefit another 15,750 by improvement. Such figures clearlyindicatea shift of poLicypriorities by the fall from 3,497 to 1,010 householdsaffected by clearance schemes and the increase from 9,791 to 15,750 households benefiting from improvement. This shift is a notable achievementwhich can be credited to the MUDPI experience where project covenantsexplicitly called for restrictionof expenditureon costly slum clearance,and the perfectedslum improvement componentgave greater credi- bility to the approach. In the opinionof the audit the shift was made pos- sible through there already existing some political commitment to slum improvementas evidencedby the fact that 48,954 slum householdshad bene- fited from what were called Environmental Improvement Schemes (EIS) prior to MUDPI. The Region argues that these schemes were a waste of money for the facilitiessoon ceased to operate for lack of maintenanceand little or no home improvement took place becauseof lack of tenure. In the opinionof the audit, such failurewould not indicatea lack of policycommitment so much as inadequate concept of slum improvement(e.g., without tenure), and perhaps ineffective implementation and maintenance. It was preciselyin overcoming some of these problemsand making such an approachmore viable that the major contribution of MUDPI lies. But that was only made possiblebecause of the initial politicalcommitment, whatever limitations there might have been in its implementation. This is a very importantpoint to make since it would not be reasonable to expect to obtainresultsas satisfactory as in Madras in another city or country if there were resistance on the part of the Governmentconcerned to the proposed approach. Thus instead of an almost completereversal of policies on slums, as implied by the PCR (Highlights, p. v), the audit considersthat the policy impact of MUDPI could be better understood as a shift of prioritiesin favor of slum improvement whose con- ception was at the same time considerably enhanced particularly through the introduction by The project of the concepts of security of tenure and cost recovery. Of course, there is no doubt that the curtailment of slum clear- ance schemes, uith their excessively high construction standards and wasteful use of resources,achievedby MUDPI made an importantcontribution to this shift, but that is only part of the story. In this approach to the policy impact evaluation,the audit wishes in no way to diminish the pro- ject's achievements, but instead to identifythe key facets that would need  - 23 - to be reproducedelsewhereif these importantachievementswere to be repli- cated, especially in cities in other countries where similar commitments might be identified. 40. The sites and servicescomponent is creditedby the PCR with having increased the potential output of the TNHB more than threefold (PCR, Highlights,p. v) for its being more cost effectivethan conventional pro- grams. While TNHB's total output has in fact increased,this cannot be exclusively attributedto the introduction of sites and services; the output has also gone up because the expansionof other types of housing as funds from the Housing and Urban DevelopmentCorporation(HUDCO) rose in recent years which has enabled TNHB to significantly raise its overall expendi- tures. Thus TNHB's total investment in 1982/83was Rs. 22.0 crores in con- trast to the Rs. 6.0 crores level of 1976/77on the eve of MUDPI. What the project succeededin doing here, and this had a greater policy impact than the slum improvement component, was to provide a feasibletechnical alterna- tive for TNHB that could, if so desired,be applied to more closelyattend the needs of the lower income population of Madras. MUDPI's sites and ser- vices effectively demonstrated solutionsfor the need for low income housing in Madras that were cost-effective,affordable and involved full cost recoverynotwithstanding the lack of subsidies. In this respect they offered a viable alternativeto TNHB's traditional approach of complete units that favored higher income groups and thereby induced an importantpolicy shift that favored higher income groups and thereby induced an importantpolicy shift. 41. While the engineeringof sites and services was not new to TNHB which had formerly prepared them in the form of what it called "open developed plots", the project's innovationwas to demonstratethe use of sites and servicesfor low income housing. In this respect,it was success- ful in leading GTN to direct TNHB to concentrate 80% of its housing invest- ments for the benefit of the lower income groups (PCR, para. 4.06). In 1976/77, the last year prior to MUDPI, TNHB had an actual output of 2,043 housing units of which some 35% were for low incomegroups. By contrast, the audit mission was informed that the planned output for 1984/85 was for a total of 12,000 units of which 83% would be destined for the lower income groups. If achieved and sustained,such a result would representa major change of housing policy attributable to MUDPI, althougha definitive conclu- sion would have to await an evaluation of the results of TNHB's actual out- put, the data on which was not made available to the audit mission. The Region informs that GTN issued a specificorder to TNHB to implementits 5 year plan which focusses TNHB constructionupon land servicing(sites and services),schemes of which many are to be developedin other major utban areas of Tamil Nadu also. The slow development of two of the three MUDPI sites and services components(PPAM, para. 22), however, suggests that a longerlead time may be necessaryto achieve the radical improvement in per- formance necessary to benefit these lower income groups to the degree expected.  - 24 - C. Problemsof Land Acquisition 42. Especially in Madras itself, the audit mission was frequently informedhow lengthy land acquisition procedurescaused delays to the physi- cal start up of the shelter and employmentcomponentsin particular. Some transportcomponentswere affected too; one of the reasons, the mission was told, why the Inner Ring Road still did not constitutea continuous link is protractedlitigationover land required for its completion. Accordingto the PCR, land acquisitiontook three times longer than expected,and the delays have persistedthrough to the follow on project, MUDPII (PCR, para. 3.02). The gravity of the problem becomes apparentwhen one considersthat seriousdelays are not only restricted to the acquisition of privateland but also occur in the transference of public land between Governmentagencies. This was the case of the Villivakkan sites and servicesschemes,for example, where the land was only transferred to TNHB in February 1981,some three and a half years late, even though it was alreadyGovernment owned,althoughpre- sumably by a differentdepartnent or agency. As far as privateland is con- cerned,the fact that acquisition procedures can take three to four years has been attributed to the antiquatedprovisionsof the relevant legislation, namely the Land Acquisiton Act of 1894. Under this Act private landowners, if they so wish (and often do), can challengeeach step of the proposed expropriation through protracted court proceedings. This situationhas led to the suggestionthat within the existing legislation the emergencyprovi- sions of the Act be made available in the case of housing sche,nes (see StructurePlan for Madras Metropolitan Area - Vol. 1, page 53). Even if the proceedingcould be streamlined,however, land acquisitionwould remain a very expensiveopcion for local authorities since compensation under the Act is fixed at market rates which presumablywould rise rapidly at the simple announcement of the intenticnto expropriate, especiallywhen much land is monopolyowned (see Gupta, Devendra, op.cit.,p. 119 and p.125). Even though MMDA has taken steps to set up special land acquisition units (PCR, para. 3.02),the problems to overcome remain formidable and their solutions possi- bly go beyond merely administrative remedies,being rooted as they are in the very nature of the real estate market itself. 43. While the procedures themselvesmay not be fully adequate, any dis- tortions in the urban real estate market can contributeto the difficulties in acquiringland for public sectordevelopment and can also help explainwhy the private sector owners themselvesdo not develop their own land. The Madras StructurePlan itselfalludedto such difficulties and calledfor fur- ther study of the land ownershipconstraintupon the supply of urban land (Structure Plan for Madras MetropolitanArea - Vol. 1, page 133). Given the difficulties experienced, Madras could benefit considerably from a complete legal and fiscal cadastre to provide adequate information on land holdings. With such information it might be possible,for example,toapplythe existing Urban Land Ceiling Act more effectively(see PPAM, para. 3), or, alterna- tively, induce private land which remains vacant for speculative purposesto be brought into use through taxation at higher rates. As matters stand, TNHB's current large scale land acquisition program, whereby it intends to acquire some 3,500 ha of urban land, (equivalent to nearly a quarter of the  - 25 - total residentialland in MMA in 1974) might not be feasible and may be counter productive because it would reinforce monopoly power among landowners,force land prices up further and provide windfallcapital gains at the expenseof the publicsector. 44. Whateverthe completeexplanation of their causes,the difficulties experiencedin acquirinig land for the project componentsled to considerable delays in executiontherebyincreasing projectcosts which contributed to the 19.8% cost overrun. Furthermore, progress in granting lnd titles, in the form of lease-cum-sale (LCS) agreementswas slow. By February 1983, some 9,746 LCS agreementshad been made, corresponding to only 40% of the total householdsin the improvedslum areas. There can be no doubt that project performancewould have improved had there not been such difficulties with land acquisition.9/ It has been suggested that, in future, all land requiredfor this kind of projectshould be acquiredprior to presentation to the Association's Board. While project implementation would apparentlybe much faster if land acquisitionwas formally excluded from this phase, the real gains for the urban development of Madras would be minimal since such a change would also mean that a project could be presentedto the Board only much later. In any case, it is most importantfor the local authorities have completecadastralinformation about the city's land (in particular, wlioowns it and what its value is) to enable them to effectively assume their respon- sibilities for metropolitanreal estate management. Only through a full understanding of the workings of the land market (especially its monopoly aspect), will it be possible to devise a more effectivearrangementfor Governmentintervention in the housingmarket. With such an understanding, a judgment will need to be made to what extent the Government should seek to own land and develop it (and whether land acquisitionproblems are to be judged in that context)and to uhat extent Governmentpolicy should be based on a fuller participation by the privatesector. After all, in Madras,it is on private land that more than half of all slum households live and where it has provenmost difficultto undertakeimprovements preciselybecauseof the land issue. Immediately behind the problemsof land acquisition experienced through MUDPI lies the question of how to improve metropolitan real estate management,and beyond that, the more general issue of what should be the respectiveroles of the public and private sectors in the urban development of the MMA. MMDA are exploringto includealternatives to land acquisition includingGuided Development and Land Readjustment. Acquisition of land from private owners would not be required, as owners would in effect become partners in the development (or redevelopmentin the case of slums) of specialized areas. 9/ As the Region has pointedout, there were other causes for slow progress with LCS agreements, includingGTN's reluctance early on to grant them.  - 26 - D. MetropolitanManagement:the Role of SpecialAgenciesand Authorities Traditional 45. It is widely believed that, in metropolitan areas in general, traditional local authorities have difficulties adaptingto the new tasks of management that a rapidly growing metropolisrequires. Thus, instead of relying upon the existing municipaladministrative machinery,it is often felt to be more efficient to start afresh with a completelynew special agency or increasethe responsibilities of an existingone. Such an option raises a number of importantissues. Firstly, although they can indeed be more efficient,such special agenciescan sit uncomfortably between central, state and local authoritieswith ambiguouspoliticalauthorityand unclear terms of referenceoften overlappingthose of the traditionalauthorities. Furthermore, as a more centralizedbureaucracy,a special agency may find itself more remote from the local populationthan municipal government, althoughit could also be less prey to specificunrepresentative local inter- ests. In addition, when such agenciesare remitted to only implement a par- ticular policy or project (which they may be able to do morc afficiently), the issue of who will be responsible for long term maintenanceand controlis left open. Specifically from the Bank Group's pointof view, the implementa- tion of urban projects can be much more straightforward when dealing with only one special agency than a multitudeof traditional authorities pursuing gcals that may be out of tune with Bank thinking on the development of the urban sector. It is particularly interestingto considerthe experienceof MUDPI in the context of this debate since it strengthenedthe role of a special metropolitan agency (MMDA) while at the same time actually led to some institutionbuildingwithin the traditional municipalauthorityof the Madras Corporation (MC). 46. MMDA had existed on an ad hoc basis since 1972, some five years prior to appraisalof MUDPI, althoughit was only given statutory backing in 1975. The fact that is was thus already incorporated within the system of metropolitan management helps explainhow it was able to successfully coordi- nate the implementation of the project. Its responsibilities included the preparation of surveysand plans for the MMA, overseeingthe area's land use and the assembly of land for its development. Prior to MUDPI, MMDA had alreadyaccumulated experience in exercisingthese functions with some effect and had become quite an influential agency in its own right. The basis of its influencewas its ability to negotiatewith sectoraland local agencies in the metropolitanarea whose interestswere directly expressedto MMDA throughtheir being represented on the Authority's Governing Board. In this way, MMDA has avoidedthe criticisms of being excessively powerfuland over- bearing levelled at the CalcuttaMetropolitan DevelopmentAuthority by tradi- tioaialauthorities in West Bengal, for example (see PPAR No. 4023 on First CalcuttaUrban Development Project).The audit mission was impressedby the way in which the MMDA had successfully participatedin the development policy process of Madras withoutseeminglyhaving incurredmajor antagonisms on the part of potentiallycompeting sectoral and traditionallocal authorities. One of the lattercould have been MC, the municipalauthorityfor the city of Madras which, althougha candidate for being weakenedby the conso,idation of  - 27 - a strong metropolitanagency (i.e., MMDA), instead was strengthenedand managed in fact to improve its own institutionalperformancedirectly as a result of project technicalassistance(PPAM, para. 30). While a lot more still has to be done, especiallyto build up MC's capacity to handle the maintenancerequirementsof the physical assets of MMA and to strengthen MMDA's programingcap&-ity,the simultaneous institutionbuildingin the two types of authorityis . notable achievementto which MUDPI made an important contribution. 47. The MUDPI experience has demonstrated that a projectneed not make a clear-cutdecision between opting for a special agency or reliance upon traditional authoritiesfor implementation. In the Madras case, one impor- tant ingredientto the success of the simultaneous institution building in two potentiallycompetitiveagencies was, no doubt, the history of sound public administration for which GTN is known (PPAM,para. 1). Where institu- tions are weak and respon3ibilities unclear, they might feel that their powers could be easily usurpedby any new arrangement of metropolitan manage- ment and might resist attempts at coordination. In Madras, where institu- tions are stronger,this does not seem to have happened. In this respect,it is significantthat MMDA has not been given responsibility for controlling sectoralbudgetsin the metropolitan area, a move which could radicallyalter the present power structure of the metropolitan managementsystem. In order to sustain the basis of negotiationwhich has so successfully prevailedso far in Madras' metropolitan planning, the audit considers that such a func- tion of budget controlshouldremain with the highestauthority of GTN. What is needed is for MMDA to enhanceits role as planningagency so that it could monitor metropolitaninvestmentand help identifyeconomicopportunities in the MMA. For this it would need to strengthen its capabilities for under- taking economic analysis. At the same time, efforts must continue to con- solidate the traditional authorities, especiallyMC, so that they can fully cope with their responsibilities to maintain the fixed public assets of Madras as well as undertaketheir own programs. It remains to be examined whether it is possibleto replicateelsewherethe effectivemetropolitan man- agementmodel of Madras. E. with Cost Recovery The Madras Experience 48. At appraisal,considerable attentionwas paid to cost recovery,in particular for MUDPI's sheltercomponents (SAR, paras. 4. 10-4.16), and refer- ence to the introduction of the conceptof full cost recoveryfor key invest- ments was includedin the PCR as a re-statement of a basic objectiveof the project (PCR, para. 2.07). A, far as the sites and services componentis concerned,there have beea some positiveresults in this respect,since TNHB has been able to build up a fund on the basis of the proceeds from costs recoveredfrom the component. However,the PCR does not indicatewhetherthe amount in the fund is in accordancewith appraisalprojectionsnor does it indicate the cost recovery performanceof the two later sites and services schemes. The Region informs, on the basis of TNHB's detailed monthly records, that on average 90% of billingsare collectedin the sites and ser- vices qchemes, and furthermore,believes that these collectionscover all  - 28 - costs given the financialplan and pricing policy adopted. Such a perfor- mance would indicate a successful cost recovery exercise and confirm the affordability of the sites and serviceslots by the beneficiaries. L-1order to make the fund accumulated into a truly revolving one as intended,however, it would be necessaryto apply it to an ongoing permanentprogram of sites and services,somethingthat remains to be done. The reason given for TNHB's inaction with the fund it holds was that it was not yet large enough to financean appropriately sized scheme which, it was felt, had to be at least as big as those implemented under MUDPI, although the audit sees no reason why smallerschemesshould not be financed out of the fund. Even recognizing that there is still an urgent need to desig- a permanentprogramin order to appl) the fund, the achievement of building it up in the first place should not be underestimated. 49. As far as other componentsare concerned,cost recovery has been less successful. Progress has been very slow with regard to the slum improvementcomponent,for example, where the MMDA's PCR reports that the beneficiaries have yet to realise that they are supposedto pay somethingfor the improvement. This failurehas been attributed to inadequateinformation at the time of the improvementand insufficientcommunity participation throughoutthe process, something that the communitydevelopmentstaff of MMDA is trying to rectify (MMDA PCR, page 32). The Region, on the other hand, attributesthe arrears to lack of politicalwill, although it noted some progress by TNSCB in this respect. The audit agrees with the Association's PCR that the revenue side of TNSCB needs to be strengthened (PCR, para. 4.07); this could be a part of a reassessmentof the agency's overall financial and operational capacity. 50. Considerable attentionhas been paid in Associationreports (and also in the Indianpress!) to the issuesof cost recoveryon the PTC bus com- ponent. The failureof GTN to raise bus fares eventually led to the suspen- sion of disbursements by the Association on the follow up project, MUDPII (see PPAM, para. 26). More than anything,that episode serves to demonstrate that cost recoveryis not simply a technical or administrative matterbut can involve importantpoliticalissues for governmentespeciallyin relation to its low income constituents in the particularcase of bus fares. In future projects of this kind a political assessmentof the feasibilityof cost recovery needs to be made as well as an estimate of the likely costs (in terms of delays,etc.) to be incurredto ensure its achievement. F. and Sustainability Maintenance 51. The maintenanceof any physicalasset has to be carriedout contin- uously and thus must be seen as a permanentactivitythroughoutthe economic life of any project. If not, in the best of hypotheses, projectperformance will be inferiorto that planned,or worse still, the assets may suffer pre- mature obsolescence, having their economiclife shortenedaltogether. Thus, lack of maintenanceadverselyaffects the economicperformanceof projects, lowering(if not making negative)their internalrates of return. Likewise, maintenance incursreal economiccosts throughout the life of a projectwhich  - 29 - ought to be computed in the calculation of their economic rate of return, somethingthat was not done in MUDPI's re-evaluationof sites and services, for example. The neglect of maintenancecan be encouragedwhen project execitingagencies,as is usual, have no specifie responsibility in their brief for maintainingthe assets they produce. Such an arrangement is also an inducementfor maintenanceto be overlookedin the actual projectdesign itself. If explicitly recognized the responsibility for maintenanceIs usually assigned to the traditional local authorities which rarely find themselvesequipped or with adequate finance to undertakethe tasks before them. 52. SinceMUDPI has only just recentlybeen completed,it is still too early to get a full grasp of the extent of the maintenanceproblem with almost the entire economic life of the completedcomponentsstill ahead of them. The Region sees some hopeful signs of solutionsin the form of the overall strengtheningof MC (PPAM, paras. 45-47) and the earmarking of revenuesfrom a surcharge on the State Sales Tax (USSST). On the ground,the sites and servicesand slum improvement works inspectedby the audit mission were generally in good physical condition; this observationsays little, however, about the state of maintenanceof physical infrastructure that in most cases had only just been completed. In fact, the missionwas informed that even routinemaintenancehad yet to begin given t,ne reluctance of MC to accept this responsibility in such areas. It is evident that MC has rneither sufficientinformation about the overall maintenancerequirements of Madras nor has it sufficientfunds to deal with those problemsit is able to iden- tify. This is a serious matter which should be addressed urgently, especiallyin view ' the fact that the type of low cost solutionsimple- mented by MUDPI are considered to need more maintenanceto ensure their physicalintegrity than higher cost, more durablesolutions do (see for exam- ple: Gupta,DevendraB., op.cit.,page 62). In the case of slum improvement especially,without constant maintenance,there is a risk of the improved areas reverting to their original conditionif the physical infrastructure were allowedto disintegrate. The apparentpremature loss alreadyof some of the footpaths and cycle tracksprovidedunder MUDPI (see PPAM, para. 28), for example,so soon after completion demonstrates the precarious nature of this kind of investment if it is not properlylookedafter beyond completion. 53. The reluctanceof MC to assume responsibility for improved slum areas parallelsexperiencein other cities. The municipalauthorities are reluctantto take over areas which they regard as contributinglittle to the tax base and where the provisionof services is costly because the infra- structuretypicallydoes not conform to the standardsof the municipality. While such standardsmay be deploredas unduly rigid, they providean excel- lent rationalefor the municipalityto refuse incorporation into its juris- diction. The establishment of independent authoritieswhich are exempt from the standardsthus only postponesa problemand projectsneed to be designed to ensure that the assets which are created will be taken over by an authorityable and willingto maintainand operatethem.  - 30 - 54. The inadequateprovision for maintenanceobviouslyalso affects sustainabilityof the project,that is, the physicalfacilities if not main- tainedwill cease generating benefits, Other sustainabilityissuesconnected with this projectcannotbe fully analyzed. Institutional capabilitieswhich have been improved should enhance sustainabilitybut institutionaloverload which is claimed in the case of some agencies may be detrimental. Cost recovery has been improvedbut whether its levels are sufficientto support sustainabilityremains to be seen. Severalproject components, particularly slum improvement, represented elaborationsof existing programsand to that extent the project clearlyevidencessustainability. Note: In its comments on the draft PPAM, the Governmentstated as follows: "FFFF. Paras. 52 and 53 of PPAM. MMDA's view is that the logistics of plann4 ng and implementinga large number of projects is not cost effectiveand full benefitscan be achieved only through projectsof reasonablesize. Smaller schemes can be handled more effectively by concernedsectoralagencies. This view may be incorporated."  - 31 - ANNEX1 Page 1 AUDIT MEMORANDUM PROJECT PERFORMANCE INDIA FIRST MADRAS PROJECT URBANDEVELOPMENT (CREDIT687-IN) ProjectDescription (Schedule2, ProjectAgreement,April 1, 1977) A. Sites and Services(to be carriedout by TNHB) 1/ The provisionof sites and serviceson about 175 ha in the Arumbakkan, Villivakkam and Kondungaiyurareas or Madras, which will includethe follow- ing: 1. Land acquisition,site preparation and provisionof on-site infra- structure includingroads, drainage and individualpropertywater supply and sewerage connectionsfor (i) about 13,470 residential plots of between about 40m 2 and 225m2 each: (ii) about 9.5 ha of commercialland; (iii) about 11.4 ha of industrialland; and (iv) communityfacilities as in sub-paragraph4 below. 2. Constructionof about 9,760 sanitary cores of which about 2,740 will, in addition,have a core housing unit with two side walls attached with attachedroofs and about 790 will, in addition,have two rooms. 3. Provision of materials costing approximately Rs. 6,000,00 for self-helpcompletionof core housing units. 4. areas of two primary Constructionin each of the above-mentioned schools, one high school, one clinic-health center and one communityhall. 5. Construction includingaccess roads, a of off-site infrastructure, bridge across Captain Cotton's Canal, a drainage course from Villivakkan to the CotonouRiver,a trunk water main at Arumbakkan, water wells with overhead tanks at Villivakkamand Kondungaiyur, trunk sewers at Arumbakkamand Kondungaiyur, a sewage oxidation pond at Villivakkamand all necessary pumping stations including equipmenttherefor. B. Slum Improvement(to be carriedout by TNSCB) 1. The provision servicesin about 85 slum areas of of basic infrastructure Madras with a resident populationof about 23,000 households(about 127,000 persons) and covering an area of about 185 ha based upon the following standards: 1/ For abbreviationsand acronyms see beginning of Project Completion Report.  - 2- ANNEX 1 Page 2 (a) one flush toilet per 10 households; (b) one bathingfacilityper 10 households; (c) one standpipeper 1C)households; (d) vehicularaccess to septic tanks servingcentrally locatedtGiletsand bathrooms; (e) vehicularaccess to within 50 m of each hut; access to each hut; (f) paved pedestrian (g) one street light per 40 m of vehicularroad and footpaths; and (h) one primary school per 3,000 persons, and one secondary(high) school per 10,000persons. 2. of: The preparation (a) sites for about 7,500 open residential plots with an area of approximately20 m2 to 40 m2 each; (b) additional sites, where feasible, for commercialand irdu3trial plots;and (c) sites for the construction of the trainingand productioncenters and child welfare centersreferredto in C.2 and D, respectively. C. Business(to be carriedout by SIDCO & TNHDC) Small-Scale 1. Construction of serviced plots in the areas specifiedin Part A above of about 120 workshedshaving floor areas of between about 56m 2 to employabout 4,000persons. 2 and 186m 2. Construction,equippingand staffingon servicedplots in the areas specified in Part A and B above of about 100 training and productioncentersto train and employ about 5,000 persons. 3. Provisionof about Rs. 11,000,000in loans for the purchase of machineryand equipmentby smallentrepreneurs. D. Materialand Child Health (to be carriedout by DSW) The provision of centers and facilities to provide supplementary nutrition,health examinations and immunization for about 17,000childrenup to six years of age and about 7,000 expectant and nursing mothers, with health, nutrition and family planning education and functional literacy trainingfor about 20,000 women from the householdsin the areas specifiedin Parts A and B above,which will include:  -33- ANNEX I Page 3 1. Construction of, and provision of vehicles,furnitureand equipment for about 200 child welfare centers;and 2. Provisionof staff, teachingmaterials,medicines,foodstuffsand operatingexpendituresfor the above until March 31, 1980. E. out by Tamil Nadu Water Water and Supply and Sewerage (to be L.arried Supplyand DrainageBoard & MC) Constructionof and provisionof equipment,materialsand servicesfor, such works as shall be agreed upon by the Association for the rehabilitationand extensionof the existing water supply and sewerage systemsin Madras. F. Roads and Traffic Improvement(to be carried out by DHRW, MC and the PoliceDepartment) 1. The construction of about 12 km of the Inner Ring Road in Madras between G.S.T. and the Madras-TiruvallurRoad to providea 7 meter wide, two-lanesinglecarriageway, with a 2 meter wide cycle track on each side. 2. The improvement of approximately200 km of footpathsand about 50 km of cycle tracksand the provisionof streetlightingat about 24 intersections, all in Madras. 3. The constructionof nine pedestriansubways;a grade separationat two road/rail intersections; the improvementor constructionof seven minor bridges over rivers and canals; the provision of traffic engineering measures and carriage way improvementsat important and the provision road links and intersections; of street signs, guard rails, road markings, road painting equipment, flashing beacons, traffic signals and traffic enforcement equipment,all in Madras. G. Bus Transport(to be carriedout by PTC) system in Madras by: of the bus transportation The improvement 1. The provision buses. of about 285 single-deck 2. of two major and one minor depots for bus repairs The construction and servicing. 3. of eight bus terminals. The construction 4. of about 400 passenger The construction shelters.  - 34 - ANNEX 1 Page 4 H. TechnicalAssistance(to be carriedout by MMDA) 1. Employment by MMDA of up to five advisors for a period of approximatelytwo years each and the provision of short-term consultancy services for improvement of development planning, capital progranmaingand project monitoring and evaluation proceduresin Madras. 2. Consultants' serv'-es, if necessary, to assist in the reorganization of the accounting and improvement system of MC. 3. Trainingof staff of executingAgencies. 4. The provisionto MMDA of aerial photographyand related equipment for the preparationof a long-term slum improvementprogram in Madras.  - 35 - ANNEX 2 Page 1 2^a*.s WJfI 43 ^APR390 ml 4 3 W-jf,.Dt'FA 2414a 31661S7FINEIN FROM SHRISUNIARAAM KRISHNA UNDEk SELrECOFAbkS NEVDELHI TO SHRICh VASUDEV ADIUtSER TO ED C INIIA) IN'BAFRAD WAISINGTON KINDLYPASSON THEFOLLOWING COMMENTS ON PPAROF MUDPOK TO YiEINORI WATANABEOF OED(.) IOTE: WHILEWE AREIN GENERAL ASREEMENT UITH THEOBSERVATIONS IN COtTAINED tHE DRAFT PFARFORTHEFIRSTMADRASURDAN DEVE'OPMENT PROJECT WE HAVEFOLLOWING COM4ENTS TO OFFER(C) AA. PARAS 2.03, 3.06, 4.01 AND4.02 OF PCR(. IT IS NOT OED Comment: The Region has made FAIR TO CONCLLJDE THAT MMEiACANNOTFNCIION UITHOUT CONTINUINGSUFPORT adjustments in the tPPropriate parts of OF TrEASSOCIATION (.) AS THEPPANPARA46 ACKNOWLEDGES MMDA the PCR. HADAL.READY ACQU'RE EXPERTISE INMETPOPOLITANMANAGEMENT AND HAPBECOMEIN IJF. INFLUENtIAL AGENCY IN ITtOWNRIGHT (.) SECONDLY AS FAr' 5I0 3 OF SAPNOTESTHEMMDA'S ROLEWASEXPECTED' TOPfEAPVISCRk IN YATUfE DUI AFTEREXPERIENCE HADBEEN GAINED ON ETOPOLITeN Cf 11k, DBUI,ETINCGTNLiOULD COi;IDIERGIVINGM1IiA PFOWER To DEIEREMINE IN7tR SECTOR.L ALLOCATIONS(.) INVE'TMEHT IT IS THEREFORE NOTCOF:FECt T. SAYTHAT ROLE Or> IN INVESTMENT MMh'l PLANNING STILL STRICTLY REMAINS A:.l'.IS2R` (.) FINALY PMXkYXXXX THE Gr DECENTRALISEDMANAGEMENT REQUIREMENT FUNCTIONS REGU!RE THAT MONTFOIS AREDISFERESED GTN AMNiOST MMDA LOCAL ANID U+ AUTHORITIES AS ACKN:.W'WEDGEr ItJ PARA47 OF PF'AM(.) BDb.PARA.3.01 OFPCR(.) IT IS ACKNOWLEDGED INPCFPARA 4.6t THATHIGH RANKINGCIVIL SERVI.SERVANTSASSIGNEDTO MHDA mUICLYKECAMEStJPPORTIVE OF PROJECTObJECTIVES (.) SINCE FACTS 9H04THATNME'AEMERGEDAS A 5.ICCESSrUL COORDINAINGAGENCY THE SLIPPOSITION IN PAkA CONTAINEII 3.1C OFF'COIS IRRELEVANT C.) ECC. PARA 6.01 Or PCRE.) THE SHIFT IN POLICY FROMCOSTLY SLUMCLEARANCE SCHEMESTO SLUrIMPROVEMENT HAI! ALREADYbEEN INITIATEDDY GTN PRIORTOMUDFONE IN THE FORMOF ENVIRONMENT IMFROVEMENTSCHEMSC+ SCHEMES (.) THIS IS ACKNOWLEDGED BY PPAR ITEM(-3)PAGEVIII AND PARAS2,14AND 38 TO 39 OF PPAR C. THE ENTIRECREDITTHEREFORE DOESNOT BELONGtO THE AC+ASSOCIATION(.)  - 36 - ANNEX 2 Page 2 BDL. PARA 2.10OF PCP(.) COMPLIANCE CONVENm;.TS UITHFINAtM:.AL B' PTC UNDERMUDPTUO HAS SINCEBEENAFFECTEDAND DISBURSEiENT SUSPENSION ACCORDINGGY LIF'ED(.) PARA MAY DE O0'DFIED q.) 42 TO44 OFPPAM(.) MMDA 3 AND EEE. PARAS EXPLORI1G AREALREABIY APPROACHES ALTERNATE TOLANDACQUISITIONPROBLEMS'btCH AS GUIDED OED Coment: LAND LAND DEVELOPMENT. AND READJUSTMENT LAND PPOLINGWHIC This is acknowledged SHOULDBESUITBLY ACKNOULEDGED ACKNOULEC+ (*) by insertions in paras. 3 and 44. OFF. PARAS52 AND 53 OFPPAM(.) MNDA'SVIEU IS TIO TKE LuAISTICS NUMBER A LARGE ANDIMPLEMENTING OF PLANNING OF PROJECTS 1S NOT EFFECTIVE COST ANDFULLBENEFITSCANBE ACHIEVED ONLY lHROUGH OF REASONABLE (.) SMALLLER PROJECTS SIZE SCHENES MORE CANBE HANDLED C.) AGENCIES SECTORAL BY CONCERNED EFFECTIVELY (.) THISVIEWMAY BE INCORPORATED HHHI. FARA21 O' PPAM(.) ACTIONHAS ALREADYBEENINITIATED OED Comment: BY MMDA CONSULTANCY THROUGH FOR STUDIES COMPREHE''"VE SOCIO-ECONDOICThis is acknowledged 'byan insertion in EVALUATION TO COVERSUCHDATAGAP DEFICIENCIES UHICHIS EXPECTED para 21 IN THE FINALREPORT(9) THISACTIONMAY BE ACKNOULEDGEI' (.) UNQUOTE REGARE,S (.) REGARDS(.) NO 1954 ;FSSAGE 29/4/1986KEUAL tAhTEDl KKK :04291424 :04291147 MSSYSTEM WORLDBANK  - 37 - PROJECTCOMPLETION REPORT INDIA FIRST MADRASURBAN DEVELOPhENTPROJECT (CREDIT 687-IN) I. INTRODUCTION A. SectoralContext 1.01 Tamil Nadu is among India'sthreemost dense,highly populated and urbanizedstatesand Madras is India'sfourth largestcity. With a metropolitan population over 5 million,Madras is about five times as large as Madurai, the next largestcity in the State. Madras is the trade and commercial center of southernIndia; expansion of the manufacturingsector (machinery, transportequipment, chemicalproductsand textiles) over the last three decadeshas been a major reasonbehind the rapid 4-5% annual population growth, over 40% of which is due to in-migrationfrom poor, economically stagnantrural areas. Expansionin Madras' organized manufac- turing sectorhas slowed in the last decadeand nearly half the work force is self-employed in petty vending,servicesor transport activities.Small businesses, employingless than 10 persons,account for almost 40% of the jobs in Madras, which had a low per capita incomeof Rs 875 in 1975.Almost half of the populationhad incomesbelow the absoluteurban poverty line. 1.02 Local government within the Madras Metropolitan Area (MMA) consists of the Madras Corporation, four small municipalities, four townships and 20 panchayats. The Government of Tamil Nadu influences urban development activities throughMMDA, the localgovernments, its own departments, statutory bodies such as the TNHB and the TNSCB,and public corporations such as PTC which have major sectoraldevelopment responsibilities. Madras is also characterized by the predominance of privatevoluntaryagenciescarrying out major socialworks programsand playingan important role in providing health services and education. Nearly half of all primaryschoolsand the majority of the high schoolsin the city are private. B. SpecificConditionsLeadingto the Proiect 1.03 The provision of servicedland and housingfor low incomegroups in Madras laggedwell behind the demand for severaldecades. GTN public invest- ment programsin the housing sectorwere primarily aimed at middle-income groups and were highly subsidized. Land development controls, building regulations and other administrative factorsgrossly inhibitedthe legal production of shelteraffordable by low incomefamilies. The slum population of Madras was increasing more rapidly in proportionto the total population, accounting for more than 30% of the total. By the early 1970's,Madraswas sufferingfrom seriousdeficiencies in key servicesectorsaggravated by a long history of inadequate water supplyand sewerage. Water is a  - 38 - scarce and expens.veresourcethroughout Tamil Nadu. Madras'per capitawater supplywas averagingless than 70 litersper day (comparable figuresfor Bombay and Calcuttawere 138 and 145 lcd, respectively) and the poorestareas receivedless than 40 lcd. During drought periods,per capita availability has dropped to 10 liters per day and much of the manufacturing sectorhas been forced to shut down temporarily. In the mid-70's,about 1.2 million personswere living in slum hutmentareas without servicesand legal tenure. 1.04 Attemptsto eeal with this burgeoning problem in the 1970's resulted in a slum clearance/improvementprogramthat producedannually3,000 public rental units for resettlementwhile the slum population was increasing each year by some 14,000households. The programwas highly subsidized as the rents chargedcoveredless than 10% of the costs. Slum clearance was increasing the problemrather than helping to solve it. Other sourcesof newly built housingwere insufficient in quantityand unaffordable by the poor, unless supportedby substantial public subsidies. Public transport was likewiseheavilysubsidized and the bus fleet was increasingly aged, ineff- cient and very overcrowded. C. PrincipalReport Sources 1.05 The principaldocumentsused in preparationof this PCR, other than supervision are reportsand generalcorrespondtence, summarized in Annex A. AND APPRAISAL PREPARATION II. PROJECT IDENTIFICATION. A. Originand Background 2.01 The projectwas first mentionedin the Association files in April 1972. It was discussed with a Bank reconnaisance mission in October1973 and was preparedby the Madras Metropolitan Development Authority with a view to appraisalin 1974. MMDA was assistedby numerousState and localgovernment agencies, most of which subsequently became the executing agencies. During the same period a water supplyand sewerageproject in Madras was also under consideration by the Bank; this led to incorporation in the MadrasUrban Development Projectof urgentlyrequired water and sewerageinvestments while the longer term investment programwas being preparedunder a UNDP financed study. The urban projectwas well preparedand possiblycould have been appraised about a year earlierthan it was, were it not for the complication of relatedwater and sewerageissuesand considerations of priorityto be given to urban investments elsewherein India.  - 39 - B. Preparationand Appraisal 2.02 MMDA coordinatedprojectpreparation, details of which were prepared by the followingagencies: Project Component Agency Responsible A. Sites and Services Tamil Nadu HousingBoard (TNHB) B. Slum Improvement Tamil Nadu Slum ClearanceBoard (TNSCB) C. Small Scale Business - Small industry Small Industries Development Corporation(SIDCO) - Cottage industry Tamil Nadu Handicrafts Development Corporation (SIDCO) D. Maternal& Child Health Departmentof SocialWelfare (DSW) E. Water Supply and Sewerage Water Supply and Madras Metropolitan SewerageBoard (MMWSSB) F. Roads and Traffic Improve- Madras Corporation (MC) and Department ments of Highways& Rural Works (DHRW) Inner Ring Road, pedestrian subways,grade separations, minor bridges G. Bus Transport MetropolitanWing of Pallavan Trar.sport Corporation (PTC) H. TechnicalAssistance Madras Metropolitan Development Authority(MMDA) 2.03 AlthoughMMDA's functions had been focusedprimarily on physical planningand land-usecontrol,in preparingMUDP I it began to initiate studiesfor capital investment programming in the MetropolitanMadras Area (MKA). It has since begun to compileoverallestimatesof public investments in the key sectorsand by differentagencies;these could serve as a basis for rationalizingpublic investment strategies for MMA. The role of MMDA in  - 40 - investment planning, however, is still of an advisory nature, although CTN has under consideration giving MMDA power to determine inter-sectoral urban investment allocations. 1/ In preparing the project, MMDA did not have direct authority over the other GTN agencies, aside from its normal land-use control activities. However, for want of agency capacity, MMDA did (and continues to do) the layout planning and financial analysis for cost recovery in sites and services and slum upgrading. Most of the key GTN agencies involved in MMA development were represented on MMDA's governing board through GTN Departments to which they were responsible. MMDA was made responsible for reviewing and sanctioning all agency scheme proposals under the project. All these factors and the civil service seniority of its chief executives enabled MMDA to act as an effective inter-agency project coor- dinating body. MMDA also acted as Secretariat for the Center/State Review Committee which was set up in October 1975 tc review projects and schemes undertaken for the development of the MMA. 2.04 The appraisal was, in most respects, thorough and dealt appropriately with the number of components, the complexity of the issues and the institu- tional arrangements. The resulting project was generally realistic in size and scope, was well executed, led to a follow-on project, and resulted in important institutional improvements and major shelter program changes. 2.05 The employment and health components were small and not influential over programs and institutions in their sectors. The small scale business component was not appraised in any strict sense, and the maternal health component appears to have been adopted with little program evaluation. However, the two components did provide necessary local employment oppor- tunities and neighborhood social services. There was a strong demand for the 120 small industrial plots and sheds in the site and service areas and they are contributing significantly to cost recovery and balanced community land use. The attempt to create cottage industries in the slums through the efforts of voluntary agencies and the Tamil Nadu Handicrafts Development Corporation failed because of inadequate organization and the factory sh,eds were sold to entrepreneurs to achieve the same purpose. 2.06 As to water and sewerage, the Urban Projects Department decided at the time of appraisal not to address financial aspects of the water sector of Tamil Nadu or the long term water and sewerage problems of Madras as part of this project, because a UNDP/WHO study underway would provide detailed proposals for reform about a year later. Funds for a limited component, along with financial covenants, were included in the project on an interim 1/ Response to DEA telexed (30 April, 1986) comment AAA: "MMDA's role was expected to be advisory in nature but after experience had been gained...GTN would consider giving MMDA power to determine inter-sectoral investment allocations."  - 41 - basis. This resulto.d in a US$10 Million i,Amediateworks program which was implemented during the early years of project implementation. ImplementatiL responsibility as transferred during project implementation from the State Water Authority and MC to the MMWSSB, treated after appraisal. The Associa- tion did not eventually negotiate a project to deal with the critical probLem of water supply in Madras, but in March, 1984 a state-wide water supply project for Tamil Nadu was presented to IDA's Board. C. Project Objectives 2.07 The basic objectives of the project were to: (a) Redirect a significant part of public investments in the key sectors of land development and housing to benefit the low-income population of metropolitan Madras; (b) Introduce the concept of full cost recovery for key investments; (c) Substantially increase the supply of affordable low-income shelter and institute a replicable long-term program; (d) Improve local government financial and administrative capacity to deliver and maintain services; and (e) Strengthen government's capacity to plan, coordinate, implement and evaluate land servicing, shelter, basic services and maintenance programs. 2.08 Related key policy measures included: (a) Provision of secure tenure (freehold titles) to the inhabitants of the improved slums and recovery of improvement costs through hire-purchase agreements; (b) Introduction of a market-related mortgage interest rate of 12%; (c) Establishment of a ceiling on annual expenditures for slum clearance; (d) Establishment of revolving funds to finance future sites and services and slum improvement programs; and (e) Raising of bus fares to eliminate the deficit of the Pallavan Transport Corporation;  - 42 - D. ProjectDescription 2.09 The projectcomprised: (a) The provisionof sites and serviceson about 175 ha, with about 13,500 servicedresidential plots and about 21 ha of servi-edland for small industryand commercial uses; (b) Infrastructure improvement in 85 (laterreducedto 53) slums coveringabout 185 ha with a population of 23,000 households, plus provision of open plots in and around the slums for an additional7,500households; (c) Work sheds,equipment,trainingand financeto generate about 4,000 jobs in small industryand about 5,000 jobs in cottage indus-ryin the sitesand services and slum improvement areas; (d) Training,equipment and buildingsto supplysupplementary nutrition,health examinations,immunization,health servicesand family planning,and nutritionaland fLnctionalliteracy training; of existingwater supplyand seweragesystems; (e) Rehabilitation (f) Road and traffic improvementsemphasizingpedestrian facilities and cycle tracks; (g) Replacementof 285 over-aged buses and construction of bus depots, bus terminalsand passengershelters; and (h) Technicalassistanceand training,mainly to the Madras MetropolitanDevelopmentAuthorityand the Madras Corporation. E. OreditCovenants 2.10 As to major credit covenants,the Slum Clearance Board compliedwith the covenantlimitingexpenditures on slum clearance to Rs 3.75 crores per year and this has been extendedto MUDP II. PallavanTransport was able to complywit' its financialcovenantsfrom 1977/78through1979/80becauseof a CTN sanctioned fare increasein 1976. Due to tardy and inadequate fare increases for four consecutive years since 1979/80,PTC was losingabout US$10 million (Rs 10 Crores) annually,contraryto projectfinancial  - 43 - covenants.1/ During 1983/84CTN converted a portionof PTC's debt to equity. Fares were increasedfrom 1 April, 1985 and during FY 1984/85PTC's financialperformance conformedto agreed targets. CTN also commiA ted itself to biannualreimbursements of PTC losseson studentsubsidies. 2/ The MunicipalCorporation kept debt servicebelow 20% of revenueand has achieved the agreed annual increaseof 8% in self-generated income. UntiL 1983/84,MC was in no positionto submitannual auditedaccounts. MMWSSB has not com- plied with the covenantto set tariffsat a level sufficient to meet costs and is losingabout US$7-8 million per year on its regular(not drought related)operations. 2.11 Other covenantson which there has been progress, but at a much slower pace than originally anticipated concern: (i) the revisionof Madras Corporation's accountsby 1978, actuallyachievedin 1982; (ii) the provision of sufficient VTN funds to improveall slums by 1985 is unlikelyto be real- ized before 1990, and (iii) only PTC has been able to reasonablycomply with audit covenants. III. PROJECTIMPLEMENTATION AND COST A. Start-Up 3.01 There were encouraging early policymeasures: GTN issuedorders on security of tenure and cost recoveryin August 1976, raisedbus fares by 22% in March 1976 and experimented with staggeredworkinghours to improvebus utilization. Start-upwent smoothlyin the PTC component, but in sites and services and the Ring Road schemes therewere delaysdue to land acquisition and contractual problems. Slum improvement startedslowly becauseof ini- tiallyweak provisions for communityinvolvement.Overall,most of the implementing agencies performed well, overcominginitialreservations about new shelterconcepts. Considering the institutionslenvironment and number of agencies, MMDA did an excellentjob of coordinating the project. More could have been done to explainthe programto the publicand prospective beneficiaries and to coordinate and improvethe processing on a large-scale of plot sales, tenure agreements and settlementdocuments. Becauseof infla- tion, and even higher rates of price increasesin construction costs, some contractors abandonedtheir work uncompleted becausetherewere no e-calation 1/ Becauseof the failureto comply with the financialcovenants,prin- cipallyby PTC, in June, 1984 the Association formallysuspended disbur- sement3on the SecondMadras Urban Project. Disbursementswere rein- statedfollowingthe fare increaseof 1 April, 1985. 2/ Responseto DEA telexed (30 April, 1986)commentDDD: "Compliance with financialcovenantsby PTC under MUDP 2 has since been affectedand disbursementsuspensionlifted."  - 44 - clausesin their contracts. During 1978 and 1979, there was an acute shortage of buildingmaterialsand inflation was runningat an annual rate above 15%, very high for India, and in excess of the 12% annual price contin- gency providedat the appraisal. Schedule B. implementation 3.02 Implementation startedin April 1977,but did not gain momentum untiL late that year. Althoughthe projectwas well executed,there were sig- nificantcontracting problems, particularly with respect to largercontracts of more than 12 months duration,and land acquisition problemson the Vil- livakkam and Kodungaiyur tank sites (old water irrigation ponds) and on the Rin Road alignment. Even though the tank siteswere Government owned, there were seriousand unanticipated procedural difficulties, relatingto water rights,in securingtheir transfer. Land acquisition for the sheltercom- ponents,took approximately three times the amountof time estimatedat appraisal. Eventually separateland acquisition units set up in MMDA helped to resolvethe problem for the first urban project. But the problemhas reemergedin MUDP II in the form of legal challengesto the acquisition of privately-owned land. Procurement was mainly delayed in the water and sewerage component, whereasobtainingan adequatesupplyof certain foreign equipment was difficult. There was also a long delay in commencing the issuance of lease agreementsin slum improvement areas, which was only started when it was made a conditionfor negotiating the Second Madras Urban Project. At the time of the PCR, about 50% of MUDP I households have acquiredleases. Home construction by the settlerswas in many cases delayed due to lack of funds. To facilitate funding, MMDA and TNHB helped arrange construction loans throughbanks ana GOI's housingagency,HUDCO. C. Costs 3.03 Total projectcosts were estimatedat US$52.0Million (Rs 468,000,000), of which US$8.0Million constitutedthe foreignexchangecom- ponent. The IDA credit of US$24.0Million financed46% of projectcosts net of taxes and duties. There was a cost overrunof approximately US$10.0 Million(see Annex B), primarily due to increasedconstruction costs on sites and services, water supplyand sewerageand road components. Civen the high rate of inflationin India in 1978-81and the even higher rate of increasein building materialcosts, overrunscouid have been much greaterbut for the relatively good quality of projectadministrationand managements' capacity to respondquickly to implementation problems. 3.04 Cost overrunsof almost 35% were experienced on two out of three sices and servicesschemes. This necessitated successiverecalculations of cost recovery and prices in order to meet the affordabilityrequirements and provide75% of the plots to the 10th to 45th incomepercentile of the popula- tion. Althoughconstruction cost increases exceededhouseholdincome increases, the market value of the more desirableresidential plots, and the  - 45 - prices commandedby the industrial and commercial plots,were much higher than expected.Togetherwith adjustments in layoutand land use, the market sales should be sufficientin aggregate to recoveryaveragecosts without pricingthe cheaper lots beyond the financial capabilities of the intended beneficiaries.However,the actual achievement of cost recoverytargetshas been threatened by initiallypoor collections on the industrial plots and improvedslums. This performance has been caused by delayedprovisionof infrastructure servicesto the entrepreneurs, and, in the slums,by inade- quate collection proceduresand a reluctance to apply sanctions. D. Disbursements 3.05 Disbursements were US$2.0 Millionahead of the appraisalestimatesat the end of the first disbursement year of the projectand remainedahead of schedulethroughout the secondyear. This was largelydue to early and rapid procurement of buses by PTC. The final disbursement of US$24.0Million, estimatedto be reachedby June 30, 1981,was howeverachievedone yea- later. The main causes of disbursement delays were contractual and land acquisition problems. Althoughthe Kodungaiyur site is to be completedin mid-1984,all but a few minor parts of the rest of the physicalelementsof the projecthave been satisfactorily finished. The creditwas closed on December31, 1982,extendedfrom the or ginal closingdate of September30, 1981. For the project, GTN had assignedto MMDA the authorityto review and approveall disbursement applications.This not only expeditedexecutionof the project but also strengthened MMDA's capacityto coordinate effectively. of Contractors, E. Performance and Borrower Consultants 3.06 MMDA plannedthe qite layouts, designedthe sheltertypes and did the cost recoveryand other financial analysis,particularly for sites and serv- ices and slum upgrading. It also organized and managed staff for the com- munitydevelopmentfunction, which was a new and criticalprojectactivity. The executingagenciesprepared detaileddesignsand tenderdocuments, awardedthe contracts,supervised construction and handledestate transac- tions. There was littleneed for consultants in preparing or implementing basic projectactivities. However,the selective use of consultants for specialized tasks (such as the reorganization of the Madras Corporation's accounts)provedefficacious. MMDA did an excellentjob of coordinating and monitoring the physicalimplementation of the project. But MMDA's reports to IDA have tended to focus on cost and engineering targetsand could have evaluated more thoroughly the achievement of projectobjectives. The Authority's abilityto monitoragency progress was primarily due to the effective working relationships alreadyestablished with the GTN agencies durijigprojectpreparation and the improvements by the agenciesin their own systemsfor planningand monitoring, projectimplementation and coordinating with each other for common purposes during implementation. A serious shortcoming by MMDA has been its failuieto developthe necessaryprofes- sionalcapacity for in-depth sectoralfinancial planningand programming.  - 46 - it does not AlthoughMMDA has the power of review of projectexpenditures, oi budgetarycontrols.1/ have broader programmatic 3.07 All the civil works were executedby medium and small-sized local contractors (most contracts were in the range of US$20,000-300,000) and performance was varied. Shortagesof raw materials,particularly cement, caused some delays in the project. Some firms abandoned their contracts midway, necessitating fresh tenders. Larger contractors were reluctant to take up scattered smalL works because of unattractive ratesand unfavorable conditions of contract. Eventually, smallerworks in sitesand serviceswere consolidated into somewhatlargercontractpackages with improved terms and conditions. These are however still too small to attractlargercontractors. Escalationclausesstill are not includedin civil works contracts, although recommendations to do so, by the State's Board of Chief Engineers, are under active consideration by GTN. Further improvements in termsof contractand contractsize will be needed as the scale of sites and servicesprograms increase. 3.08 The OperationResearchGroup (ORG)of Barodawas retainedas consult- ants for a number of economic studies. An Employment Base Study of the Madras Metropolitan Area (MMA) surveyedt1lemain featuresand problemsof the economy,but was unable to come to grips with the underlying causal factors for changesin the economy,since extensivebasic research was required. Also, this study did not examineGTN policytoward employment and industrial growth or consideralternatives that could help the formulationof poiicy on industrialization. ORG satisfactorilycarriedout a Resource Mobilization and Investment PlanningStudy which is servingas a basis for rationalizing the propertyevaluationsystemfor propertytax. These con- sultantsalso helpedestablishMMDA's monitoring and evaluatingsystem,which is operatingeffectively in the executingagenciesand they did a number of other individual studieswhich enrichedMMDA's understanding of the MMA economy. 3.09 Alan Turnerand Associates(UK) assisted MMDA in preparing a Struc- ture Plan for the Madras Metropolitan Area. Tnis was funded by the Overseas Development Authority(UK) under the Colombo Plan. After a detailedsurvey of the existingsituation and trends,guidelines for growth and alternative development strategieswere proposedfor the next ten years. Land require- ments and distributionwere assessed;a preferredstrategy was proposed;and prioritiesfor action in the main sectorswere specified. While the result- ing Plan was thoroughand MMDA uses it to screenthe suitability of individual projectsfrom the point of view of land use, further studies are needed to develop sectoralinvestment priorities, financial policiesand specificrelatedaction programs. Althoughbegun late in the project 1/ See para 2.03 for DEA's commenton this point and ILA's response.  - 47 - implementationperiod,consultant assistancefor developingand implementing a new accountingsystem for the Madras Corporationis graduallyprovidinga betterbasis for the management of municipalfinanceand services. 3.10 The dorrowergenerallyperformed well in preparing,executingand monitoringphysicalaspects of the project. Throughoutthe project implemen- tationperiod, the implementing agenciesand GTN demonstrated a substantial capacityfor rapid adjustnentto problems,particularly those which could be solved throughadministraEive decisionsor technical changes. Areas in which improvements have been slow to take hold are the capabilityfor financial analysisand managementin all agenciesexcept PTC and middle term pl_nning and programming except for PTC and TNHB. From time to time, MMDA's develop- ment as an investment coordinating agencymay have been affected by too frequentchanges in its top management.1/ of Association F. Performance 3.11 Early missionswere helpful in guiding projectpreparation. A pre-appraisal missionwas convertedinto an appraisalmission in the field in May 1976 and the projectwas appraisedin a realistic manner despite the many components and institutional complexities. Althoughthe immediate action programfor water supplyand seweragewas satisfactorily executed,the Association and CTN did not come to grips with Madras'criticaldeficiencies in water supply, thougha state-wide projectwas presentedto IDA's Board in March, 1984. The Long delay in identifying and implementingimprovements in water supply in the MMA carriesa heavy socialand economicprice. 3.12 Supervision of the projectwas effective,including concertedefforts to follow throughon the financial covenants, maintainthe implementation schedule and achieve the overallprogram,as well as projectobjectives. There were about three supervision missions per year (22 man weeks) in the initialyear of the year of the projectand about two per year (6-8 man weeks) in the later years of the project. Missioncoverageof the most important projectcomponents, (shelter,slum improvement, transport, and institutional development was thorough). Strong and collegial working relationships with the implementing agenciesand Departments of GTN were 1/ Responseto DEA telexed (30 April, 1986) commentBBB: "Since facts show that MMDA emergedas successful coordinating agency the supposition contained in para 3.10 of PCR is irrelevant." The Associationmaintains the view that from time to time too frequentchanges in managementhave adverselyeffectedMMDA's development as an investment programming agency.  - 48 - maiitained. Supervision was concentrated on the housingand transportcom- ponents, as the most significantparts of the project;becauseof organiza- tion and budgetaryconstraints supervision of the healthand employment components was thin. 3.13 A follow-up projectof similarcomposition, for a US$42.0Million Credit (which, from the first project,almostdoubledthe number of improved slum units and the number of buses, and added solidwaste and drainagecom- ponents)was prepared and appraisedduring the third year of MUDP I. The comparative successof both projectsand the resulting institutionalimprove- ments and program reformsin the shelter sectorindicatethat the Association was effectivein addressing these problemsat a criticaLstage and in sup- portingthem over an initialeight year period. 3.14 The size of the componentsfor worksheds/equipment, job training,and health/nutritiondid not permitany leverageeffecton the existingGTN programs,nor was there any comprehensiveprogramevaluationin the course of preparing,appraising or supervisingthe project. PERFORMANCE IV. INSTITUTIONAL AND DEVELOPMENT A. Madras Metropolitan Development Authority(MMDA) 4.01 The MMDA staff preparedthe projectwith assistance from the various executing agencies. Through this preparation work, MMDA developedthe lead planning role which was continuedthroughmonthly inter-agency meetings chairedby the Authority. These meetinsworkedwell in coordinating the physicalimplementation of both the firstand the on-goingsecond urban projects. MMDA has developed an effective(but slim) capability for the dynamicphysicaland financial planningconnected with pricing,cost recovery and affordability criteriain sites and services schemes. Althoughthe leadership of MMDA changedseveraltimes during the projectimplementation, hign rankingcivil servants were assignedto MMDA and they quickly became supportive of projectobjectives. MMDA's reportingsystemhas concentrated on engineering and expenditure progressand has providedreasonable com- parisonsof estimated and actual costs and the time requiredfor completing components. With experience, the staff is tryingto developother indicators for assistingprojectmanagement by objectives. The Authority's staff has receivedtrainingin planningand technical assistance in economicstudies. However,the expectedrecruitment of financial and economicpersonnelhas rot happenedand MMDA'sability to do financial and economicplanninghas not improved substantially. 4.02 MMDA's objectiveswere to have been the preparation of economic, socialand spatialprograms; and the Government of Tamil Nadu was to consider enhancingMMDA powersto carry out metropolitan-wide financial planningand to determineintersectoralinvestment allocations for metropolitan Madras.  - 49 - However,such powers have not been granted since they involvesensitive issues of state, local governmentand agencies'relativepowers,particularly the power to determinebudgetaryallocations and programo jectives. Also, MMDA has not yet developedthe expectedcapability of carryingout metropolitan-wide budgetaryplanningand programming. Considerable further efforts by MMDA are needed to improvethe qualityof its monitoring and reporting activities aiidto developa capacityfor evaluating shelterand other sectoraldemandsand ways of meeting them from both the public and privatesector. The strengthening of MMDA's investment coordinating role will require increased GTN supportfor planningurban development 1/. B. Madras Corporation 4.03 Like many municipalentities,the Madras Corporation had been over-staffed, had poor financial management, had an inadequate fiscalbase and had grave deficiencies in the provision and maintenance of infrastruc- ture. Three significant impr)vements were broughtabout under the present project. First, throughextensivetrainingand recruitment, the qualityof the financialstaff has improved greatly. This has allowedan accrual accountingsystem to be introduced and the true costs of municipalservices to be more accuratelyportrayed. Secondly, based on studieswhich were an integral part of the project,the Corporati^nis consider.ng ways and means of reformingpropertyvaluationfor propertytax, its basic sourceof income. Thirdly, the Corporation has improvedthe management of its maintenance program,particularly for improvedslum areas which were mostly neglectedin the past. 4.04 Experiencedemonstrates that more time is requiredfor the fundamen- tal changesenvisioned by the projectfor MC. While the Corporation has adhered to the project financial covenants,it has yet to develop,and get CTN to approvea program for substantially increasing its very inadequate self-generated revenue base. MC's per capita revenues of about Rs 50 are still much too small for the effectivedeliveryand maintenance of services in Madras, the capitaland economiccenter of Tamil Nadu. It would be help- ful to MC's planningactivitiesif its GTN-appointed chief executives were changedless rapidlythan has been the case in the past few years. C. PallavanTransportCorporation 4.05 The projecthelped to further improvethe management of PTC, which was alreadya relatively well-managedand efficientoperating company. Through trainingof the top financialstaff and work with the IDA missions, 1/ Responseto DEA telexed (30 April, 1986) commentAAA "It is not fair to concludethat MMDA cannot functionwithoutcontinuing supportof the Association."  - 50 - PTC started to do five year financial projections and to institute stricter financial controls and indiators for management information. The project impact however on transport policy has been mixed. Although a substantial fare increase in 1976 enabld PTC to meet its financial covenant until 1979, since then fare increases have been delayed and inadequate. PTC's operating ratio has deteriorated to about 112 in 1983/84. The covenanted operating ratio of 95 has not been achieved despite operational improvements such as reducing the ratio of staff per vehicle to 7.5 and financial adjustments such as reducing the non-IDA bus procurement program and converting GTN loans to equity. PTC's management has been very receptive to advice from the Associa- tion, implemented the project quickly and has tightened up its operations in the process. D. Tamil Nadu Housing Board 4.06 TNHB's management has reoriented its approach tn housing as a result of the project. It has moved from providing 3,000 high-cost flats per year in the MMA to providing more than 6,000 serviced lots per year with about 95% of cost being directly recovered. The Government of Tamil Nadu has directed TNHB to allocate 80% of its investment to shelter for lower income households - a sharp reversal of earlier programs. Through project experience and training, TNHB's top engineers have enthusiastically developed the Board's capability to formulate and execute a long term housing program on a scale commensurate with the shelter problem. Although TNHB's enginee':ing staff has become more adept at implementing low cost solutions, a capacity for finan- cial analysis and management remains to be formed. Planning, construction implementation,coordination wth other agencies and monitoring of collection performance steadily improved during both this and the follow-on project. Administratively TNHB still does not yet have the organization required to rapidly process a large number of applications and lease-cum-sale agreements so that settlement does not occur as rapidly as plots are available for occupancy. E. Tamil Nadu Slum Clearance Board 4.07 TNSCB's program has switched from one basically oriented toward slum clearance to slum improvement. The management and staff have effectively changed their orientation during the project period. The initial delay in starting the slum improvement program seems to have been due to failure to mobilize the community development staff to explain the program to beneficiaries. Once started, TNSCB executed the slum improvement works efficiently and developed the capacity to upgrade about 15,000 slum units per year. However, the issuance of tenure agreements and the commencement of maintenance and garbage collection by MC lagged several years behind improve- ments. This not only delayed, but severely damaged cost recovery from beneficiaries. Cost recovery in the slums was an innovation of the project and the poor initial cost recovery was improved somewhat to the 35% collec- tion level by requiring households requesting home improvement loans  - 51 - (providedunder MUDP II) to fully pay the chargesdue on their plots. The revenueside of TNSCB will have to be strengthenedto improvethe collection rate further. V. ECONOMICRE-EVALUATION 5.01 The economicbenefitsof the sites and servicescomponentare based on the imputedrental valuesof the servicedresidential plots and the market land. Slum improvement value of the industrial/commercial benefitsare also valuedat the estimatedrentalvalues. Costs includeall infrastructure and amenities, net of taxes. 5.02 At appraisal, the economicrate of returnof sites and serviceswas estimated at 12% (when combinedwith the connercial/industrialdevelopmentit was 15%) and for the slum improvementsit was estimatedat 18%. The weighted averageof estimatedrates of return for the components which could be quan- tified (about 55% of total projectcosts) was 21%. 5.03 Chargeablecosts are supposedto be fully recoveredfrom beneficiaries for the sites and servicescomponent(throughlot sales), for the small-scale businesscomponent(throughsales of workshedsand loans advancedto beneficiaries), for buses and relatedfacilities(throughbus fares),and for water supplyand sewerage(through charges). An attempt is also being made to fully recover the chargeablecosts of slum improvements throughmonthly paymentsfrom the beneficiaries.All remainingcomponents and costs (largelyfor off-sitefacilities) are financedfrom general revenues of CTN and Madras Corporation, or from user charges by public utilities. 5.04 The only componentfor which sales and collections are sufficiently advancedto allow rates of return to be re-calculated is the Arumbakkamsites and servicesscheme. This shows a financialrate of return of about 4.5% and a positiveeconomicreturn of "bout 30%. The main reason for the dis- crepancy, betweenthe economicand financialreturns,is the relatively high imputedrental value of the units versus the actual price charged especially for the economically weaker sector (EWS) units. The appraisalestimateof at least a 12% financialrate of return is not being achievedon this site becauseinterest during construction and marketingdelays were not adequately reflected in plot sales prices. MMDA has correctedthis oversightin sub- sequentpricingexercisesfor other sites. 5.05 The economicrate of return for the slum improvementprogram is estimatedto be about 23% even though the sale of some vacant serviced land in the slums (envisaged at appraisalto createa surplus)did not material- ize. As elsewherein Madras, the imputedrentalvalues in the slums are at least 50Z higher than those assumedat appraisal. The bus componentwas implemented very well and there are substantialcost savingsfrom the newer buses,efficientoperations and de-congestion.But the failureto raise bus  - 52 - fares and water chargeswill adverselyaffect the financial rate of return in the relatedcomponents. In economicterms, it is expectedthat the overall projectwill generatea rate of return moderately higher than that estimated at appraisal. VI. CONCLUSIONS 6.01 The most significant aspect of the projecthas been its impact on housingpoliciesand programs. Essentially, the projectsupportedefforts which GTN had alreadybegun to initiateto replace the policyof high cost, highly subsidizedslum clearancewith a more cost effective, equitableand financially sustainable slum improvement programwhich minimizesrelocation and demolition. [t also supportedthe reorientation of GTN's housing program from costly flats co sites and services,land servicingand core housing, which were much lower in unit cost and self-financingwithout interestrate subsidies. 1/ 6.02 This projecttogetherwith MUDP II, have had a substantial long-term programmatic and institutional impact. Taking the two projetstogether, about 24,500 servicedplots and 75,000 shelterunits in the upgradedslum areas, mostly cateringto the EWS, will have been provided. This amounts to a net annual increasein the public sector sheltersupply from approximately 6,000 to 13,000units per year. Simultaneously, the level of slum clearance, cappedat Rs 3.75 Crore expenditure per year under the project,now finances only about 1,500 units a year. The unit cost of a servicedlot and improved slum plot average about Rs 12,000and 2,500, respectively, versus 25,000 for the smallestwalk-up flat. The Slum ClearanceBoard has developedthe capacityto improve15,000units per year and has formulated a Five Year Plan, althoughmore effort is requiredto make this operational. The Housing Board now has the capacityto produce 10,000shelter units per year and has sanctioned a Five Year Plan for the MMA which would put 45% and 35% of is resourcesin shelterfor the EWS and LIC respectively. 6.03 The combinedsupplyof Legal,environmentally acceptableshelter from sites and servicesand slum improvementschemes irderboth urban projectswas intendedto start reducingthe absolutenumber of 250,000 familiesliving in illegal, unservicedslums by 1981/82. Becauseof construction, land acquisi- tion and settlementdelays,this target slippedto 1985, when the annual supplyof 7,000 IDA sitesand servicesplots, improvements for 15,000 slum families and 11,000privatesector houses/plotsshould exceed the annual MMA 1/ Response to DEA telexed (30 April, 1986) commentCCC: "The shift in policy from costly slum clearanceschemesto slum improvement had already been initiated by GTN prior to MUDP 1 in the form of Environmentimprove- ment schemes. The entire creditdoes not belong to the Association."  - 53 - householdgrowth of about 40,000. If achieved,this must rank as one of the major interventions Association's in the housing sectoracross regions. 6.04 Particularlimitations revealedduring the implementationof the first projectand correctedin the second projectwere the provision of financingfor housing improvements and the developmentof municipalmain- tenancecapabilities. Generally, the agencieslearnedrapidlyfrom experience and togetherwith GTN demonstrated a capacityfor making opera- tional adjustmentsduring implementation. 6.05 Even though construction costs, due to inflation and cost overruns, exceededincreasesin householdincomes, the serviced plots were suppliedto the EWS at affordablepriceswithoutexternalsubsidies becauseof differen- tial pricing, i.e. prices for lots which had locational or other advantages could be adjustedupwards to market pricesbecauseof strongdemand so that averr3e prices still coveredaveragecosts. Thus the sheltertarget of accommodating primarilythe 10th to 45th income percentile, on about 70% of the plots, is being achievedwithout sacrificing financial or socio-economic objectives, despite the financialshortfall on the initialsite (para 5.04). 6.06 Institutional improvements were very significantparticularly in: the accountingsystemat the Madras Corporation; the coordinatingand monitoringcapabilities at MMDA; the continuedimprovement in the operation and managementat PallavanTransport; and the programmaticreformsat the Housing Board and Slum ClearanceBoard. 6.07 MC has starteddeveloping a capabilityto maintainimprovedslums. Unfortunately, GTN has stillnot foun. a way to improvethe extensiveslums on private lands,which remain in the same poor environmentalconditionas the slums on public Land were when the Association'ssuccessfulintervention began. 6.08 MMDA regardsthe acquisitionof land as the most seriousproblem encountered. It recommends that land with high developmentcosts, such as water tanks, shouldbe avoidedif possible. Since about 18 months are requiredto obtain privateland under emergencyprocedures and about 30 advanceplanningof land acquisition months under normal procedures, for public sector shelterprogramsis a prime necessity. 6.09 SIDCO was initially mark ting subsidizedplots a short distanceaway from the projectplots where full cost recoveryis being sought. This policy has been changedand also the responsibility for selecting entrepreneurs has been shiftedto TIIC. Cost recoveryon industrial and commercial plots, initially delayedbecauseof TNHB's delays in completing infrastructure servicesto plots, is now beginning. While some new small industries started and jobs were created in the site and servicefactorysheds,the attempted employmentcreationand trainingthroughcottageindustries in improvedslums did not affect GTN's employment creationprograms. An extensive analysis of  - 54 - the factorsand policies affectingeconomicactivityin the MMA would be needed in order to developa workableemployment creationprogram for the HMA. 6.10 The most importantspecificobjectives not achievedrelate to the financial covenantsfor the bus companyand the water and sewerageauthority. Large subsidies are requiredfor both these operations and run at an annual level of about US$10 million for PTC and US$8 million for MMWSSB. 6.11 One conceptualshortcoming of both the first and the second urban projectswas not to includethe private sectorin the analysisand approach to improvingthe supplyof servicedland and housing. However,at the time of project preparationand appraisal,the problems needingcorrectionin the public sectorwere so fundamental that the more limitedfocus by IDA and CTN is understandableand may well have been tne only prudentapproachat the time.  - 55- ANNEX A PRINCIPALREPORTsOURCES (Other than Supervision Reports and General Correspondence) Report No, Name Date 1320 a-IN Staff Appraisal Report - The Madras Urban Feb. 15, 1977 Development Project Credit Agreement Project Agreement Madras Urban Development Project I August 1983 Project Completion Report Madras hetropolitan Development Authority (MMDA) Structure Plan for Madras June 1980 Metropolitan Area, Prepared by Alan Turner Associates, Consultants to MNDA Vol. 1 Vol. 2 Guidelines for Growth Vol. 3 - Quarterly Reports Project Monitoring Division MKDA - Memo on Proposed US$24 Million Credit for Feb. 18, 1977 Madras Urban Development Project (Mr. Blobel to Mr. Knapp) - Issues Paper for Second Project June 11, 1976 2893 a-IN Staff Appraisal Report Nov. 26, 1980 Second Madras Development Project 4707 - IN Staff Appraisal Report Dec. 21, 1983 Tamil Nadu Water Supply and Sanitation Project - Industrial Economy of Tamil Nadu and Madras Dec. 1979 Metropolitan Area-Prepared by ORG, Consultants to MMDA - Sites and Services Program at Arumbakkam Nov. 1983 - Financial Evaluation by MIDA IV N\ 2Žt e\'  - 57 - ANNEX B Page 1 of 2 INDIA MADRASURBAN DEVELOPMENT PROJECTI Appraisal Estimates and FinalCosts Appraisal estimates Cost to base cost in 1975 prices Completion (Contingenciesshownin CategoryI) A. Sites and Services: 1. Land & Infrastructure 4.9 2. HousingCore Units 2.1 3. Self-help buildngmaterials 0.6 4. Community facilities 0.2 SubtotalA 7.8 16.9 B. Slum Improvement: 1. Infrastructure 4.7 2. Community facilities 0.2 SubtotalB 4.9 6.2 C. Small-scale Business 1. Work sheds and machinery loansfor small industry 1.6 2. Training,equipment and sheds for cottageindustry 0.8 SubtotalC 2.4 3.0 D. Maternaland ChildHealth 0.7 0.9 E. Water Supplyand Severage 9.0 13.1  - 58 - ANNEX B Page 2 of 2 F. Road and TrafficImprovements 1. 12 km of InnerRing Road 2.0 2. 9 pedestriansubwaysand 250 km of footpathsand cycle tracks 2.3 3. 2 grade separations 1.1 4. City streetimprovement, minor bridges 1.5 5. Trafficengineering, management and enforcementequipment 0-8 SubtotalF 7.7 15.2 G. Bus Transport 1. 285 buses 4.2 2. 3 depots,8 terminals and 400 passengershelters 1.7 SubtotalG 5.9 6.3 H. Assistance Technical 1. Advisersto MMDA (Training) 0.4 2. Consultants' services and aerial photography 0.4 0.8 0.7 SubtotalA-H 39.2 I. Contingencies 1. Physical 3.1 2. Price 9 7 SubtotalI 12.8 Total 52.0 62.3  - 59 - ANNEXC Scheduleof Disbursements Us $E000 IDA FiscalQuarter CumulativeDisbursements Actual % of appraisal AppraisalEstimate FY 1978 Dec. 31, 1977 50 March 31, 1978 200 - - June 30, 1978 500 2,200 440 FY 1979 Sept. 30, 1978 1,300 2,400 185 Dec. 31, 1978 2,500 4,100 164 March 31, 1979 4,500 4,800 107 June 30, 1979 7,500 6,700 89 FY 1980 Sept. 30, 1979 9,500 6,900 73 Dec. 31, 1979 12,000 7,100 59 March 31, 1980 15,000 8,900 59 June 30, 1980 18,000 11,200 62 FY 1981 Sept. 30, 1980 20,200 12,900 64 Dec. 31, 1980 22,400 13,800 62 March 31, 1981 23,200 15,300 66 June 30, 1981 24,000 19,400 81 FY 1982 Sept. 30, 1981 - 19,900 83 Dec. 31, 1981 - 21,100 88 March 31, 1982 - 24,000 100 June 30, 1982 - - -  .- 'BURMA AA I j U R SI LANKA *1~~~~~~~~~~~~~~~~~~ ~~~~MANALI ) ' IIUVTTVU ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~4,IA[ tA vA RAM 7 A V t'~~~~~~~~~~A V,  I~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~OI i J ( 4 i.S--t . A d UO4P>A OA4G / ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~mC ITV{ ; _L I ~t\$-voE K:;ArMZ F3A- °MADRAS MTR ITAN AREA THE -- . ' ~~~~~~~~~~~~~~~~~~~LOCATiON g >F: OF MAO PROJECTCOPNENTS - ,, ... \ . \) S:f /\ ), ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~Progeci ccn0uV , Arrport / > ~~~~~~~~~~~~ iZ ~~~~ tX > ~~~~~~ Z ~~~~ -t~nn.r R.ng Road ) _ \ /A} A:ANDt.Iv p .<) Bus termi.noJ ^~~~~~~~~~~~~~~~~~~~~~~~~~ ~~/ \ ; Mj-/r- & / / / * *o~~~~~~~~~~~~~~~ / ~~ ~ ~ ~ ~ ' - Wtrworks - Th.rid _a supplycoedwit g / r 0: SJ }~~~~~~~~~~~~~~~~~~~~~~i E..istiq band use: / -4n~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~t.R ~ ~ ~~~~/ J ESTATEjZIItS mIndtiustria eQ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ( IlxMBASAM /7r ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~Agr.cultural-P I ma &-y*3s .rc DoIo opn seennW* Pr- t_ mL~~~~~~~~~~~~~~~~~~~~~~~~~~~~~MDA --- i- MERPOIA ARE ,, TH )/< NH Nat,onal 5~~~~~~~~~~~~~~~~~~ b.ghoays t ~~~~~~~~~~~~~~~~~ ~~~~~ Other main - \~~~~~___ roads I~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ -. _. Po.Iways ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ,_,Rers ) / t ~~~~~~~~~~~~~~~~~~~~~~~~~~ and tobnjip boundn. ,/ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ AAun.capal.t,.s AVAD/ Townsh9ps _~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ A. 4. 5 6S 7- 8 7 ., ,, >~~~~~~~~~~~~~~~~~~~~~~~~~1 , \ ....... _I.ETR 2 _ 2 .r N MIoLES3 - _ \- os# /-57 -~~~ i* -/ - ___/____I_____________________