Tanzania - Kigoma Rural Development Project

Document of The World Bank FOR OFFICIAL USE ONLY Report No. 4858 PROJECT PERFORMANCEAUDIT REPORT TANZANIA KIGO14A RURAL DEVELOP?4ENTPROJECT (CREDIT 508-TA) December 28, 1983 Operations Evaluation Department This docmment has a restricted distribution andt may be used by tecipients only in the performance of their official duties. Its contents may not otherwise be disclosed without World Bank authorization.  ABBREVIATIONS ATTC - Agricultural Trials and Training Center CAT - Coffee Authority of Tanzania DCA - Development Credit Agreement DDD - District Development Director IDA - International Development Association KCU - Kigoma Cooperative Union KRA - Kigoma Regional Administration KRDP - Klgoma Rural Development Project NMC - National Milling Corporation MATI - Ministry of Agriculture Training Institute OED - Operations Evaluation Department PCR - Project Completion Report PME - Project Management Exectitive PMO - PrilmelMinister's Office PPAM - Project Performance Audit Hemorandum PPAR - Project Performance Audit Report PPT - Project Preparation Team RADO - Regional Agricultural Development Officer RDD - Regional Development Director RHC - Rural Health Center RIDEP - Regional Integrated Development Project RMEA - World Bank Regional Mission in East Africa TANU - Tanganyika African National Union - later CCM (Chama Cha Mapinduzi) - The Party TCA - Tanzania Cotton Authority TRDB - Tanzania Rural Development Bank UNCDF - United Nations Capital Development Fund  FOR OFFICIALUSE ONLY PROJECT PERFORMANCEAUDIT REPORT TAJZANIA KIGOMARURAL DEVELOPMENTPROJECT (CREDIT 508-TA) TABLE O' CONTENTS Page No. Preface ..... * * * l Basi e Data Sheet iii Hig -'l.ghts .................... ............. iv PROJECT PERFORMANCE AUDIT MEb%RANDUM 1. NAIR ISSUES. S ..... *.... .. . . * .. .. 1......* 5 A. Project Organization * * . . a . 5 B. AgricuLlttral Inmpact ............... *.... * * *..... * 7 C. Long-Terma Support for Ruxral Development .............. 10 PROJECT COMPLETION RE1PORT I. Background 13 II. Fortutlatlon and Design 15 TII. Implemeaitation 21 IVt. lIrgpac t a 29 V. Economic Analysis * ................. ....... * 34 VI. Bank Performance ....................., 35 VII. ConcLusions: Lessons for Rural Developaent ............. 37 Tables Charts Map rhis docuntent has a restticteddistributionand rnaybe used by recipients only in tthe performance of their official duUes.Its contents inaynot otherwise be disclosed without Vorld Bank authorization,  PROJECT PERFORMANCE AUDIT REPORT TANZANIA KIGOMA RURAL DEVELOPMENT PROJECT (CREDIT 508-TA) PREFACE This is a project performance audit report oa Kigoma Rural Develop- ment Project, for which Credit 508-TA in the amount of US$10.0 million was approved in August 1974. The Credit was fully disbursed in September 1982, before the closing date in December 1983. The closlag date was originally scheduled for December 1980. The audit report consists of a Project Performance Audit Memorandum (PPA1, prepared by the Operations Evaluation Department (CED) and a Project Completion Report (PCR) prepared by the Eastern Africa Regional Office. The PPAM is based on a review of the Appraisal Report (453a-TA) dated July 15, 1974, the President's Report (P-1453-TA) of July 23, 1974 and the Credit Agreement dated August 21, 1974. Correspondence with the Borrower and inter- nal Bank memoranda on project issues, as contained in relevant Bank files, have been consulted. The audit has also been able to study a report on an evaluation of Kigoma Rural Development Project undertaken by the Economic Research Bureau of the University of Dar es Salaam.l/ The audit mission travelled to Tanzania in December 1982. Discussions were held with officials from the Prime Minister's Office, the Cotton Authority and the Tanzania Rural Development Bank in Dar es Salaam. The mission travelled to Kigoma Region and visited a number of project facilities in Kigoma and Kasulu Districts. Discussions were held with the RegionaL Development Director, the Regional Planning Officer, the Project Coordinator, the Financial Controller and other regional staff. The audit generally supports the conclusions in the PCR. The pro- ject was directed primarily at Increasing farm incomes and improving educa- tional, health and drinking water facilities in newly established villages. At the time of appraisal, improvements of this type were desperately needed, for Kigoma was one of the poorest regions in Tanzania. The project has not, in general, succeeded in achieving these development goals, primarily because project implementation capacity has been weak and certain aspects of the improved agricultural technology provided under the project were unattractive to farmers. This situation has been seriously aggravated by the country-wide economic problems which have affected Tanzania. Even in areas where the project has succeeded, Government has had difficulty maintaining these activities now that external financial support is no longer available following completion of the project. Tbese issues are discussed in the PPAM. 1/ "An Evaluation of the Kigoma Ruxal Development Project", Michael Loft, Ellen Hanak, and Ahmed Ndyeshobola, Economic Research Bureau, Paper Number 821 (1982).  - ii - A copy of the draft report was sent to the Borrower in October 1983 for comments; however, none has been received. The audit mission wishes to thank the Government officials, both in Dar es Salaam and in Kigoma Region, and the staff of the Cotton Authority and the Tanzania Rural Development Bank for the valuable assistance proziideddur- ing Its visit to Tanzania.  - ilt - PFtRWgCTPERFORNANCYAUDI? RtPORT TANZANIA KICOHA 711gAL DEVELOPKElfT PRtOJCT (C_REntT 508-TA) HASIC DATA sHEET tLY FU0JUCT _DATA Appraloal Act ucl or Actual as 2 of Item Est!stte Stistted Actual Appraisal Ettitete Total ProiJE Coste (US$ mtillion) 13.3 158 1S1 Credit Amount (UDS m illoun) 10.0 1OO 100 UNCIF Grant 1.5 1.5 I00 Date Hoard Approval 08/06/74 Date Vffectivonees 11/20/74 Date Physical Components Completed 12t79 12/82/a 155/b Proportion then completed (X) Apprximately 822 of re-ised targets or 501 of appratosl taroete. Cloaing Date 12/80 12/83/c tranoirc Rate tf Return (I) 22 negoative Insf tat iconal Performadsca Poor, With some impraocvent later Agrono-lc P-rfornance Very poor Number of Diroct donefiliarica (1982) Approximately 250,000 Approximatcly 340,000 in 130 to 135 villages by vl.Isse by 1982./d 1979. VE DISBURSEMENTS CUMVIUATA FY75 PY76 Yf77 F78 FY79 PY80 F101 FY82 Appralsal estia4te (US$ million) 0 3 2.6 4.6 6.6 9.0 16.0 10.0 10.0 Actual (US1$ rllton) 0.6 1.5 3.1 3.9 5.6 9.0 10.0 Actual as t of etitate - 23 32 47 43 56 90 100 Date of final disburoeeent; September 1982 HISSIION DATA T Date lo. of Hfaod4ys Specializattons Perforgonce ype. of mission (mo./Yr.) Persons tn Field gepresented RatinDIf Trrttd/s Problema/h IdeotIficcaton/L 8-9/71 6 7 a,f - - - Preparatlonjl 5/72-6173 12 113 a,f.d Appraiml 10-1173 63/74 7 258 a.d,f,h,1 - - - Subtotal 378 Suryrvision I 10/74 3 30 f - Supervision 2 1/75 2 24 f 2 2 Supervision 3 7-8/75 5 90 a,d,e.f 2 2 e supertleson i 12/75-1/76 4 56 a,f 2 2 N Supervision 5 5-6/76 3 42 d,f'j 2 I H Supervision 7 6/77 3 42 * f.d 2 2 H,T Supervision 8 11/77 3 30 s f 2 2 H1T Supervision 9 6/78 2 tO a,f 3 2 H,T Supervtsion I0 1O-/llt7 6 91 a,c .4,f 3 1 H,T + DUd-tera Evalestton Supervision 11 3/"' 3 42 a,d.f.g 3 2 H,S Supervision 12 7-6/79 3 42 a,d,f 3 I K,T Supervision 13 3/80 2 36 s.d 3 1 H,T Supervision 14 11-12/80 1 18 d 2 1 H,T Supsriliaton 15 3-4/81 3 40 a,d,f 2 1 M,T Supervision 16 11/d1 2 24 d.e 2 I H,T Subtotal 659 Total 1,037/k o8HER PROJECT DATA Sorrover Government of Tanzania tsxecuting AgSency Kigolon Regional Adsitdstrrti*n Fiscal Year July I to June 30 sMeeof Currency (abbreviation) Taeos-lan Shilling (Tah) Currency E=change Rates Appraisal Year Average (1973) US$ 1.00 - 7.14 tntexveaing Years Avera-e (1974-81) US$ 1.00 - 7.90 Caspletfon Year AveroSe (1902) US$ 13.00 - 8.2 Follov-on Projects None 1a Slow ,implesentatien. late forailly extended to 12V82 Dose structures out completed when project funds eoxhausted. /b Calculated frees Hard epproval date 7Le Date extended to 12/83. the final disbuteseot wase nde in Septe8bet 1982. /4 Population in project villagos based on 1978 censce figures. enetfits vary asose villages depeodtig on servtces provided. / -a3riclcorc a b * 08t econoticl c - financial at0l070i1 d - engioeerfeo; e -o gciogy/anthropology; a grconouica 8 * hesithl h0 - figheri@5a I - eduatioonl j * land use plannint h tourtag I housing end urhan develepnent. /f I - probile free or minor .e al 2 - moderate problems; 3 - eajor problems. I - Ieprovmng; 2 - tatioenry; *nd 3 doterlorttng. a tE - financittll N - masnagerial; T - technical; 7 - politi.all and 0 - other, 7f Port of an ecsnosic nsisaioo. Sleues separate utesioaL P during the period 5/72 to 6W73. 6 About 152 to 202 spent on in-country travel. COPY BEST AVAILABLE  - iv - PROJECT PERFORMANCE AUDIT REPORT TANZANIA KIGOMA RURAL DEVELOPMENT PROJECT (CREDIT 508-TA) HIGHLIGlITS This project was the first integrated rural development project to be supported by IDA in Tanzania. It was also one of the first "new style" rural development projects to be supported anywhere by the Bank Group, fol- lowing the new lending guidelines announced at the 1973 annual meeting in Nairobi. The project aimed to develop agriculture and village Infrastructure in 135 villages in Kigoma Region, one of the poorest regions in Tanzania. The project Included components for agricultural extension, training and research, farm credit and input supplies, rural access roads, hcalth and edu- cation. Cotton was to be grown as the principal cash crop, while maize, beans and groundnuts were to be the main food crops. It was expected that this project would be the first phase of a longer-term development program for Kigoma region. Actual achievements under the project fell far short of those expected at appraisal. During the project period agricultural production has increased in Kigoma Region, but most of this increase was obtained through farmers increasing the area cultivated using traditional methods. The pro- ject aimed to help farmers to increase yields through using improved cultiva- tion techniques, fertilizers and better seeds. This approach has been largely unsuccessful, mainly because the technology being offered was unat- tractive to farmers, while implementation capacity was very weak. Develop- ment of village infrastructure was also poorly implemented. The project coi-c pleted about half of the school buildings which were planned, but less than 10 of the project villages had a functioning water supply system as planned at appraisal. The PCR estimates that the project shows a negative rate of return and this project must be regarded as a failure. Some other lessons learned and points of interest are as follows: ~ - at the beginning of the project period the countryside was in a state of upheaval because of the massive villagization program being implemented at the time. At appraisal it was expected that villagization would proceed much more slowly and the project would contribute to successful village development through improving pro- cedures used for village site selection and planning. This proved impossible, for the pace of viLlagization was increased substan- tially after appraisal and it had been essentially completed by the time the project started.  v - capacity was weak in Kigoma Region and this project implementation situation was aggravated by the complicated project design with many participacing agencies. When the project started, the decen- tralized system of Government administrationwas not well estab- lished, for it had been introduced only a short while. One key agency, The Kigoma Co-operative Union (KCU), which was to have played a key role in the project, was disbandedby Governmentright at the start of the project. The functlons should have which KCUT fulfilled were assumed by the regional authorities and several parastatalorganizations, but these agencieswere unable to perform these tasks effectively. The Tanzanian Rural DevelopmentBank, which was expected to play a major role in supplying farm inputs on credit, ceased doing so part way through the project period. from the outset it was accepted that this was a risky project, in part because the agriculturaltechnology to be used under the pro- ject had not been tested effectively in Kigoma Region. For this reason, high prioritywas given in the project design to establish- ment of a crop trials program under the project. Unfortunately, during project implementation this was not given the priority it deserved and by the end of the project period an effective trials program had still not been established in Rigoma.  PROJECT PERFORMANCE AUDIT MEMORANDUM TANZANIA KIGOMA RURAL DEVELOPMENT PROJECT (CREDIT 508-TA) I. SUMMARY 1. This was the first rural development project in Tanzania .o be sup- ported by IDA. It was also one of the first "new style" rural development projects to be supported anywhere by the Bank Group, following Mr. McNamara's speech at the 1973 Nairobi annual meeting. Detailed project proposals were completed by Government in 1973, with assistance from the Bank's Resident Mission in Eastern Africa. The project was appraised in October/November 1973, and the Credit became effective in November 1974. 2. Within the context of Tanzania's policy of villagization, the pro- ject aimed to help develop Kigoma Region, one of the most remote and poorest regions in Tanzania, although it was a region with relatively high agricultural potential. The project was to contribute to regional development through a broad range of measures with an annually increasing number of selected project villages. Principal project activities included village site feasibility studies, for the proper location of new villages, and provision of vital village infrastructure, including drinking water supplies, access roads, health, education and crop storage facilities. Farm incomes were to be increased through intensification of production. The project aimed at strengthening agricultural research, training and extension, farm credit, input supply and marketing services. The project was to encourage production of cotton as the main cash crop and maize, beans and groundnuts, primarily as food crops. The project also involved strengthening the regional administration, while it included a component for preparing further rural development projects in other parts of the country. 3. The Kigoma project was expected to be the first of three successive phases of this project, for, from the outset, it was acknowledged that rural development was a long-term process and not much could be achieved in one phase of five years. The project was also considered to be an experiment which would provide valuable lessons for the design of poverty-oriented rural development projects in Tanzania and elsewhere. On this basis, the Bank was willing to accept the higher than usual project risks posed by the project's scanty data base and by untested new country policies. 4. Total project costs were estimated at appraisal to be T.Shs 94.6 million (US$13.3 million). This was to be financed tiith an IDA Credit of US$10.0 million, and with a grant of US$1.5 million from the United Nations Capital Development Fund (UNCDF) for the drinking water and health com- ponents. the balance of project cost was to be financed by Government. Actual project costs amounted to T.Shs 125.0 million (US$15.8 million), equi- valent to an overrun of 19% when expressed in US dol.l.ars and 32% in Tanzanian Shillings. The cost overrun was funded by Government.  - 2 - 5. Great difficulty has been experienced in implementing this project, and it took eight years to disburse the Credit compared with five years expected at appraisal. The project had many different components with a relatively large number of agencies responsible for implementation. This would have made project implementation more difficult even in "normal" cir- cumstances. In practice, conditions were far from normal. The decentralized system of government administration had been introduced in 1972 and had not been firmly established when the project started. The countryside was in a state of upheaval during the early part of the project period as a result of the massive villagization campaign, which involved resettling most of the region's farmers in new villages. Project implementation thus proceeded very slowly during the first few years. Although the pace of development picked up later on, project implementation became increasingly difficult during the last few years of the project as a result of the deteriorating country-wide economic situation. This was characterized by increasing financial strin- gency and serious shortages of essential project inputs, such as building materials, fuel and spare parts for vehicles and equipment. 6. Project implementation was also influenced by two key events which took place soon after the project became effective. Firstly, the Kigoma Cooperative Union (KCU) was disbanded. The original plan for the project expected that KCU would play a central role in proJect implementation, pri- marily through supplying farm inputs on credit and marketing crops from vil- lage cooperatives. Although the functions which KCU should have fulfilled were transferred to several parastatal organizations and to a new Project Management Executive (PME), established in the office of the Regional Development Director (RDD), these tasks were never fulfilled adequately. Secondly, Government implemented the villagization program much faster than expected at appraisal. At the time of appraisal, about 25% of the rural population in Kigoma Region had been established in about 76 villages. It was expected that villagization would be completed by 1980 and that the pro- ject would play a major role in improving procedures used for selecting sites for and establishing new villages. This proved impossible in practice, for Government decided to accelerate the pace of villagization, and this had been essentially completed throughout Kigoma Region by the time the project started. 7. The procedures which had been designed under the project to assist with improved site selection and better procedures for development of new villages were thus never used for that purpose. Instead, they were used for deciding whether villages which had been established should be provided with assistance under the project. The result of this was that ths project tended to concentrate assistance on the best endowed villages, while neglecting the poorer villages, at least until the later stages of the project period. 8. Primarily because of the problems encountered in implementation, a mid-term evaluation of the project was undertaken in 1978 and many of the original targets for the project were scaled down. The mid-term evaluation concluded that "the estimates or targets set out at appraisal were beyond any reasonable probability of realization within the time frames set. One con- sequence is that the real achievements of the project are overshadowed by its  - 3 - very evident shortfalls in relatiou to the appraisal targets.l/ This remains largely true today, especially in the areas of education, village storage and feeder roads. Ilowever, in the case of agriculture end drinking water, two of the principal project components, progress under the project has been clearly unsatisfactory. 9. The available statistics relating to agriculture are scanty and an accurate assessment of the project's impact oiiagriculture cannot be made. During the course of the project period it appears that there has been a sub- stantial increase in agricultural production in Kigoma Region, especially for food crops. The audit believes that this increase has taken place for two main reasons. Firstly, now that the initial upheaval of villagization has passed, farmers liave been able to devote more effort to cultivation, rather than constructing new buildings or clearing land. Secondly, over the course of the project period, prices, especially for foodcrops and coffee, have risen significantly on unofficial markets, and this has provided a major stimulus to production. However, it appears that most of the increase in production has been achieved through farmers ex-anding the area cultivated, using traditional cultivation techniques, and ;he project can claim little responsibility for this increase. The proJict aimed to increase yields of cotton and foodcrops through promoting the use of fertilizers, improved seeds and pesticides. This has been largely unsuccessful. Production of cotton, which was to have been the principal cash crop under the project, has increased very little. The total quantities of fertilizer used have been far below the appraisal projections and increased foodcrop production cannot be attributed to much higher input useage, as expected at appraisal. 10. At the time of appraisal, there was considerable uncertainty about the suitability of the technical packages being offered to farmers. For this reason, the appraisal report proposed that a crop trials program, designed to test alternative crop development strategies, should be started early in the project period. The need for this crop trials program was again emphasized during the mid-ternievaluation. Unfortunately, a trials program of this type has not been successfully implemented in the region, and at the time of the 2 audit mission's visit no trials were being conducted. / Even in instances where appropriate improved technology was available, it was felt that the extension staff did not make sufficient effort to promote this during the course of the project. 11. The appraisal report expected that 110 villages would be provided with drinking water supplies under the project, including 55 villages with piped water systems and 55 villages with water supplied from shallow wells. As progress with development of water supplies was very slow during the first few years of the project, the targets were reduced during the mid-term evalu- ation mission to 24 piped water systems and only two based on shallow L/ Mid-term evaluation mission Back to Office Report of December 7, 1978. 2/ Even if the trials program had been implemented as planned at appraisal it would have been difficult to have obtained significant trial results and have used them througnithe extensiornservice within the original time frame of the project.  - 4 - wells, for the latter proved to be less suitable than expected. By t:hcend of the project, nineteen piped systeimsand two shallow well systems had been completed find this was very close to the revised project targets. Howe,er, only about onci-thirdof these water systeinswere funct~.oningproperly. Rost of the piped systems were operated by ptimps,and operation of these proved to be a major problem. Spare parts and fuel were frequently not available, while some engines were stoLen. Construction costs for piped systems also proved to be very high. Some of the water schemes also failed to function properly -due to poor design. Overall, performance of the water supply corn- ponent of the project was unsatisfactory. 12. Progress under the project with improvement of educaticnal facili- ties has been relatively good. Two hundred and twenty one classrooms were constructed and this was equivalent to 96% of the revised project targets, or 58% of the appraisal targets. Roughly comparable p.rgress was made with con- struction of teachers' houses, offices and scAioolworkshops. However, the project did not succeed in supplying all of the furniture and equipment which was needed for use in these buildings. 13. For the health component, five new dispensaries were constructed, and this was 50% of the target set at appraiqal (and retained following the mid-term evaluation). Three rural health centers were upgraded under the project, as planned at appraisal. Seventeen dispensaries were improved, although those were not included in the original project proposals. While progress with physical construction of health facilities was relatively suc- cessful under the project, most of these do not have adequate equipment, and there has been a serious shortage of drugs. This, combined with the inadequate water supplies in most villages, means that the health services available to people are still generally very poor. 14. For feeder roads, the project was able to exceed the appraisal targets; 619 km were upgraded compared with the appraisal estimates cf 380 ku. However, for some of these roads only very minor improvements were carried out. These roads were simple tracks which fulfilled the important function of linking project villages with the main roads in the area. The project also completed construction of 22 village stores. Although this was substantially less than the appraisal target of 95 stores, this was the same as the revised target established following the mid-term evaluation. The project was also able to train 180 village bookkeepers compared with the appraisal tavget of 100. 15. In addition to the project comuonents which were specifically aimed at development of Kigoma Region, the project included support for a team of people who were responsible for preparing proposals for rural development projects in other regions. Three regional projects were prepared by this team, for Mwanza/Shinyanga, Mara and Kagera ]Regions. The project in Mwanza/Shinyanga was appraised by the Bank and became effective in March 1979. This project is itself now experiencing many problems similar to those encountered in Kigoma, although these are to a considerable extent due to country-wide economic problems. The project in Mara Region was appraised by the Bank but it was decided not to process this any further. The project for Xagera Region has not been appraised.  -5- 16. During the course of Kigoma Rural Development Project, more than 50% of all development expenditures in the region were derived from the project. Now that the project has been completed, a very difficult situation has been created. The major source of financial support for new development in the region has been terminatea and the regional authorities have had great difficulty maintai.aing activities which were started under the project, or completing certain work which was incomplete at the end of the project. This lack of continuity in support for the project, combined with the many problems experienced during implementation, particularly the failure to achieve the expected economic benefits from agriculture, means that the project must, in general, be regarded as having been unsuccessful. Although, it has not been possible to re-estimate a precise economic rate of return, the PCR suggests that this would be negative (PCR, para. 5.05). II. MAIN ISSUES A. Project Organization 17. The project involved a range of different components, all of which were to be implemented through the existing administration in Kigoma Region. The regional administration, under the Regional Development Director (RDD), was responsible for overall project coordination and implementation of components such as health, education, water, agriculture and construction of feeder roads. At the village level, each village formed a cooperative society responsible for marketing farm produce and providing farm inputs, while the Kigoma Cooperative Union (KCU) was to have been responsible for these functions at the regional level. The Tanzania Rural Development Bank (TRDB) administered credit through cooperatives, while the Tanzania Cotton Authority (TCA) promoted production of cotton. The National Milling Corporation (NMC) was responsible for purchasing foodcrops produced in the region. 18. The project required a substantial increase in the volume of work to be performed by all of these agencies and all of the components were to be implemented simultaneously. Yet, most of these agencies had been established 3 in Kigoma for only a short while and/or were not firmly established /. The decentralized system of Government administration was set up in July 1972 and had not had a chance to become well established when the project started, especially in view of the scarc'otyof experienced staff to fill the increased number of positions created for the new decentralized system. The regional administration also faced an unprecedented workload during tl.eearly part of the project period because of the very rapid pace of villagization being implemented under "Operation Kigoma". This was carried out far faster than expected at appraisal. TRDB had established an office in Kigoma Region in 3/ At the outset of the project there was very little activity by TRDB, TCA or NMC in Kigoma Region.  -6- 1972 and had only a handful of staff when the project started, while KCU was financially weak and poorly managed. 19. The risks created by the weak administration in Kigoma Region were clearly recognized at appraisal, as the following extract from the Issues Paper indicates - "the project has a greater than normal degree of risk due to the unproven and revolutionary nature of ujamaa settlements and its organization of agricultural production, the scarcity of administrative and technical skills in the region, the extreme weakness of regional agricultural input and credit delivery systems and marketing, and the logistics needs created by the region's moving from a subsistence to a cash economy".4 / In order to try to compensate for some of these risks, the appraisal report proposed that project implementation capacity should be strengthened in the region, primarily through providing key additional staff and other support for KCU. In retrospect, these measures seem to have been very modest in relation to the need for a substancial strengthening of regional administration. But during the period when the project was appraised, the Bank adopted an optimistic stance towards Tanzania's untried and revolutionary policies. 20. The measures designed to strengthen KCU never became effective, for Government decided to disband KCU at the beginning of the project. The functions which KCU should have performed were transferred to a specially established Project Management Executive (PME) in the RDD's office and to TRDB, TCA and NMC. The Project Management Executive was a small unit headed by a project coordinator and included an expatriate financial controller. Although this unit performed useful planning and coordinating furctions, it was not sufficiently well integrated into the regional administration, and it was not in a position to strengthen the implementation capac'ty of the agencies which had direct responsibility for project implementation. The parastatal organizations, TRDB, TCA and NMC, which had assumed responsibility for some functions given up by the defunct KCU, were weak or were subjected to substantial political interference. The "extreme weakness of regional agricultural input and credit delivery systems" referred to in the Issues Paper was something which was to plague the project throughout its life. TRDB was not able to implement an effective credit program for farm inputs and, at times, there was confusion about the respective roles of TRDB and TCA. During the initial stages of the project, TRDB and TCA both distributed significant quantities of fertilizer and other farm inputs to villages, as part of the Government's program for supporting newly established villages. However, these inputs were distributed in a haphazard manner, the quantities provided were not based on a realistic assessment of the amounts which farmers could use effectively, and, in many cases they were "dumped" on the villagers. Farmers did not understand from the start that these iTLputswere being distributed under a credit scheme, and there had been a previous history of free distribution of farm inputs in Tanzania. TRDB also had very few staff available to administer credit for villagers. As a result, recovery of credit proved very unsatisfactory, and TRDB stopped providing credit for farm inputs in 1979. This was a vital weakness in the project, 4/ Para. 1.06 of the Issues Paper of December 11, 1973.  - 7 - for one of the principal aims of the project was to increase crop yields through the use of fertilizers and other inputs. TCA continued to supply some farm inputs in Kigoma, for it was able to withstand the poor repayment position better than TRDB because of its income from cotton marketing. 21. The complicated project design and the problems experienced with implementation also made it more difficult for the Bank to supervis-4 this project. For Kigoma Rural Development Project the Bank has used an average of 59 weeks of staff time per year for supervision. This is almost 5 times the average for Tanzania as a whole (12 weeks) over the same period and is in marked contrast with the requirements of a simple single product/single agency project, such as Sao Hill Forestry Project, which required only an annual average of 8 weeks oL supervision time. Even with this substantial supervision input, it has not neen possible to supervise each individual component adequately. For example, a specialist in health has vSsited the project only once, while no education specialist has ever visited Kigoma on a supervision mission. An engineer with experience in water visited Kigoma on only one or two occasions, and a financial analyst has visited only once, despite the fact that there have been continual problems with the project accounts. Thus, members of supervision missions have been expected to deal with specialized fields in which they have little expertise. This seems inevitable where the Bank supports projects like Kligoma with so many different components. The audit feels that supervision would be both easier and more effective if project desigr.were simplified.5 / 22. In concluding this section, the audit would like to emphasize what it feels is the most important lesson to be drawn from the experience with the Kigoma Project. This is that complicated project designs should be avoided and the scale of development proposed must be consistent with the capacity to implement. Where these conditions dn not hold, as was the case in Kigoma, less ambitious development plans s ould be proposed, and the project should concentrate attention on developing the implementation capacity of local institutions. This conclusion is supported by experience with many other projects apart from Kigoma Project. For example, the PPAR (OED Report Number 4617) for the Integrated Rural Development Project in Mexico (PIDER I) mentions that this project experienced serious problems because "large scale diffusion was initiated before either the technological packages or the institutional mechanism to implement them were perfected."6/ This is precisely the same experience as in Tanzania. B. Agricultural Impact 23. Statistics on agricultural production for KigouiaRegion are scanty and some are difficult to interpret, unreliable or even misleading. One of the reasons for this is that a large proportion of trade, especially in 5/ Problems with supervision of Kigoma Rural Development Project are discussed more thoroughly in OED Report No. 2858, Operational Policy Review, The Supervision of Bank Projects. 6/ PPAR, para. 114.  - 8 - foodcrops, takes place through unofficial channels. For this reason, an analysis of changes in agricultural production which have taken place in Kigoma Region over the past few years must be to some extent impressionistic. The audit believes that over the last few years there has been a significant increase in agricultural production in Kigoma Region, especially for foodcropcs The official statistics for crop marketing show that the volume of crops marketed through official channels increased substantially from the beginning of the project period in 1974(75 to 1980/81 (see Table 2.b in the PCR). Dtiring this same period, crop marketing through unofficial channels has probably increased even more, as prices in unofficial markets have recently been as much as three or four times the level of official prices.7/ The audit supports the view expressed in the PCR that this increase in production has occurred primarily through farmers expanding the area cultivated using traditional methods of husbandry, not through using modern farm inputs to increase yields in the way intended under the project. This increase in production has occurred for two main reasons. Firstly, the process of villagization seriously disrupted agricultural production during the early part of the project period when farmers were moved to new villages and had to concentrate on construction of houses and buildings and clearing land. Although farmers have since been able to concentrate more on farming and increasing production, this is not to suggest that all is now well in these villages. There is considerable evidence suggesting that this is not the case. Many villages were poorly sited and it was unfortunate that the project was not abla to help prevent this in the way expected, as villagization was completed so rapidly in the period between appraisal and the time the project started. Villages are now experiencinig increasing problems due to the inadequacy of the available land within their immediate neighborhood. Some land near to the villages is also now getting exhausted, and it is becoming increasingly necessary to apply more fertilizer to existing land or to cultivate new land a long way from the village.8 / 24. The second factor which has encouraged production has been the existence of a thriving unofficial market for many crops, especially foodcrops. Although this type of market is probably significant all over Tanzania, it is particularly important in Kigoma, which is on Tanzania's international border with Zaire and Burundi. In some cases crops can be sold at relatively high prices in international markets or they can be bartered for other comodities which are not available in Tanzania. These same factors have probably been very important in discouraging production of cotton in Kigoma Region. Unlike foodcrops, cotton can be sold only to TCA at the official price, and this makes cotton production unattractive. This has been a major problem for the project, for cotton was expected to be the principal cash crop under the project. 7/ See chapter II, Section F of the Tanzanian Agricultural Sector Report, August 1983, for a discussion of this subject. 8/ See the Tanzania Agricultural Sector Report, pp. 144-149, for a discussion of this. 25. The project was intended to increase crop yields through the use of better methods of husbandry and purchased inputs. Large increases in yields were expected, as the following data indicate:9 / Traditional Advanced Husbandry Husbandry Increase (kg/ha) (kg/ha) (%) Maize 600 1900 217 Beans 280 670 139 Cotton 390 1120 187 Groundnuts 390 790 103 In the case of maize and beans, intermediate stages of development were suggested between the traditional and advanced levels. 26. The evidence available from Kigoma indicates fairly conclusively that the project has not been able to achieve these targets. In the case of cotton, total production statistics are available and, as there is no other market for cotton apart from purchase by TCA, the statistics can be considered reliable. These indicate that, while there has been some increase in cotton production in Kigoma over the last few years, the total quantity of production is way below the amount expected in the appraisal report. Total seed cotton production in 1980/81 was only 2,365 tons compared with the appraisal targets of 10,700 tons in 'Year 5 and 13,900 tons in Year 8 (PCR, Table 4.b). Even this level of production couLd be maintained only through a considerable amount of official encouragement or coercion, for all farmers in cotton growing areas were expected to grow a minimum acreage of cotton. As mentioned above, the principal reason why farmers were unwilling to increase cotton production was that production was unattractive compared with other crops which could be sold for high prices on unofficial markets. Cotton is also a labor demanding crop and was unattractive for this reason also. However, even if cotton prices had been higher, it is unlikely that farmers would have been able to achieve the high cotton yields forecast at appraisal. 27. For maize and other foodcrops, the audit feels that none of the available statistics is reliable, and it is therefore difficult to be confident about how production has changed in recent years. However, the quantities of fertilizer and other inputs being used in Kigoma Region are well below the amounts expected at appraisal, and this confirms that yields cannot have increased in the way expected at appraisal. In 1982, for example, the total quantity of fertiLizer supplied to the region was of the order of 400 tons, whereas the appraisal report fore.-ast that incremental fertilizer use would be more than 4,000 tons by Year 8. TRDB, which was expected to be the principal supplier of fertilizer under the project, has suspended credit sales of fertilizer since 1979, due to its very poor repayment experience. and only minimal quantities of fertilizers have been sold on a cash basis by the regional agricultural staff since then. This is particularly unfortunate because the need for fertilizer is increasing In the 9/ Data from the appraisal report, Annex 3, pp. 13-16.  - 10 - area, now that cropland near to villages has been under cultivation for several years. 28. It was clearly recognized at appraisal that crop improvement packages being offered to farmers were based primarily on expjrience outside Kigoma Region, and there was considerableuncertainty about how appropriate these packageswould prove to be in Kigoma. For this reason, the Development Credit Agreement (DCA) required that the Borrower should complete constructionof an agricultural trials and training center (ATTC) within 18 months of project effe(tiveness(DCA, Section 3.05). In practice, this ATTC was not constructed wi.hin the time expected, and the buildings at the site selected in Kasulu had not been 100% completed at the end of the project period. More important than progress with physical construction is the status of the agriculturaltrials program for Kigoma Region. This has been very disappointing. Although a few trials have been conducted at several locations during the project period, the results of these seem to have been inconclusive. During the audit mission, none of the agricultural staff at either the ATTC (now renamed a Ministry of AgricultureTraining Institute,or MATI) or from the Kigoma Regional Office were able to indicate the nature of the results obtained from these trials. During the current crop season, no program of crop trials is being carried out. It seems clear that the original aim of developing a crop trials program in Kigoma, so that the most appropriate advice based on a local experience could be offered to farmers, has not succeeded. 29. The failure of the project to achieve its agriculturaldevelopment goals is a major shortcomingof the project, for agriculturewas expected to be the source of economic benefits from the project. The PCR has not attempted to recalculatea precise economic rate of return from the project, due to lack of adequate data. However, the PCR suggests, and the audit agrees, that this would probablybe negative. C. Long-TermSupport for Rural Development 30. Kigoma Rural DevelopmentProject was originallyexpectedto be the first phase of a long-term project,probably involving three phases totalling 15 years. Experience from this and other projects supports the idea that assistance is needed over a relatively long period of time if worthwhile achievementsare to be made. Typically, during one phase of about 5 years, most of the project infrastructure can be completed, but development activitiesbecome fully operationaleither towards the end of the first phase or during later stages of the project. It is not generally realistic to expect that a process of self-sustaining rural development can be initiated without support stretchingover a period of at least a decade or more. 31. Unfortunately,the experience with Kigoma Project has been the reverse of what was needed for sustained rural development over the long-term. Too much was attempted during the initial stages, and this could not be implemented properly. The technical basis for the recommended agriculturalpackages was weak, and an agriculturaltrials program to correct this deficiency was not carried out. Now that IDA support for the project has terminated,Government is having difficulty completing some of the work  - ,1 - which was unfinished at the end of the project and provlding adequate recurrent financing to maintain activities started under the project in an operational status. The audit feels that it would have been better if the original project design had been simpler, iv high priority had been given to agricultural trials, and if uajor emphasis had been given to strengthening the implementation capacity of the region through development of staff and institutions. The audit acknowledges that, even if project design and implementation had been improved in this way, and donor support had been available to continue the project beyond the end of the first phase, the increasingly unfavorable country economic situation would have made project implementation very difficult.  - 12 - TANZANIA KIGOKA RURALDEVELOPMENTPROJECT PROJECT COMPLETION REPORT August 16, 1983 Eastern Africa Projects Department Southern Agriculture Division  - 13 - TANZANIA KIGOMA RURAL DEVELOPMENTPROJECT, CR. 508-TA PROJECT COMPLETION REPORT I. BACKGROUND A. The National Economic and Policy Context 1.01 Tanzania had roughly 90% of its 16 million inhabitants living in the rural areas at project identification in 1971, and derived 40% of its GDP and 80-85% of its export earnings from agriculture. In 1967 the Tanzanian Government and the sole political party (TANU, later CCI) embarked on a policy known as Ujamaa (broadly tran:latable as African Socialis-), which was to transform the economy and society according to the principles of equity and self-reliance. 1.02 At the time of project formulation in the early 1970s, Tanzania was in the midst of some major institutional reorganizations with respect to its rural policy. First, since the Arusha Declaration the leadership had been encouraging the resettlement of the scattered agricultural population into villages, where it would provide basic social infrastructure and encourage them to practice communal farming. Beginning in 1971, the Party and Government began to take a more active role in pushing what had been a lukewarm initial response to this policy among rural dwellers, by undertaking area-specific resettlement operations, with the effect that over 2 million people were living in villages by the end of 1973. Second, with the aim to facilitate local planning and implementation of rural development programs, the Government decentralized admdnistration in mid 1972, placing all functional ministry officials (agriculture, works, health, education, water, cooperatives, etc.) under the responsibility of top administrators in each of the 17 regions, with a similar structure in the smaller district units the regions were composed of. B. The Project Area 1.03 Kigoma, one of Tanv!ania'spoorest and most undeveloped regions, did not have a viable cash economy in agriculture because of its remoteness from national marketing links, and had traditionally been one of the country's major labor exporters, both to the East Coast sisal estates and more recently to individual tobacco farming in neighboring Tabora region. Estimated per capita income was U'$20 per year, compared to a national rural average of US$85. Educatioral facilities were also sub-standard, leaving over 80% of the adult poptlation illiterate and only 32% of the school age population in primary school in 1972. Health status, thought to represent national standards, was a debilitating force for the rural population, many of whom were chronically affected by malaria, tuberculosis, and other parasitic and infectious diseases. Beginning in mid-1972, the leadership chose Kigona as one of the areas for accelerated villagization under "Operation Kigoma" and began moving people from the overcrowded highlands to the intermediate and lowland miombo zones, previously unsettled because of dense tsetse infestation.  - 14 - C. The Project Rationale 1.04 The Kigoma Rural DevelopmentProject (KRDP) estimated to cost US$13.3 million,was approvedin August 1974, and became effective on November 20, 1974. It consistedof an IDA Credit of US$10.0 million and a paralLelUNCDF grant of US$1.5 million for water and health facilities, with the remainder to be met by the Government. Though it followed an establishedBank Group involvementin Tanzania's agricultural sector,'/ it represented a significantbreak from past lending strategies,in that it was one of the Bank's early experiments in implementinga multi-sectoral, "integrated" approach to rural development,itself an outgrowth of discouragingexperiencewith the earlier more narrowly defined rural projects which failed to lift the multiple productionconstraintsfaced by the rural poor. 1.05 The choice of Kigomnafor the experimentwas based on several favorable conjunctions. First, the Government'sstated objectivesof focusing of the rural poor in developmentefforts coincidedcompletelywith the Bank's new directionsin lending,as expressed in Mr. McNamara'sspeech of 1973. Second, the Bank consideredthe recent decentralization of the Tanzanianadministration as conducive to an integratedproject. It would allow area-specific integrationof the various administrative sectors,and funds could be channeleddirectly to the region rather than through each individualcentral ministry. Finally,the choice of Tanzania, as one of the world's poorest,primarilyagriculturalcountries,and Kigoma, as one of the most remote and deprived regions within Tanzania, was well in line with the Bank's intentionsto focus its lending on the "lowest 40% of the populationin developingcountries."2/ 1.06 The project'smain objectivewas to double rural incomes through a range of economicand social infrastructural support to newly established villages in Kigoma Region in conjunctionwith OperationKigoma. The implementation experiencesuggests that the project's risks, considered higher-than-normal at appraisal,were inordinate, and that the project has significantly overspentand underachievedin relation to its original objectives. It failed to generate most of its intended impacts,its strongest legacy being unintendedimpacts, both positiveand negative in nature. I/ Six IDA credits,totallingUS$62.1mill.on (one each in agricultural credit, tea, tobacco, cotton and two in livestock),a loan of US$21 million for a cashewnutproject, and US$3.3 million earmarked for agriculturaltrainingin a larger educationalcredit had also been appraisedfor Tanzania. The performancerecord in these projectshad been mixed (Appraisal Report, p. iii). An Agriculturaland Rural DevelopmentSector Report was being prepared at the same time as the preparationof the Kigoma Project. 2/ As expressedin IDA press release No. 74154, August 8, 1974.  BEST AVAILABLE COPY Follow-up and Completion 1.07 In November 1980, a Bank team in the region requested regional officials to prepare components for a possible Phase II funding. Over the course of 1981, the Bank's views changed because it was felt that national economic priorities were no longer consistent with a second IDA-financed phase of the Kigoma Project. Regional officials understandably lost their enthusiasm with the Bank over this experience. No completion mission took place. 1.07 In the absence of a completion mission, this completion report is based on: a. a working draft prepared by the Agriculture and Rural Development Department which had supervision responsibility for the project; b. an independent evaluation conducted in the last year of the project by Michael Loft, Ellen Hianakand Ahmed Ndyeshobola, Economic Research Bureau, University of Dar-es-Salaam, An Evaluation of the Kigoma Rural Development Project (IDA-508), E.R.B. Restricted paper No. 821, March 1982 (herein referred to as Project Evaluation Report); c. the report of the project's farm management survey Sy Michael Loft and Joris Oldevelt, Economic Research Bureau, University of Dar-es-Salaam, Developments in Peasant Agriculture in Kigoma Region during the Post-Villagization Years, 1981; d. Supervision reports and other documents in the project files, including Report No. 453a-TA, Appraisal of Kigoma Rural Development Project, Agriculture and Rural Development Department, July 15, 1974; and "Agriculture and Rural Development Department, Rural Development Division, Tanzania: Rural Development Project - Kigoma, Mid-Term Evaluatioia Report", July 1979. II. FORMULATION AND DESIGN A. Introduction 2.01 Many of the problems experienced under the KRDP were indelibly stamped on the project during the formulation and design phase. While the Bank and the Government agreed in principle on the advantages of the Government's rural policy, discrepancies in intent emerged on a number of practical points of implementation, and the compromises which ensued undermined the possibilities of effective implementation of a number of components.  - 16 - B. Preparationand Appraisal Process 1. Identification 2.02 Followinga short visit to Kigoma, a Bank Economic Mission to Tanzania in August/September 1971 identifiedthe region as an area with considerableuntapped agriculturalpotential,suitablefor a low-cost, integratedregional rural development scheme in line with the Government priorityof assistingthe more backwardareas of the country. It is reportedlyon the basis of a draft version of the mission's report 3/ that President Nyerere decided to request Bank assistancefor such a program. 2.03 Meetings between the Governmentand RMEA began in late May 1972, and includeddiscussionswith the Presidenthimself quite early in the process. The Presidentand and his aides articulatedthe leadership's view of the rural developmentpolicy: the first objectivewas to bring people togetherinto villages, whereupon basic social services-- health, water and primary education -- could be providedeconomically and whereby a common sense of purpose and motivationcould be generatedamong them. Productiveinvestmentswould then follow in a subsequentstage. Villages would be carefullyplanned so that they could evolve into viable self-reliantunits. Bank and Governmentofficials thereforeagreed that the proposed Bank-assisted project should be related to the regional authorities'(KRA) village developmentprogram,as this was to form the core of Governmentactivity in the region for some time to come. 2. Preparation 2.04 After these preliminaryagreements,another eight RMEA missions visited Tanzaniabetween July 1972 and June 1973 to assist in preparing the Government'sproject proposalto the Bank. The Governmentproved less able to assist in the technicalaspects of this process than the Bank had initially envisaged. 2.05 RMEA prepareda draft RegionalDevelopmentProgram outline largely reflectingits own concerns of establishing villagesand promoting agriculturalproductionat minimum costs. As this was the first proposal from the Bank's side with some degree of elaboration,its characteris worth noting in relation to subsequentchanges made in project design. 2.06 The program provided for essentialvillage social and economic infrastructure, crop and livestock production,improvements in the region's marketing network, a credit unit to handle short and medium-termloans to farmers,and a programming and evaluationunit which would assist in the preparationof village land use plans, give advice on procurements and forward planningand evaluate implementation progress. In order to provide village serviceseconomically,a satellitevillage structurewas proposed, within which large scale services would be shared. The proposed approach was cautiousand modest, spanning a developmentperiod of 12 years, financedover three 4-year phases. 3/ "The Economic Developmentand Prospectsof Tanzania",Volume II, June 22, 1972.  - 17 - 2.07 The regionalauthorities' reactions indicatethat they felt considerable pressureto move faster than this. They wanted a much larger project,providing full servicesto all of the region'svillageswithin five years, therebyrejectingthe conceptof shared servicesamong satellite villages. In retrospect, it seems that if therewas a turning point anywherethis was where the projectgot onto che wrong track. An active debate ensued betweenthe Government and the Bank teamA. The compromise reachedwas, in effect,that both sides would do what they wanted -- the KRA would proceedwith villagization at its own pace,and the projectwould come in its wake, providing essentially the above-mentioned program,but only to selectedvillagesconsidered to have a sound agricultural potential. The Bank fully supported the idea of moving people into villAges, which it considereda prerequisite to rapid increases In agricultural productivity (throughtechnological improvements).4 / The Village Site Feasibility Studieswhichwere to have helped locategood sites for villagesthus became simply criteriafor selectingalreadysited villagesfor the projectbecausevillagisation had gone too fast. 2.08 RN$A drafteda workingpaper containingthis compromise. It differedsubstantively from the initialRMEA proposal;the first phase (now 5 years)would cost twiceas much (US$9.8million)and would cover more than twice as many farmingvillages(150),in additionto 14 fishing villageson the lakeshore. 2.09 A number of 'substantivedisagreementsremainedbetween the Governmentand the Bank, and to a lesserextent within the Government. The most importantof thesewere: a. the KRA had only concededto the inclusionof the satellite villageconcept(paras3.83-3.84 of proposal)under pressure from the centralGovernment; b. regionalauthorities did not agrez to the 300-350family villagesize which the RMEA team had considereda maximum; c. the ntmber of projectvillageswas a compromise betweena much largernumber that the KRA had wanted (216), and a smalleramount that RMEA felt was feasible; and d. agreed criteria for the selection had not been established of projectvillages. 2.10 These points of difference of frictionin the gave indications future. J Appraisal 2.11 A seven person appraisalteam visited the region in October/November1973 with a smallurfollow-up visit in March 1974. Followingthe creationof a RuralDevelopment Divisionin the 4/ Letter from RMEA to HQ of June 18 and memorandum from Preparation Team to RMEA Div. Chief of Jtne 19, 1973.  - 18 - CPS/Agriculture Departmentin 1973, responsibility for the Kigoma project changed from RMEA to AGR/RD. Kigoma was to become the first and for a while the only agricultureproject in Tanzania to be superviseddirectly out of Washington* The project was approvedby the Board on August 6, 1974. C. Project Design 2.12 In its final post-appraisal version, the project's scale was again altered, the level of costs increasedto over $13 million, and the number of project villagessliglhtly reduced to 135. The Bank had acquiescedto the Region by droppingthe satellitevillage concept and the 350 family limit to village size, meanwhile introducingcriteria (included in the credit agreement)to screen village eligibilityin project participation. 2.13 The project would consist of: a. investmentsin eligiblevillagesfor public water supplies, education and health facilities,tsetse clearingand feeder roads; b. loans to eligibleand creditworthyvillagesfor: (i) production- infrastructureincludinggodowns,maize mills, fish catchingand processingequipment,and (ii) seasonal productioninputs; union for facilitiessuch c. loans to the regional cooperative as godowns,offices,and transportequipment; d. investmentsin regionalsupportingservices,including:(i) constructionand rehabilitation of rural trainingcenters; (ii) a small regional radio-telephonesystem; (iii) a livestockpilot project; (iv) aerial photographyof areas to be developed;and (v) a regionalagricultural trials and training center. e. technicalassistanceto include: (i) a financial controller, an operationsmanager and a trainingsupervisorfor the regionalcooperative union, a land-useplanner and trials and trainingofficersfor the agriculturaltrials and training center togetherwith training;(ii) back-up support such as housing and transport;and (iii) a project special fund for use by projectmanagementfor specificproblem-solving; f. monitoringand evaluation;and g. project preparationfor rural developmentprojects in other regions. 2.14 The principalobjectivewas to double annual per capita income (from US$20 to USS40) for the roughly one half (250,000)of the region's rural populationresidentin the project villages. Income increases, mostly in cash fort, were expectedto arise from greatly increased agriculturalproductivity, most directlyfrom the widespreadadoption of  - 19 - technical packages for cotton, maize, beans and gioundnuts extended to villagers on seasonal credit. These loans were only to account for 11% of base costs, however (Table la). Far more significant were village level support components at 55%, and regional support for both the administration and the cooperative union at 25% of base costs. The "piggyback" project preparation component, introduced during appraisal, was allocated the remaining 9% of base costs. This change also meant the abandonment of the initial RMEA proposal to have a project evaluation team (lead by an internationally recruited staff member) resident in the project area. The supervision, technical assistance, and design of monitoring and evaluation was now loosely delegated among the responsibilities of a "Project Preparation Team" (PPT) which would be situated somewhere else in the country. 2.15 As anticipated from identification, the project would be carried out under the existing organizational structures of the decentralized regional administration. The Kigoma Cooperative Union's (KCU) role was central to the project's economic viability as it would be responsible for the entire range of crop production and marketing activities. 2.16 Implementation was to extend over five years, with full benefits accruing by the 12th year after project inception, as exemplified by the attainment of the principal objective of doubling total incomes. Income from agriculture alone, principally through cotton sales, was projected to double after eight years. The role of cotton in the marketed output mix had been doubled during appraisal, despite its consistently declining regional production trend since 1967 (from a low 1,000 tons to a meager 390 tons). Though this was never made explicit, the change was probably in acquiesence to growing Government pressure to increase foreign exchange 5 earnings. / On the basis or this projected agricaltural growth, the project's economic rate of return was set at 22%, exclusive of the costs of social infrastructure, or between 12 and 18% including varying degrees of these costs. 2.17 A master plan or schema of the inter-relationships of the projects many components was not explicitly provided. But an understanding of the logic of this schema is essential, both to evaluate the project's design, and to assess project impact. An outline of a schema deduced from the appraisal report is found in Chart 1. The numbers in circles indicate the offices responsible for implementation. 2.18 Schematically, the project's support package was to streamline the adv.ancementof villages through three distinct stages: a. identification of suitable village sites; b. provision of social and economic infrastructure; and c. provision of technological packages. 5/ This would explain the way the regional authorities pushed cotton production on the villagers so vigorously in the first year of the project (see Ch. I1I, Sec. C.1).  - 20 - 2.19 The project was considered to have greater than usual risks, stemming from "the unproven nature of ujamaa villages and the need to create a completely new marketing and distribution network". These risks were not considered prohibitive, however, both because of the project's provision for te::hnicalassistance, but more importantly because of its "potential for improving markedly the incomes of large numbers of rural poor in a disadvantaged region at a relatively low cost". In retrospect, the appraisal seems to have underestimated the risks and not taken adequate account of the following flaws in the logic of the design: a. Villagization. After all the indications the Government had given that it intended to push on with and even accelerate the resettlement program, the final design was still geared to influence Operation Kigoma by inserting technically more sound methods of site selection. The quantity and level of local skilled manpower to do this were simply not available even if the timing had been appropriate. b. Inputs. The widespread adoption of input packages was a cornerstone of the project (from the Bank's point of view). Yet the technical recommendaf:' 7IiSwere based on little hard evidence, as there were no reliable data on agronomic response for the crops in question. Cotton production increments were based on recommendations for the Western Cotton Growing Area, about 150 miles away. However, even in that area the recommendations were starting to be questioned at the time of appraisal. In the 1973/74 Ukiriguru Research Station report the soil scientist noted that "at all sites cotton yields were depressed by nitrogen", and in 1975 he noted that "failure of ammonium sulphate in the recent past has been one of the most marked features ... yield depressions have been more common than increases". Also, the relationship between the provision of input credit and the project's research, trials, and extension training components was not executed in a logical sequence. With such levels of uncertainty in the technical packages a pilot program with crop trials, followed by credit for inputs would have been better. c. Overall timing of impact. Most components directed at the village level were sequenced in the correct chronological order. However, the plan assumptions to have the Stage 2 elements (extension officer, trained bookkeeper, godown and access road) in place and working smoothly within 2 seasons were unrealistic, as was the intention to fine tune the input recommendations early in the project when the I Agricultural Trials and Training Center could not possibly have been operative in less than 2 - 3 years. d. The Project Area's Overall Implementation Capacity. In general the assumptions the appraisal made about the implementing capacity in Kigoma were optimistic in relation  - 21 - to what Bank teams had experienced during preparation. For example, without any technical assistance, the regional water office was expected to double its annual implementing capacity, and introduce methods to ensure the installation of least cost systems. This was a general weakness of the project. Relatively insignificant strengthening of staffing, either local or expatriate, took place alongside very substantial transfer of funds. e. Monitoring and Evaluation. A major design inconsistency was the loose treatment given to monitoring and evaluation. Considering how innovative and untried the project's approach and target area were, i1. seems in retrospect to have been a mistake not to insist on an operational 14& E system within the region from the outset. It could have been estabLished during an early adaptive research phase. III. IMPLEMENTATION A. Immediate Design Changes 3.01 Before the project even got underway, two changes took place in Kigoma which undermined the basis for a number of design premises. First, Operation Kigoma was superseded by the countrywide villagization program which began late in 1973. While the regional program had been relatively gradual the latter was a crash operation. By the end of 1974, villagizatlon was essentially completed in Kigoma, with 95% of the inhabitants accommodated in 193 villages. The authorities did not give much regard to the Bank's concern about size - village settlements often consisted of the legal limit of 500 households, and soue even larger settlements were divided nominally into two separate villages for administrative purposes. The Bank adjusted by turning the Village Site Feasibility Report system into an ex post screening device -- setting the availability of 8 arable acres per household as the principle criteria for project villages. 3.02 The second event was the dissolution of the KCU by Government order late in 1974. The Bank had a warning this would happen about a month prior to the scheduled date of project effectiveness, though the Government's formal announcement was not issued until over a week afterwards. The Bank and the Government agreed to somewhatmakeshift arrangements for parceling out the KCU's project responsibilities. Crop marketing was taken over by the National Killing Corporation (food crops) and an expanded branch of the Tanzania Cotton Authority. The Tanzania Rural Development Bank became directly involved in the village credit program. To fulfill the KCU's general supervisory and coordinating role in the project's economic components, a "Project Management Executive" (PME) was created, under the RDD's office, with a core consisting of the technical assistance positions originally intended for the KCU (some of which had already been filled).  - 22 - B. Overall Implementation Experience 6/ 3.03 The Project got off to a very slow start, with hardly any spending in the first two years. Expendituresdid pick up in the following three years, but at a slower pace than had been anticipatedat project appraisal. Consequently, extensionswere made twice in the credit agreement,and 1982/83was projected as the last project year. Then spending took off so rapidly that by the beginning of 1981/82 less money was left than was needed to completea whole range of investmentswhich had already been started in the various sectors (see Table la for expected and actual expenditures). 3.04 The stark contrastbetween anticipatedand actual rates of expenditure in the first five years is explained by several basic constraints inhibitingthe regional authorities'implementingcapacity. The most immediate constraint,which accounted for the negligible activity in the first two years, was organizational. The KCU's dissolutionhad left the project in an institutional void, and the role and capacity of the PME was neither well defined nor well understood. In retrospect, it is apparent that the Bank and the region had unrealistic expectationsof the PHE because despite its title, it had no executive powers, even within the regional administration it was attachedto, and certainlynot vis-a-vis the autonomouscrop parastatals. As a result, implementation took on a distinctlyuncoordinated,disintegratedcharacter,whereby even agencies directlyconcernedwith the same components (like input credit) were not communicating. 3.05 As implementation proceeded the PKE's sole active role was in the sphere of project finances, where it worked to ensure that project funds were used for project purposes. 3.06 A second constraint, registeredin almost every sector, was manpower, a widespread national constraint. Manpower shortagesmeant that investmentstook longer to develop, set up, and organise, and implement than project design had envisaged. 3.07 A third constraintwas the procurement of materials. The project suffered chronicallyfrom an inability to procure sufficientbasic constructionmaterials (such as cement, corrugated iron sheets, and roofing nails) on which so many of the componentsdepended. Other more specialized equipment,such as pipes and fittingsfor water systems, or desks for classrooms,were similarly difficult to obtain. To some extent, procurementproblems reflectednational economic conditions,and therefore were not easy to forecast at appraisal,but Kigoma's great distance from the country's industrialcentres and main port was also a factor, and predictable. 3.08 While these constraints impeded the overall progress of the project,they did not affect each component equally. Some components, such as water and health, lagged far behind others in percentageachieved of implementation targets. The education component,or at least those parts which were regionalpriorities,and bookkeeper trainingwere faring 6/ See Table 2a for Key Indicators.  - 23 - comparatively well. The more complex and specialized a component was, the more likely it would run into implementation difficulties, given the region's overall constraints. Water systems, which involve a whole series of complex and sensitive operations, would fall at the one extreme, while bush clearing, which involves only a bit of organization and a group of willing peasants, would fall at the other. Construction of classrooms, which could be readily modified to suit the available materials, was a fairly simple, and hence fairly successful operation. Construction of health facilities, which required a considerable amount of planning and to which stringent building codes were applied, did not fare well. Field research trials hardly materialized, which was a major failure, limiting the technology that could be disseminated to farmers. Classroom training of bookkeepers, requiring only a couple of instructors, went smoothly; field supervision, requiring a large field staff and vehicles, did not. Because of the different "success" rates, the project desIgn's sequenced guidelines for village investment acted as a drag on overall implementation. 3.09 By the second half of Project Year 4, it was evident that the project was in serious trouble. Not only did overall implementation lag well behind targets but the input credit program, a focal point in project design and the direct link to the production objective, was doing poorly; and it began to look as though the region would never be able to absorb the funds which had been set aside for its rural development program. This led to the classification of the KRDP as a problem project in August 1978. Through this year and the following one, the Bank took a number of decisions about how to resolve these problems. 3.10 Three types of policy changes resulted from this process; each altered the character of the project in some way. First was the attempt to lift some of the major constraints to implementation, including increased purchase of vehicles and resolution of procurement problems; second, was the setting up of means to better facilitate financial flows; and third, was the decision to provide funds into some major regional and national infrastructure which had not originally been part of the project. 3.11 Towards the end of 1978, steps were also taken to alleviate the procurement problems the region had been experiencing. A Dar-es-Salaai- based procurement office was established. However, eventually, after some success, confusion arose in the PME over what was being ordered and how much was being spent, and regional authorities reduced use of the Dar-es-Salaam office. 3.12 To facilitate financial flows, the original concept of "project village" was dissolved, and the land availability requirements for village selection were disposed of, on grounds of irrelevancy - many villages with less than eight acres of arable land per household were in viable economic condition, and there was no reason to disqualify them from project assistance. A special programme in coffee rehabilitation was introduced for those villages in the previously disqualified highland areas. Ihese measures worked to dramatically increase the number of villages receiving project assistance.  - 24 - 3.1.3 As another measure to help the regional authorities acceLerate expenditure an agreement was reached between the Government and the Bank on the use of a "one-linevote". This would allow the KRA to make adjustments in the specific items to be funded under the project's annual budget once a general agreement had been made on the nature of expenditures in that year. 3.14 The most significant change in the project's character resulted from the decision to fund some major regional and national infrastructural installations. For the region, the most notable of these were the assortment of heavy road equipment and a number of large buildings including the extension of the regional office building (Regional Block). 3.15 InstitutionalprobLems of the project-financedAgricultural Trials and Training Center 7/ (AlTC) led the KRA to negotiate its conversion into a full Ministry of AgricultureTraining Institute (MATI) in 1978. Following these discussions, a sharp increase in project funds went to expand what would now be a larger, national institution. The school's management tried to acquire as much physical plant and educational equipment as possible while project funding was available. It remains uncertain whether the MATI will ever be completed sufficientlyto function as a proper MATI and whether the Kigoma Region will get anything special out of it (they have been promised a special allocation of places). 3.16 The effect of these large investments on the project'soverall characterwas by no means confined to financial considerations. As highly visible, centralised investments which had a high priority for the officials who would be using them, they became strong competitorswith village infrastructuralinvestments for scarce regional resources. Thus when the project experienced a severe shortage of cement in 1980/81,more than half of the total amount went to the regional administration buildings and the ATTC, while the need for cement to finish project buildings in the 8 village was far from satisfied. / C. Special Issues in Implementation 3.17 Input Supply. With the dissolution of the KCU, inputs were distributed to the village4 directly by TRDB, using NMC and TCA as delivery agents. The supply of fertilizers got off to a flying start, over-shooting the target for the first year by over 13 times, and then ground to a 9 complete halt after the third year because of non-repayment of loans. / rhe abrupt halt was directly related to the flying start. The failure of this component of the project may be ascribed to low demand in conjunction 0 with badly organized and compulsory supply.' / 3.18 The Low demand for fertilizers during the early project years had its root partly in the prevailing crop price relations and partly in the technology scarcity and the particular stage of development of the recipient farming system. 7/ Further documentationof individualimpacts found in Project Evaluation Report, Section 5.5.2. 8/ Project Evaluat'on Report, Appendix H. 9/ Table 4.11 of Project Evaluation Report. 10/ This is treated at greater length in Loft and Oldewelt, "Developments in Peasant Agriculture in Kigoma Region During the Post Villagisation Years", op. cit.  - 25 - 3.19 During preparationit was assumed,erroneouslyas it turned out, that the land preparationpeak was absolute and constituteda constraint for the expansion of acreage; the project was thereforedesigned to foster productionincreasesexclusivelythrough an improvementof yields, and not through an expansion of area. Of the four crops which the intensification effort was to be centered on, two, maize and beans, togetheralready formed the pivotal enterpriseof the doiznant farming system, and two, cotton and groundnuts,while not alien to the region, were not central in the system and were to be given greatly increased importanceunder the project. The project's inabilityto procure groundnut seeds 1 l/ left only cotton, to have receivednearly 80% of the total resourcesunder the seasonal input program, and maize and maize/beans mixtures, to receive the remainder. 3.20 Generally speaking, cotton is a crop which is quite suited to an intensification strategy,particularly wkaerepests are a problem,-- either it is worth growing with substantiallabor and capital inputs, or it is not worth growing at all. The failure of the input program for cotton is thus intimatelylinked with the failure of the crop itself to win a preferrred place in the productionpatteTn. Initial assumptions both underestimated the labor inputs required for satisfactorycotton yields and overestimated the ecological scope for cotton cultivation in the region.12, 3.21 Price changes for the 1974/75season then invalidatedthe crop's commercialcompetitiveness even before the project became effective. Given the official prices (usuallyan underestimation of the value of food crops) and the AppraisalReport'sproduction coefficients for cotton, maize, and beans, the returns to cotton have consistently been less than the return to the competingmaize-beansmixture throughoutthe entire eight years of the project 13/, a fact not highlighted in the supervisionreports. Beans, because of the high price and input complementarity with maize proved particularlyprofitable. 3.22 The mistake in project design of building the intensification strategyprimarilyaround cotton, could have been of less concern if flexiblyhandled, as the appraisalmission no doubt hoped, but it was exacerbatedby the efforts of the Governmentto use compulsion(ratherthan price incentives)to elicit cotton production. 3.23 The lack of denand for inputs for maize-beansdid not stem from the undesirability of the crops themselves(which grow comparatively well under Kigoma's relativelyfertile soil and stable rainfaLlregime), but rather from the inappropriateness of inputs to the resourcesand priorities of Kigoma'smaize-beansfarmers, and the fact that they did have surplus labour, once initial villagisation tasks were completed,to expand cultivatedarea on previouslyuncultivated and therefore productiveland. 3.24 It was against the backgroundof this initial low level of potential fetilizerdemand that large quantities of project-financed fertilizers were transported without farmer's requests to the first lot of 21 unsuspectingproject villages in 1975/76. Accordingto the project 11/ Project EvaluationReport, Section 4.1.3.1. 12/ Reasons elaboratedin ProjectedEvaluationReport, Section 6.2.1. 13/ Project EvaluationReport, Appendix U.  - 26 - 4 project design1 / the villages should have decided on the amount of inputs matters, they would apply for. In fact this did not happen. To complicate it was not effectivelycomunicated to the villagesthat the inputswere on credit. 3.25 Satisfactory loan recoverybecamenearly impossible.TRDB's attemptsto sanction non-payment of loans by withholding furtherseasonal input credit from debtor villages were overruled by the prevailing Governmentthinking which favoredan increasein cotton production at any cost. Projectinvolvement in the programwas finallyhalted,to the considerable consternation of the regional authorities, in projectyear five (1978/79). The repayment controversy has hauntedthe programever since,and a solution was only agreedupon betweenTCA and TRDB In late 1981. Moreoverin the processof reconciliation, it appearsthat the villagesmay have been short-changed by about Tsh 1 millionwhich they have for.15/ alreadypaid, but are stillheld responsible 3.26 The input credit supplyprogramseemsnot only to have failed to have a positive effect on production,it has had some seriousnegative consequences.First,thosepeople who did laterwant fertilizers or insecticide, or maize millsl / were preventedfrom gettingthem. Second, the bitter associations whicihmany farmersnow have with fertiliseruse may well colour theirwillingness to experimentwith it again. 3.27 Monitoring and Evaluation. Monitoring and Evaluation, under the initiation and supervision of the PPT, was supposedto be an important input,both into the -Kigoma Projectand other rural development projects yet to be.formulated.In practice, PPT involvementwas minimal,and on the whole the monitoring and evaluation _ystemwhich evolved made little contribution to the Kigoma Project. 3.28 Monitoring was left to the projectcoordinator of the PHE to sort out. At the insistence of the Bank, he instituteda systemof quarterly reporting on projectactivities, compiledfrom progressreports submitted by the regionalfunctional managersand the DDDs. Although some improvements were registered in the qualityof thesereportsover the life of the project,the systemitselfnever reallyworked well. 3.29 The deficienciesin the monitoringsystemare attributable to severalfactors. Most significant was the absenceof any traditionof monitoringin the Tanzanianbureaucracy. Transportation and communication problemsexacerbated the situation, since the regionaland district- officialswould in many cases be stuck in the townswhile their work crews in the villageshad no system for sendingback progressreports. 3.30 This K and E problembegan to surfacevery early in the project,and in 1977the Bank asked that the post of evaluation officer be created. The post has remaineddifficult to fill with personswith the high level trainingand years of experience which the job really required. 14/ AppraisalReport,para 5.07. 15/ Detailsin ProjectEvaluationReport,Section4.1.3.1,and AppendixE. for maize mill loans and an 16/ There were at least 17 officialrejections untold numberof other villageswere discouragedfrom even applying.  - 27 - 3.31 In retrospect, what would have been far more useful to the project itself, and to the region's longer term development or administrative capacity, would have been to hire a management consultant/advisor under the project, who c-uld have helped institutionalize monitoring capacity within the various regional and district offices. The PPT's principal mandate to design new projects tended to preclude the level of attention needed for Kigoma's M & E. 3.32 Two surveys were conducted in the course of the project, both with initial consultation from PPT. In mid-1976, a single visit baseline survey aimed to measure production parameters at the household level, and covered 30 households in each of 10 villages. This survey was contracted out to the Economic Research Bureau at the University of Dar-es-Salaam in an attempt to involve a local institution. The data from the baseline I7 were used to make comparisons over time with data collected in a continuous 16-month farm management survey 18/ covering the tail of the 1976/77 and the whole of the 1977/78 agricultural seasons for 132 households in 9 villages taken from the earlier sanple. 3.33 The lack of demand for answers to specific questions among the regional authorities, symptomatic of the more general problems noted above, was a main hindrance to supplying any feedback from the survey during the course of the project. 3.34 The November 1978 "Mid-term Evaluation" exercise referred to earlier, called in response to the serious implementation difficulties the project faced, sparked a major generation of data, experiences and perceptions and prompted the significant changes in implementing strategy noted above (Ch. III, B). A final evaluation exercise by ERB, conducted in late 1981, issued a report in early 1982 which is drawn on heavily throughout this PCR. 3.35 Project Preparation Team. Activities of the Project Preparation Team (PPT) were funded under the project. Tne objective of the PPT was to assist in the preparation of two or more regional development projects elsewhere in Tanzania and to support Monitoring and Evaluation in Kigoma. It was to comprise a multi-disciplinary staff of expatriates and national counterparts. Formal institutional relationships of this group remained loosely defined. According to an April 1979 supervision report, the team's "relationship has been one of an ad hoc team (identified loosely with the Prime Minister's Office (PKO) and, to a large extent, with the Bank) on secondment to the Regional Planning Office of a particular region", i.e. the Mwanza/Shinyanga and Mara Regions. In general, with the exception of M & E assistance in Kigoma this assistance accomplished its objectives. The Mwanza/Shinyanga Project Preparation Report was submitted and was subsequently approved for IDA financing, and in 1979 the preparation of the Mara Project was completed and appraised (however, it has not been considered for IDA funding). In 1980 preparation of a project 17/ The report issued from the survey was "An Evaluation of the Kigona Rural Development Project, part One", by R. Mabele and M. Schultes, Economic Research Bureau, University of Dar-es-Salaan, 1977. 18/ "Developments in Peasant Agriculture in Kigoma Region During the Post Villagisation Years" by Loft and Oldewelt, op. cit.  - 28 - for Kagera Region (West Lake Region) was completed but was never appraised. The Government in 1979 requested continued Bank financing for the extension of PPT work beyond the measures financed under the Kigoma Project for three further projects. Amotng the positive achievements of the PPT are greater efforts in data collection (although analysis fell 19 seriously behind) /, a criticil examination of strategies under the Kigoma Project, and adjusted approaches under the Mwanza/Shinyanga and Mara Project proposals. 3.36 The PPT's work was, however, also plagued by a series of problems: a. institutional ties remained unclear; the PPT responded to the PMO, but was at the same time seen by Government as an outside - i.e. Bank - team; b. the team's non-acceptance by the region as a Tanzanian group led to its comparatively stronger identificationwith the Bank, and an increasingly distanced, critical attitude towards Tanzanian developments. Its acceptanee by PHO fluctuated widely partly depending on personalities;at times it received little attention or support; at other times it was asked to write important "think pieces" for PMO on national rural development issues; c. the PPT's staffing was never at full strength for any .period of time. The average manning of the team was 60% of established strength, with the maximum level of staffing being 70% (for a 10-month period); and d. the team's budgetary dependence on the Kigoma Regional Development Director (RDD) was unsatisfactory since communicationswere difficult and the team's budget was perceived as being non-incrementalto the Kigoma region's budget ceiling. A separate channel of funds through PMO would have been a better arrangement. 3.37 Aecounts and Auditing. At project completion no audited project accounts had been produced. However, as of November 1981, the overall accounts of the Regional Administrationup to FY79/80, which include, but do not easily identify, the project expenditures, had been audited by the Auditor General's office (in March 1983, full accounts for the period were forwarded to the Auditor General for auditing). 19/ A comprehensive baseline survey covering over 850 households in 24 villages in Mwanza and Shinyanga regions was carried out, together with a substantial Farm Management Survey series, yet to be fully analyzed, and a number of issue papers were prepared regarding the economic implications of village size and layout, sustainable land capacities, and mechanisms for participatorydevelopment.  - 29 - 3.38 Disbursementapplicationsto the Bank, for expendituresrelating to the project,were lodged separatelyby TRDB, the Prime Minister s Office (PMO) and the Kigoma Project ManagementExecutive (PME). There was seldom close coordinationor communicationbetween these agencies on these matters. In addition, the Kigoma office did not always receive information regarding direct payments by the Bank. As a result, the PME was only able to effectivelykeep track of project expendituresincurreddirectly by the Kigoma RegionalAdministration(KRA). 3.39 The accounting capabilityof the PME lay primarily in the expatriateFinancial Controller, who was provided under the project. However, as noted above, although the regional accounts prepared by the regioral accounting office were reasonablyup to date, there were, at the date of project completion, no audited accounts for the project itself (the Financial Controllerwas requested to stay on to complete this work and full accountshave now been forwarded to the Auditor General). Even as late as 1981 when the evaluation report was being prepared, the report noted that, "surprisingly, the only type of data which was so inaccessible that it had to be by and large excluded from the analysis was the financial accounts. Short of getting involved in auditing-likework, the team was only able to make rough cost estimations both for overall expenditureand for component-by-component breakdowns". IV. IMPACT 4.01 Chart 2 presents an overview of the intendedand unintended effects of project components directed at the village level, togetherwith their proportionof total project costs. 2 0 / According to project design, the composite impact was to have been thtedoublingof incomes for the target population. It was expected at appraisal that this could be measured as the difference in household incomes before and after the project. There are both practical and theoreticallimitationsto apply quantitatively either this "before/after" approach or that subsequently initiated by Bank supervisionmissions to compare project and non-project villages. Thus a more inferentialapproach to assessing composite impact must be followedhere. This is done first by examining the shape and direction of general agricultural development in the region, and how these trends have affectedhousehold level income and then judging to what extent and in what areas the project influenced this outcome. A. General Trends in Agricultural Production 4.02 The second half of the seventieswitnessed an apparently substantialrise in agriculturalproductionand rural incomes in Kigoma. Two detailed project surveys in 1975/76 and 1977/78and a less detailed follow-up in 1981 give strong evidence to support this (see also Table 2b for official procurementtrends). This development can probably be ascribed to: (i) the rapid incorporationof the population into the nationalmarketing network; (ii) the gradual release of household labor from village infrastructure and domestic and agriculturalrehabilitationto cultivationon largely untapped soils 21/; and (iii) the proximity to the 20/ Further documentationof individual impacts found in Project EvaluationReport, Section 5.5.2. 21/ As soils get overworked, this is likely to be reversed.  - 30 - border and the decreasing availability of consumer goods giving a high incentive to produce for trade. National pricing policy had a decisive influence on the form and content of this development. Beans began to be officially purchased by NKC and in the wake of concerns about food self-sufficiency official prices went to very high levels (cotton prices did not keep up with food crops). 4.03 This same trend toward increased commercialization was exhibited in a number of areas in Tanzania (e.g. Rukwa, Ruvuma, and parts of the central plateau) which had previously been neglected by a marketing link to the national economy. It was in direct contrast to the trend exhibited in the historically more commercial areas (Kilimanjaro, Mwanza, Tabora, Arushs) for whom the advent of the parastatal marketing network meant a decline in the quality of services over what were relatively well run cooperative marketing networks, and where land was generally in such short supply that previously uncultivated soils were not tapped. 4.04 The trend toward increased market orientation of &gricultural production in Kigoma was initiated in 1974/75 by the introduction of cotton growing by administrative fiat in 143 of the region's viLlages. Cotton production peaked only two years later, however, and output seems to have stabilized at a somewhat lower level. Even though cotton has been gradually confined to only 97 villages, in the majority of cases its cultivation is still sustainable only by coercion. The market-orlentation of food crop production took off one year later. Hampered by the cotton obligation of one acre per household, but spurred on by good official bear and other food crop prices, the officially marketed 3uLpat of food crops kept ircreasing all through the seventies (see Table 2b). Official bean procurement has now plummeted due to falling real prlces, so unofficial marketing has increased. Trade (mainly in the parallel market) to Burundi has expanded dramatically with a major attraction being the availability of consumer goods which have been in very short supply locally. 4.05 However, official procurement peaked at the end of the decade and in late 1981, NMC estimated that its procurement of beans in 1981/82 would only be about one-eighth of what it had been in its peak only two years previously. Neither NMC, the village governments, or the villagers themselves, suggested that this dramatic decline was, to more than a marginal degree, a reflection of a drop in production. Rather, it was first and foremost attributed to NMC's loss of its previously major share of the market because of stagnant official prices, and secondly to the parastatal's severe liquidity problems in the preceding marketing season. 4.06 Although the decline in the volume of crops officially marketed at this time was not a true reflection of a similar trend in overall production, several indicators suggest that production had, in fact, probably reached its limits. Two barriers have been making themselves felt during the last years of the project. The first is found in the increasing distances farmers must walk to put new land under the hoe, and the second is found in the insufficient supply of consumer goods, especially the more durable ones, which has far from kept pace with the population's increasing purchasing power, and in some parts of the region may have registered an absolute decline in availability.  - 31 - B. Indicators of Living Standards at the Household Level 4.07 To relate the previously discussed production trends to the actual standard of living of the rural population, one must seek indicators of living standards at the household level. Two very detailed project 2 surveys l/ covering identical households fortunate;:, span the period where the eariy and most rapid increases in market orientation took place, namely from 1975/76 to 1977/78 (PY2 and PY4). On the basis of these surveys, It is estimated that the value of total agricultural production per capita in the main farming system of the reglon (the maize-beans villages) rose by 64% from Shs. 209 to Shs. 342 per capita as measured in 1978/79 official prices. The total cash incomes from agricultural and non-agricultural activities for this sample rose in the same period by 95Z, from She. 128 to Shs. 250 per person. If i~flation is taken into account, cash incomes still rose by 58% or the equivalent of the purchasing power of She. 92 per person in 1978. 4.08 Neither funds nor time permitted a detailed survey at the end oiO the project to establish the level of income for the same sample in 1981. To gauge the direction in which incomes have moved since the 1977/78 season, the evaluation team asked the households in the project villages 23 which were visited / to compare their present economic situation with their situation in the eaclier year. Sixty-three percent felt that their situation had improved while 10% felt no difference and 27% registered a deterioration. This indicated that Incomes have still generally been rising, but at an abated-rate. 4.09 Perhaps the most dramatic indicator is the improvement in housing 24 standards. / The early seventies witnessed a complete replacement of the traditional round grass huts with rectangular mud and pole houses. Lately, the use of bricks, sun-dried or burnt, has been increasing rapidly. There appears to have been a marked.increaseof durable consumer goods, although there are long lists of unfilled orders in the better-off villages because supply has not kept pace with demand: the cattle population has also been rapidly recovering from the losses endured during the villagisation process. 4.10 These apparent improvements in rural living standards have basically come about through acreage expansion. Withjut entailing any qualitative changes in the methods of production, the expansion was simply based on the eventual utilization for agricultural purposes of that surplus labour which had existed prior to villagization. Some increased differentiationamtong households must have taken place because of the differences In their work capacities. Those households whose composition even prior to villagization only allowed them to meet their minimum subsistence needs would not have experienced any or much increase in production and income as a result of the post-villagizationmarketing opportunities in the region. 22/ The baseline and farm management surveys. 23/ Of the 14 project vlllages (plus one non-project village) visited, six were overlapping with the farm managemen. survey villages. 131 households were interviewed, 114 from project villages. 24/ Project Evaluation Report, Section 5.5.3 and Table 5.2.  - 32 - C. The Project's Contribution to Rural Development in the Region 4.11 How or to what extent may the effects of the individual project components listed in Chart 2 have contributed to the type of development which has taken place in the region? 4.12 Preconditions. The project evaluation experience suggests that those project components which achieved their intended impact to any significant degree, exhibited three pre-conditions. First, the component was easily implementable with the given resource, i.e., technically and organizationally relatively simple. Second, it was in confluence with current Government priorities. Third, it corresponded to the needs of the recipient population. Where any part of this constellation was absent, a component's impact was either negligible or considerably less than intended. 4.13 Primary Impact. In general, whenever otieof the above conditions was present, at least one of the others would be absent. The single exception to this observation was in the case of expanding the official marketing system's crop procurement capability. 4,14 The comparative simplicity^of installing the elements of the marketing system (transport, roads, storage and basic accountancy skills) in conjunction with the national policy at the time of creating and expanding crop authorities, was not inconsistent with the immediate aspirations and productive capacities of the population. This ensured that at Least these project components succeeded in their designated purpose. However, only in the case of the bookkeeper training was the project the main actor. With regards to the other elements, the project usually provided only a peripheral contribution. The contribution to feeder road development was, on the face of it substantial, 619 km upgraded compared with an appraisal target of 389 km. However, the evaluation report sutggeststhat the data on this is of doubtful reliability and the extent and quality of the work was quite limited, with little or no drainage work or gravelling. Villagers interviewed did not apparently have strong impressions about the impact which might have been felt from feeder road improvement. By 1978, it was clear that serious road deterioration was setting in and approval was given for the purchase of equipment for a more capital-intensive operation. Most of it was received in 1980 and its impact has not been evaluated. 4.15 The intensification of land use was supposed to have come about through the interlocking of a multitude of components (Chart 1) but hardly any one of them achieved their intended efforts (Chart 2). The educational component had a relatively fine implementation record and the villagers' enthusiasm for this component is witnessed by the initially generous contribution of self-help labor for classrooms. The reason why the educational component cannot be said to have had any immediate impact on production is that the originally expressed policy priorities of emphasizing practical training in carpentry, farming, masonry, and smithery were superseded by the accelerated pace of universal primary education -- leaving the regional authorities with no spare energy to devote to the technical aspects of education. However, it would be reasonable to expect that longer-term, more intangible production effects may be derived  - 33 - indirectlyfrom the enhancedability of literates to acquire knowledge and skills which may improve their living standards. 4.16 The creation of area-specificcultivationrecommendations was never achieved because the task was infinitelymore complicatedthan the simple fine-tuningwhich was originallyenvisioned. Thus, the extension network never really had anything new and appropriateto offer. In particular,and of fundamentalimportance, the key crop cotton, on which a large part of the credit was to be used to achieve both increasedhectarage and yields, was largely a failure. As a result, the credit, which was to be the means to increaseproductionwas cut off, thus severingearly on the linkage through which the major part of the productivity increasesof the project were to be obtained. 4.17 While the measures to intensify land use failed to raise production,it .as the release of more labor which enabled the villagers to take advantage of the newly establishedmarketing system. But the labor in questionwas that which had been tied up in the villagization process or due to poorer incentives,had, prior to villagization, been allocated to leisure. The project componentswhich were to alleviatemore permanent labor constraintsthrough the provision of water, maize mills and health services, could have only contributedvery marginallyto the expansionof output. The mills reached only a handful of villages due to the problems with the credit system arising partly from the disastrousfertilizer scheme. Water supply and health serviceslargely failed in their productivepurposes, not because they did not have high priorityboth nationallyand by the villagers themselves,but because of seriousproblems of managing implementation. 4.18 In short, a review of the incidencematrix in Chart 2 highlights how rare it has been that the project has contributedto the recorded trend of rising incomes. 4.19 Other Impacts. In contrast, it appears that the project'smajor contributions may well have been unintentional or incidental,but neverthelesssigaificant. Certain of these impacts,notably from bush clearing,have actuallyhelped augment (over the short run) income levels, by subsidizingan e-tensification of land use rather than the intensification which other componentshad aimed at. Yet the most conspicuouscontributions of the project may have been in the sphere of nation-building, and the strengthening of village and regional administrative structures. In particularthe educationalcomponent figures prominentlyin this respect assisting,for example, in increasingnational literacyin a common language. The incidentalstrengtheningof the role of village governmentsand the population's acquisitionof vested interests in their villages (e.g. schools,dispensaries, water supplies, etc.) are factorswhich may eventuallyhave a significantimpact on the shape of future developments. However, the impact of such things on productivity is unclear at this stage, and in any case, much will depend on the extent to which over the next 20 years, the new-foundvillage administrative capacity and power can be used by the villagers to articulatetheir needs or the extent to which it becomes merely a channel for bureaucratic directives.  - 34 - 4.20 The region may also be smarting for some time to come from the negative impacts of the project, in particular the legacy of the input scheme through which 34 villages were saddled with a long-term debt to TRDB, and the cynicism which a litter of uncompleted structures have left in the minds of the involved villagers. 4.21 Lastly, the project has also left behind to the regional authorities significantly improved physical assets in terms of mobility (vehicle stock) and office and storage space. Unexpectedly, the project even transcended its regional scope by augmenting the Ministry of Agriculture's stock of national training institutes, although, as mentioned in para 3.15, whether the MATI will be completed and what its impact will be on the region remains unclear. V. ECONOMIC ANALYSIS 5.01 Production in the region increased substantially over the period. While the project had such increases as one of its main objectives, the evidence (Section C) suggests that very little, if any, of this growth can be directly attributed to the project. 5.02 A realistic reassessment of the rate of return on the project is not possible mainly because production data are inadequate and causality linkages between project investment and production are imposible to make. Nevertheless, in order to give some feel for broad magnitudes, a crude economic analysis largely based on assumptions, has been undertaken (Table 3). This is not intended to show what did happen under the project (the data is far too weak for that) but it does show what the rate of return might have been under certain assumptions, the plausability of which can then be assessed. In this analysis, Project costs were shadow priced with FE valued at double the official rate. Labour was costed at 50% of the current minimum wage (in 1977/78 prices) but estimates of incremental quantities of labour were difficult to derive. Total regional production increases were based on the 64% increase which surveys showed occurred between 1975/76 and 1977/78, and on an assumption of 32% increase (i.e. 50% of 64%) for the two-year period 1977/78 to 1979/80 and 16% (i.e. 50% of 32%) for the final two-year project period with no increase thereafter. The value of production at official market prices was adjusted upwards by 50% to a "guesstimated" shadow price to allow for higher unofficial prices on untraded commodities and to allow, crudely, for import parity prices for internationally-traded commodities. 5.03 The analysis (Table 3) showed that if project costs excluded education and health costs on the grounds that the benefits to those were not measurable and also excluded the road and coffee component costs on the grounds that the benefits to these may come at a later date, then attributing 20% of the total regional production increment to the project gives an economic rate of return of 5%. 5.04 If, on the other hand, only 10% of the total increment were attributable to the project then a negative economic rate of return of minus 4% results. 5.05 This analysis is highly speculative. However, attributing 2D% of the regional incremental production to the project is almost certainly  - 35 - extremely generous to the project. There is therefore very little chance that this project will have a positive rate of return. VI. BANK PERFORMANCE A. The Preparation and Appraisal Process 6.01 The project required a staff input of about 132 man weeks through appraisal,2 5 / a comparatively high figure for agricultural projects.2 6 / Bank performance during these early stages raises a number of questions. Of the various design flaws which were to haunt the project, some may have been a result of too little staff time in the region. Important things were missed about the region's farming systems -- underestimation of the amount of land people would cultivate and of the adjustment time required after resettlement in the new intermediate and miombo areas; and relative neglect of the agricultural potential of the highlands and lake shore areas. Also, judging from descriptions of the resettlement program, the teams apparently were not able to determine, or were not privy to information which might have highlighted, that villagisation was being conducted with a considerable element of coercion. 6.02 On the ,)re positive side, both RMEA's comments accompanying the project proposn, Ch. II, B, ii) and the issues paper of December 11, 1973, reveal that ,mber of problems were recognized -- e.g., the regional authori: o go ahead with villagisation and larger village sizes, .. and the likely manpower constraints in the years to come. These issues,however were progressively downgraded. The project which was finally appraised was far more ambitious and riskier to implement than the first, cautious draft prepared in early 1973 (Chapter II, B, (ii)). 6.03 The Bank's willingness to push ahead with the project may have been, in part, a result of the belief that the Bank could technically influence and constructively help by approving the funding more quickly. However, in retrospect, this underestimated the political motivation for the project. Yet, in other respects, there may have been no reason to doubt the Government's ability to meet its promises of priority attention to an individual project's needs. Bank staff were told that Kigoma would get special attention in manpower and materials supply. As it turned out, Kigoma RIDEP was but the first of many such regional projects (partly due to Bank encouragement) and this reduced the possibilities of priority Governm"entattention to any one of them. Moreover, the national economic decline over the project period which was to seriously influence the extent to which the project coild get support and the environment in which it had to be executed, was not predicted by most observers at the time of appraisal. 25/ See Basic Data Sheet, Mission Data. 26/ Average for preparation between 1975 and 1978 was 36 weeks, for appraisal 59 weeks in East Africa Region. Earlier figures not available in any reliable form. The preparation figures may be underestimates.  - 36 - 6.04 Some of the compromises suggest, in retrospect, that the Bank nay have been in too much of a hurry to get this project approved. The Bank saw Tanzania as an important and eager partner in efforts to expand lending in line with the new poverty-orientedfocus. 6.05 While the knowledge of KCaJ'sdissolution did cause great concern in the Bank, to the extent of serious discussion of possible project cancellation, it was finally decided to proceed on the grounds of indications of measures to improve the efficiency of the parastataLs and promises to avoid duplication of such things as transport. Purthermore, it was not entirely clear at the time that the dissolution was permanent as it subsequently became. However, it would probably have been not only sensible, but legitimate, to suspend considerationof effectivenessuntil sound Institutional readjustmentswere actually established in the project area, perhaps even involving reappraisal. Following the formal govert.ment announcement (less than 10 days after effectiveness) the Bank accepted the regional authorities suggestions for the institutional innovation of the PME suggesting, even, that the PME might be a superior arrangement. Although clearly this judgement is easier with hindsight, the PME was never in a position to coordinate the necessarily Integrated functions of marketing, credit, and input distribution--anact which a refurbished KCU, with the right kind of top management, might have been able to pull off. In retrospect, the proper action at this critical stage would have been to reappraise, and in the absence of a reinstatement of the KCU with increased controls and strong technical assistance, to design a much reduced pilot project with more modest'attempts at integration, and/or to have placed less reliance on centralised marketing channels, however, such proposals would not have been in accord with Government's strategy at the time. B. Supervision 6.06 Staff input in supervision consumed about 395 man week,2 7 1 well above the Bank average. While part of this resulted from the near doubling of the project cycle, the KRDP was always intended to have a higher than usual supervision input because of its experimental nature. While supervision doubtless made an important contribution in helping implementationover the course of the project, there were a number of shortcomings in the system which prevented more positive and tinely contribution: a. even with the higher than average supervision Inputs, the missions seem not to have been in the Region long enough to develop atndmaintain a real sense of what was happening. It apparently took several years before the Bank fully realized that cotton growing was compulsory, for example; b. this was exacerbated by some discontinuity in supervision staff; c. when the missions did flag a serious problem, there seemed to be few effective means to assure that corrective action was taken. The second supervision mission correctly noted that 27/ Basic Data Sheet, Mission Data.  - 37 - the Region was not exercising sufficient caution in Initiating the fertilizer program. Yet the most the Bank could do was sent a letter from faraway Washington stressing to the Region that it should slow down. 6.07 Greater Bank insistence in trying to correct problem areas would in retrospect, probably have been difficult in that the "Last resort" mechanisms of suspending or sharply curtailing disbursements may not have been easily reconciled with the Bank desire to respond to Government's stated intention to find ways to support production and income improvements among the rural poor, and the Bank recognition from the outset that the task would not be without problems or risks and would require perseverence. C. Mid-term Review 6.08 This review was undoubtedly a valuable exercise in gathering and processing the project experience over the first four years. In some respects, subsequent changes made in the project reflect lessons learned from earlier mistakes or oversights--the input credit program was finally stopped, the land availability criterion (which had proven irrelevant under the circumstances) was dropped, a pilot coffee program was introduced for the highlands, and the implementation constraints posed by the complete package approach were lifted, allowing more villages to benefit from components which could be implemented more successfully (especially bookkeeper training and education). 6.09 The mid-term review did not really produce a new strategy for rural development. As the review of the later years of implementation seems to reveal, the ci,angeswere rather piecemeal and, in some important respects, significantly pulled the project away from its original rural beneficiaries. VII. CONCLUSAIOT: LESSONS FOR RURAL DEVELOPMENT 7.01 The Bank's experience in Kigoma was partly intended to provide important lessons for rural development project design and it undoubtedly did. The lessons can be summarized as follows: 1. Ensure Project-Government-Target Group Complementarity 7.02 The preparation process, the implementation experience, and the review of registered impacts all demonstrate that an outside agency cannot engineer developments which do not reasonably coincide with the interests of both the recipient government and the intended target group. Over this period, Government and target group interests did not always seem to coincide (e.g. cotton growing), so the possibility of a successful project was seriously reduced. lany of the other lessons are subsidiary to, and hence qualified by, this overriding problem. 7.03 In the case of village planning for example, the project was not really consistent with the Government's own vilLagisation policy. In a number of subsequent differences between official strategy and the project objectives, the problems seem to have sprung at least partly from the Bank's misperception of the administrative structure. Preparation teams, perhaps understandably, thought that the recent decentralization exercise  - 38 - policy decisions. This was indeed the stated intention. In fact, decentralization worked in many respects in the opposite manner, as a direct channel to the regions of increasingly centralized official policy interventions. Pricing, marketing, and credit policy all came to be determined centrally, and often quite unrelated to the project's objectives. The unattractive official price for c- ton, coupled with the policy to enforce its cultivation, not only made iL impossible for the project's income objective to be implemented flexibly but made it difficult for the Bank to stop the disastrous input credit program without affecting relations with the Government. 7.04 There were a number of early hints of the difference in objectives of the Government and the target group (e.g. the extent to which villagisation was forced, and the Government determination of crop selection), but these tended to be downplayed perhaps due to the Bank's eagerness to believe that the project would nonetheless work and to interpret the structures and events in the country in a very positive light. This might argue for a larger Bank sectoral presence in the country, on a permanent basis, to follow policy developments wore closely, to gauge the environment's suitability for individuaL projects aridto contribute to supervision. Alternatively, it night simply argue equally strongly for the Bank to be very careful not to allow its wishes or hopes to sway its interpretation of events as noted by missions. 2. React More Positively to Policy and Institutional Environment probLems During Early Supervisions 7.05 It is arguable that many of the emerging "national" problems impacting on the project became apparent after appraisal. To the extent that this is true, it would appear that greater attention should have been given to some of these issues early in the supervision process, particularly the problem of the dissolution of the KCU, and the rapidly changing crop price relationships. It seems that a number of fuxndaxmental issues warranted more positive reappraisal during the early supervision period. 3. Seek Responsive Project Administration 7.06 As noted above, there was not a clear or continued coincidence of objectives, so defining a "responsive" project administration would not be an easy task. The Bank's interest seemed to focus on finding a solid production base, but even here a real gamble lay in the productive investment program. 7.07 Because of its central position vis-a-vis the interrelated aspects of marketing, input credit and distribution and productive infrastructure and because of its relative independence, the KCUJ (sufficiently bolstered by technical assistance) might have been able to respond to farmers needs and to react flexibly, at least for the production aspects of the project. Its very serious weaknesses at the time of appraisal may have made this difficult. As it was disbanded, however, the option disappeared entirely.  - 39 - 7.08 Instead, a makeshift arrangement was made for a non-executive office within the regional administration to keep an eye on the project activities - production and other - of a multitude of separate offices in the region. The prod ction-related activities of the KCU were parceled out to no less than five institutions, Utndersuch circumstances, and with weak institutional monitoring there was almost no chance that the PMX could have operated effectively, in the first instance, to try to lead the parties involved in developing and naintaining a clear consensus on the project or, in the second, to have adjusted and steered the project according to information gained as implerentation proceeded. 7.09 One of the lessons which springs from this experience is that there seems to be an inverse relationship needed i.nrural development projects between the precision of design and the implementing agency's ability and power to adjust. If, as in the case of the Kigoma project, the institutional capacity is low, it becomes all the more crucial that clearly agreed objectives and a correct resource assessment lies behind project prescriptionsfor investment. 4. Assign Priorities to the Components in a Large Multi-Sectoral Program 7.10 Despite the fact that agrlcultural development was seen as the foundation of the program, technical agricultural aspects, such as the adaptive research trials, were treated as just one of several components. The effect was to spread efforts and resources too thinly without concentrating on some of the most important areas. The appraisal report should have provided more guidance on this matter, which would have helped to focus the attention of supervision missions. 7.11 However, it must also be acknowledged that even if such priorities were made more explicit in the appraisal report, and supervision missions made the appropriate interventions, the agricultural research component may have still encountered major difficulties, since this component was aimed at developing technical packages for the smallholder farmer, while Government policy favored either communal or l.rge block farm arrangements. It would appear that this conflict, and hence, the risks, were not sufficiently highlighted at the appraisal stage. Had they been, it is not clear what steps, short of not going further with the project, could have been taken to minimize the risk. 5. Ensure Realistic Project Tining and Phasing 7.12 B3etweenthe initial preparation drafts and appraisal, the project scope was expanded and the implementation period compressed into a five-year cycle whose sequencing and spacing made its prospects for effecting the initial linkages among components doubtful. Because this compression made unrealistic demands on the region's absorptive capacity, the cycle was, in the event, doubled anyway. A much longer planned project duration would have been far superior, allowing in the early years, research on agriculture, hydrology investigations, development of an M and E base and perhaps road improvement, after which appropriate second phase investmentscould have gradually been introduced.  - 40 - 6. Carefully Prepare and Know the Technology of Low Cost "Short-Cut" Components 7.13 Imbedded in the project design were a number of short-cut approaches to keeping the project cheap which may have ended up contributing to cost ineffectiveness: a. Water. Except in cases where the hydrogeological conditions permit low cost, shallow well water systems, village water supply systems are not inherently low cost, and they stand much better chances of being designed and constructed for durability if there are qualified, experienced engineers and technicians employed to do the work. The water component was ultimately expensive and the unhappy record appears to be a direct result of the initial cost-cutting design, which assumed that shallow wells were possible in Kigoma without doing sufficient testing. b. Extension. The short cuts in extension training would have required serious attention to designing an appropriate curriculum. Unfortunately, the training was done haphazardly and with very theoretical English texts being verbally presented in Swahili. Not only were the graduates considered by farmers to be very ill-prepared to give any useful advice, they could not be admitted into the administrat4ve network (to be paid) because of their non-standard education. More detailed design of this and also the technical assistance component was needed. 7. Provide Adequate Technical Assistance 7.14 The general predeliction against a sizeable technical assistance tean to the region was explained in the issues paper with the argument that regional authorities had gained sufficient experience in implementing Operation Kigoma to implement the project. This contrasted with an earlier suggestion to explore the possibility of getting a bilateral donor to provide this high cost item. The final solution, to recruit a small handful of technical assistance staff on a piecemeal basis, probably cost more through low cost-effectiveness that the additional cost of a more substantial core team of staff to help implement the major components.  - 41 - PIRJZCT COSTSAMOF7HAt4C0NG 2//15, 6/ tprtraClt t.tioent 1/ Actual C.ctiaied) Actual a. 1 Coa.nnata taff 8iioO - YiWTh06uSs 00) it AUo pireLoal 1. Vii le4e2tCfrsltructure t) water Supply .12,665 1.774 17,21 2,184 12) IS) Health 1.550 i1t 1.802 :'S 104 t1t) iecatic 13.436 2.161 34.235 4,333 200 Iv) Feeder Roads 98 134 10.I94 1.316 982 V) ?&atc. Clia9rig 842 118 1.165 147 124 S. Total 31,449 4,405 ".847 S.208 186 1. Locen t V1o :) l3Lsgo., Cr 7.706 1.079 4*,700' 199 5 11) = tut coaer Cgedlc 5474 767 1.473 La8 14 S. Total 12,Lo 1.846 6.175 782 42 3. Loaneand Znuity to Reicadl cI peraVis Unioa 5,59 796 - - 4. r,tL9uL1tn aestru~ctie etc. t) 2 ss CancerC enter7 248 3.634 460 185 *1) Agric. trisis, traintng center 87S 123 13,348 1.71l 1t,35 Lit) Vehicle., aroial phoc- araphy & land use planniol agri ul ture, mnagement. 9c0jacc preparatitn 4 *Valuatioo, sp.tutn LP"diLture, etc. 884s 27.931 3.535 2 S. Toctl 11,4.0 1,610 45,113 5,710 355 5. Tactnltcal aseit^actce 5,892 825 8,870 1,123 136 6. CaSt lagenc1*i 26.900 3.767 TTAL PrIcsCT COST 96,109 L3,251 12S.005 15,823 119 rINAMCflG TahdHtIllon USS.itlllion Tah.Htllion USS.Mtllion zDa, ---- Yl-.o 10.0 79.0 10.0 100 huGlO? t1.0 1.5 11.9 1.5 tW coveruoant 13.0 1.8 36.1 4.3 239 WrAt. 9S.0 13.2 123.0 1t.S U.9 i7jith1nua race at appraisal USSI.Teh.7.41 -Z (I) Average *bchango rate over projece period USSLteTh.7.9 (II)Avardle *xchafn; rAt. by year '74-7.14; 175-7.1.1; 76-0.38S '77-8.27; '78a7.69 79.8.25. '80-4.18; '81-8.20; 82-8.20. 1/ IAact flgurea repreasnclng actual e:cpandltuiree have been very dlffieultta nbtaIn b-ccuse thne project account. have stILI noc been fltnoaitd. (A finAl supervision mIssion planned for harch-April 1.982 had to be cancelled because the country was unable to accomodcsc It).In the absence of flInalzed records of expendtture It hss b.een vAceasaty tO mk Ce 4us of whatawva lnforsadtoo use *avliabla and construct !tgur-a of expenditurt vhicht vould porcray ch- actual Poatclon ae cloaely ad possible. The procedures adopted for thlx putrpos ore as follows: t) tn ctountry expdefldtt-te to Juno 81. CxcludirOg Project Preparation and EvaluatIon Unitc costs durtig FY 79/30-SL/82. are based on worktlng sheate of the projecc's PlinnclalControlct, TRODSrecorda And the Project Evaluatlonreport dated 1982 preparedby Che Untversity of DCar-2-3swa4s t1) Project Preparation and FvaluatIonUnit Ceac Ft 79/80-01/82 are based on N.Q. disbursement recoeds. taking disburseenc * 902 of full costs. as tii) A1l direec paymentsby DS are bssed an H.Q. dlabursament records. These are foreigncoats an0 ru.present the fult expeadicure. tv) rI 81/82 dnodndlturc Io based on Che year's budget. ASsuming full expendttura. */ Usobange rates used: I) rndivtduSal direcc paymetcs In corotg4 currenclae ocher chan LtSS- value at date, as per 50.Q. disbureseric docueonta. IL) USSIT.Sf.converstons - (a) For a speciftc year, vervorl rate for that year;, b) for tocals over projectperiod tihe averame race oyer Che pcotoce partod. 3/ HIgher costs due to project takinm lunter than 4ntlclPaced. Amendedproject dJeiln caid general coot lncrsaseee Ii) In to4e caoed *xpndisturee totorred on bulk Crocureaents have been charsed to one coepon-nt. elthou4h the tetras procured hAve actuaLly been used on several components. Poc *sfpl. coat cemenc coset hiwe been chsr4od to the health coaponent. vehicle and other service coccs used in leveral toanonente to recorded aUa.zetct ;ittt) of Sgtional lntrasteuct.sra. etc. In the absence of completed, detatled propJec Accounts.a re4tcrtrbution t tsas eaxpendttures co rcelect coats and .ini eaut aCttally rt tel d co tne Jifterent:ospoesncs laso notc bn poasible. In tecoar ot Vt$S Llucte. BEST AVAIUBLE COPY  COPY BEST AVAlUBLE Table 1. b _ 42 - AJNUAL PROJECT EXPEDZTURS (T.Sh. Million) 2 o£ Anticinated Actual Appraisal Per Year Cumulative Per Year CuulatLve Cumulative 1974/75 14.5 14.5 1.0 L.0 7 1975/76 23.3 37.8 6.5 7.5 20 1976/77 20.3 58.1 19.5 27.0 46 1977/78 18.0 76.1 16.4 43.4 57 1978/79 18.5 94.6 14.1 57.5 61 1979/80 30.7 88.2 93 1980/81 27.7 115.9 123 .981/82 9.1 125.0 132 _/ Prepared on the same basis as the preceeding tabulation of project Costs  - 43 - IANZ Al1IATble hji KIPOHA RURAL neVEtLUPllIET PROJECt Bghodu.L2 of 8teburfs.efti * tWA CaudtC (as Qf JusMI 30. L962) Fiscal Year Mavusulated Uisbiicseaeztsa afd coluYot from Cr. S8n6TA (ISS ,tllton) __________ Apprta1aa ACcUl Etsttmaca 2.975 .? 2t *v SV e 0.3 1976 - I 0.1 0.9 Li S 0.L 1.5 7 III 0.3 2.1 L4 tV 0.6 2.6 23 2977 - I 0.8 3.1 2. t1 1.0 3.6 28 ttt 1.1 . 4.1 27 IV 1.5 4.6 32 1978 - t 2.5 5.L 49 xi 2.8 5.6 50 XSS 3.0 6.1 49 IV 3.1 6.6 £7 199 - S 3.1 7.2 £3 it 3.2 7.8 41 tll 3.7 8.4 44 IV 3.9 9.0 43 1980 - S 4.3 9.3 46 Ut 4.4 9.6 44 ttt 5.1 9.8 52 IV 5.6 1.0.0 56 1981 -1 t6.4 64 Uz 6.5 65 EU 3~~~.0 s IV 9.0 1982 - I 9.1 91 *~t it 9 9.8 9, t KUz tSS 9.9 9,9~~~~~~~~~~~~~~~~9 g tV 10.0 100 It III BEST AVAILABLE COPY  TANZANIA DEVELOPNENT KI.CCIARURAL PROJECT - U. S. OF DiSUREENTS UY CATEGORT BREAXDOVI S 1. IDA Credit, as of June 30, 1962 . .Aountof, Credit Allocated Amunt Disbursed Present Balance 1. CivllWorks (excluding w) and health facilities, etc.) 3,900,000 3,750,312.47 149,687.53 2. Construction and road equlpment and materials 500,000 821,244.64 - 32l,244,64 cost of seasonal 3i Incremental inputs, etc. 600,000 527,665.65 72,354.35 4. Aelral photography. telecom systems, vehicles, etc. 1,300,000 2,739,220.98 -1,439,220.98 5. TROD Sub-Loans 200,000 45,082.45 154,917.55 6. Consultant's Services, etc. . 1,600,000 1,498,632.39 101,367.61 7. Salariesof supportstaff relatedto (6) 500,000 249,729.95 250,270.0S S. Projectgpecialfund 200,000 215,473.05 - 15,473.05 9. GIHof Vehicles 600,000 152,658.42 447,341.58 10. Unallocated 600,000 600,000.00 TOTAL 10,W0,000 10,000,000.00 0 IT. UNCDFGrant,as of June 30, 1982 1. Construction of V/S, etc. 1,050,000 1,309,590.81 - 259,590.81 of healthfacilities, 2. Construction etc. 150,010 160,048.97 - 10,04B.97 3. Unallocated 300,000 300,000.00 1,500,000 19469,639.78 30,360.22 u! COPY AMLBLE  - 45 - Table 2B TANZAN IA XIGOMARURAL DEVELOPMENT PROJECT Key Indicators, End Niovember1981 10/ Achi eve - Revised ments Appraispl Targets (November Targets (March 1980) 1/ 1981 2/ Number of Prulect Villagea 135 135 130 Village Infrastructure Water Supply Systems - Shallow well/borehole systems 55 2 2 (1) - Piped systems 55 24 193/(23) Bush clearing (acres) 53,500 45,654 Education Facilities: Classrooms 378 230 221 (188) - Teachers houses 335 147 106 (79) - Offices 76 49 57 (41) - Workshops 93 24 20 (23) - Toilets 120 30 1 Health Facilities - Dispensaries upgraded - 17 18 (14) - Dispensaries constructed 10 10 5 (4) - Rural Health Centers upgraded 3 (new) 4 (upgraded) 3 Godowna 95 22 22 (21) Maize mills 4 95 (5) Feeder Roads Upgraded (km,) / 389 600 619 Village Sub-loans 5/ 175 - 11 Agriculture Villages eligible for seasonal credit 95 75 55 Annual marketed surplus (tons) - Maize 8,400 6/ - 180 7/ - Beans 4,900 - 1,542 - Cotton 10,700 - 2,366 - Groundnut 700 - - Value of mixed agric. produce purchased by NMC (Tah millions) - - 8.9 9i Training Extension agents trained 100 77 8/ 77 Bookkeepers trained and still working 100 180 161 (248) Village medical helpers trained and still working 50 200 86 1/ Based on achievements to date and additional objectives anticipated to completion of project. 2/ Figures in parenthesis are from the evaluation findings and indicate discrepar.cies between evaluation team's findings from Bank supern4sion findings. 3/ At least half not functioning. Based on reports of Regional Engineer. Upgrading mnerely referg to some work being done on these roads. 5/ For fishing equipment, maize mills, etc. (excluding godowns). Source: TRDB Progress Report to March 1981. Current position is the saAu. 6/ Appraisal estimates for PY5 (1978/79). t/ Actual figures for PY7 (1980/81). * '/ The Agricultural Trials 6 Training Center (ATTC), was officially converted into a Ministry of Agriculture Training Center (MATI) in July 1981. The new two-year training course has commenced with 50 students. 9/ Total value of agricultural produce purchased by NMCin Kigoma region increased from Tah 300,000 in 1974/75 to Teb 22.2 million in 1979/8U. Drop to Tsh 8.9 million in 1980/81 attributed to increased sales through unofficial channels and local consumption. 0/ In addition to achievements indicated on this table the project providedt (i) land u planning - 134 villages demarcated (some with land classification ), 23 assisted with alining black farns; (ii) transport - 95 vehicles provided to region (about 10 times appraisal expectation), including 46 four-wheel drive and 39 lorries; (iii) education - 46 text book and library book sets and 3,676 desks provitded; (iv) coffee - 11 nurseries established, 2,100,000 seedlings produced, 30S estimated survival rate. DreT PnDV AVAIIAR  K=1AhtL I T C affidcb p Profraee (Crp Year _ - IW) -1974175-- -1975)76-- -1976/77- -19778- -1978/79- -1979/80-- -1980/81- Cwp TWO Tab Tas Tih TUo Tab TODS Tab TOMs Tab TMs Ye1 TOM Mb 'OM IO1 '0000 To00 _ 0n 'ODDx _Htdi) 11 5 53 38 60 552 737 626 n 79 558 558 180 270 Bem (Uur4wr) 20 27 23 37 531 1,063 1,189 4,162 2,014 6,984 2,865 10,028 1,542 4,R2 F' Pes ) - - - - - - 442 1,325 489 1,468 919 2,756 1,065 3,196 casava O*xtw ) - - 9 4 135 67 465 279 385 250 23 149 6 4 E (lo rgm) 343 225 418 448 817 877 953 1,143 640 798 568 851 327 572 F. Hillet (ezi) - - - - 12 11 19 38 8 15 13 19 1 2 nWUD ) - - - - 129 116 169 169 75 75 4 4 3 3 (1zeti) 9amni Swd - - - - - - 14 21 - - Now ax (a) - - - - 10 143 - - - - 6 132 - - cwea= (a( ) - - - - 10 15 2 6 3 10 1 4 2 7 m(Uf.ta) - - - - 1 2 - - - - 7 2 - - _ramt (ya=*3) - - - - - - - - - - - - - Ootc (PabE) - - 1,102 2,036 2,905 5,546 1,991 4,224 2,409 5,444 2,4n 7,172 2,366 7,121 TOWValue of Proht 1(a*ted Officially In Kigm (irobw1iig Sm iteE Dt aboNe) dxw 297 2,563 8,58M 11,973 24,296 29,35D 16,a02 Smx.e: Kfgom NeWmal Bmc-b of NC, TrA ad GM (Gem1 hgWotI PbtM EXIIWtGOiptLu), M 31, 1983 BEST COPY AVAIUBLE  TAW-4NM RNELM crM, CR.50 1974/75 1975/76 1976(7 1977/78 19789 1979180 19BD%V8 19811U2 19828 TotD Pradect Costat (Onrent Idce) I/ 1,250 5,0C 20,0 16,250 15,O0D 31,250 26,250 10,00W aF BOfA=*r 21 14.5 15.0 19.8 8.0 8.5 15.0 12.5 9,9 Factor for Cceatnt 1977/78 1TIm 1.69 1.3) 1.08 1.0 0.93 0.80 0.71 0.65 Ftoject Con xi Cmtt 1971/78 MPi 2,113 6,5m 21,6o 16,250 13,950 25,OOD 18,638 6,5(0 PXoJct Cost dlh Sbakm P1icd FE (1.61)3/ 3,402 10,465 34,776 26,163 22,46D 40,250 3D,07 10,465 Day IaIked al Tal Productin ('100) 5 (I dayper Tah 7.07 mtpt) / 7,332 7,590 10,367 13,327 16,007 18,855 21,08 23,420 Itrctxital Lator Sbwa rImd6f 0 397 4,675 9,234 13,360 17,747 21,15 24,778 24,778 project Clt After %dxctio 7 an Mialiul, litIh, WMds a.A Ccfee / 2,109 6,488 21,561 16,221 13,925 24,955 18,604 6,488 0 Total Project Ca After Altoer 2.109 6,885 26,236 25,455 27,285 42,72 39,729 31,266 24,778 Per LEa Prtion 4/ (194) 194 256 318 369 420 453 487 Ppiattci 340,0tO 1981/82 9 3.51 &wtl 267,20D 276.6W 286,300 296,350 3D6,70D 317,400 328,5 340,0W Total ProdEto Valm at 1977/78 OfliclA Prim, (t,COD) 51,836 53,660 73,293 94,223 113,172 133,308 148,810 165,5,0 Toil ritLm Wme at AdjoetM 8 Pd$ / 77,754 80,49D 109,939 141,334 169,758 199,962 223,215 248,370 increael Productiom at Adueted Prims 0 2,736 32,185 63,580 92,X4 122,20B 145,.61 170,616 Thaeortal P Attrtbled miction to tie Projet: (3=) 0 493 6,383 12,662 18,346 24,3B7 29,038 34,069 34,069 (+5I ER1) (I(1 0 247 3,192 6,331 9,173 12,194 14,519 17,035 17,035 (-4 EBR) IT fron Agriculture 11nwiI eloput 1tt. Ag *. FM draft fcr total project cotn by catwu* sead ove projea pd in SW diavleant. SO 2/ Sept. 24, 1981mm frm Idtaloe,EAMi -wIac pricg aama FE ia 61:;iei obaw rate daie offidal rate. lit 612 FEis d¶ ai appwief estijtes adjoi for it ie FE am sof 90 (39 + (61 x 2) - 161 - 100 - 1.61). 4 Me lt MmOzai of agonsi l= relot Project (Loft, 1.* ad Fyesmb*ol 1982)_ats (R 111) tdit tde v1c atel f agr Ltural mv&vdit in tk ae/bejs syste mm fat T1YB 25W9pr caita in 1975/76to Tale 342per capit in 1977/7 to 1978/79 aixae. Gwth fto tie to-ear pd 1975/76to 1977/78si forel to be 6S2fwo amva. Foltwp aaress (mct qensifId) abed a cmtimed Iif In bfxmlih at a re&tnid rate. lt Uterfom n it 1977t78to 1979t8D x*txdt Ii to bt If 62 at 32r wa 1979180 to 1981/82e be lif vin at 18X, 5/ Mm 1977/78 cesy sboe Tle 362 pe npita in 197/79 pxi for 45 wo days par capita, or, adjusted Tal 318 In 1977/7 prie to 45 wsk das, ie. Tib 7.07 outpt per t& day. D.ayswke ate tis dwivea fxs de veli of p dztion rt ,ag tle TrI 7.0 figus. 61 tabor is dili primd at Tat 7.7 per day mo6ed, ie. 511 of tPt axrret il*m wi In 1977/78 pdos b.d n te 20 e af iinea1 atpat to be attribited to project. 7/ Otso of E6natbz4, uilth, mtob ad coffie repaa 3Z2 (T w forud Wt to daP). t/ To elw for tle fact that ,magt prees hwe. genrany eb h r dsai afidal price ad that l two rt parity prices b i hbe eW -i offia prime atwl "he of proedacio nn 19778 prI ha bs m aIjusted q-=* ? March 8, 1*3 Rlr*T "MPV AAlUAIIAR  kI,bP COMPO1ENTJS FIGURE 1 EFFECT ULIKAGES OF THE (Ol.GIiJL 1 U TAKAETS) beSiG b'Aild l0) mile CarkateA trpwif.Et ~~~vI C&I~ UAp t S _ s elf 4 ( < lk Icep ( 9 t efen H g rl~~~~~~~~~~~~~~4 _ eVed AwCU_SI ~~~~t~F socAbs juft r."fi ^h; m W*"s 4 O'(prowq^ ~~~~~a ~ /StWt es' & ittin*C g sr / b stft ~ ~~~~~~~~~~ etl t AILhc,; RgIos1g. rz gsetis M.osSt'tr Of Tiii MIMP; tMf'LtUI(aTATg@U (1) CoojI.r.ti, unIo. le c AvlsU e Ai;c Teel ma,^..;. tt Sf wiloesa !} @ ^s~~~~~~~~~~~~~~~~~~~~~~-.Se. rsn Can 03 094C.4Iwl a Ay,;c.beftJop,.. ot,k Report, Figure 1.. Source: Project Evaluation AVAILABLE COPY BEST  -49- CIART 2 impact Incidence Matrix Component landu e 1nn bush crnL water up hvaLth *r,es _m1o Village __sodcwns of pro45sat 1% 14% 1% 0.3% 1% assure nat * encourage reduce reduce reduce provids Intended resource livestocks water-borne illnees tim. far marketing impact basns too early diseaes,gS nFg5ieikbl food pre- links ve on - Villagess m* lprove parations population anns$ health humt of reduc Zduet Unintended prevent land reduce facilitate lower morale improve impact on disputes vormins ty house con- in villages nutri- target among vilLagesao*bsidi* atructiona with un- tional populatLon -!s* finished standardos exclude appro- acreag. possibl projects priate popula- expansions inrese - tionfroa support sz i aso project ua'adne activities, ln4 usme .t -ve fir Other provide data subsidise income to income to unintended base for village village Village impact planningsvy govt. govt. Ive govt.s.ve , . . _ .. ._ _i.come,.±_ Chart 2 Continued coinpone tbookeepe itr,h8Dt-t" roads education w6Zc trial.s/. - uCoffee, Component bootr eeper _arketing xtension eo trnp loans Cpos 31 n. 27% 117U 47 i% provide expand maintain, induc- duption of increaa-e rv.iv-e Intended x rketing region iprove la receptivity recomemnded Cotton & corree impact links Marketing feeder to and practieesa maize growing onv capacitys roadateo_ediffusion Desticides, yiolds: in high- target Yt install of approp ytgotherr,no et irilj landzs population capacity technical increase -xten- for MaJor knowledge aion statie ti road works on produz- create region 2t Yes tim.t Ii_ l__t specific b1owltedge.us Unintended enable promaot bring pr-w-ns impact on coll*etive unofficial willagers needy from target economic marketing, into rain- getting in- population activitions ±ve for sreamn of putsepre- eve & -vt r.a-rve a nat'l life vent other town dwel-. v conponentc lr,: -w Mftwe fromresch- for NEtC brick lug villages. teobmoloC: 'iususe lega-t C .ve tive twd. attitude Inputs put Villages 5 in long-term Other rais ve create -ATZ, - a tva_n unintended of admini.- relations impact trativ* di ert projett between Competence funds from TCA & elsewhero regional pur- TRDB,.-A Source: Table 5.3. Project Evaluation Report; Costs in Table 1 of this Report and in Project Evaluation Report, Chapter V. BEST COPY AVAIUBLE  IBRD 10942 --- in=ternaionl n77,e hmndarlr' doo, . Itol 1,, dlop 32 NlRWAN&jK VIttotria j boundarles lmpiv .dn0,'".uIf tr op-/t "hy MARCH 1974 ¢>_[' q > > > ~K E N YA </ )BURUp46l . PROJEcr N *Klgoina ) AREA L TANZANIA - l I lat ho _ \ olooj _ Tangnyika 'Dar es Sal,a OF ZAIRE AM B I)A4 Lake f- NNyGsO ;`.,MiZAA6l0* E00,0 . 0 <\~O~UR'U NDI 45, a)\\ t sus V _} 40Q / j0 dnEogu0 0001 0 00 C Z c,ike 0 00 - Kigorenal bou 't 21,000 UjiV \T7oganyiy/ka 0 I N -.. Sag.ri 7ga9ao N~~~~ 4 N ~~~~ TANZANIA _ |. X / 0\ 0eKigorrni Rural Development Project POPULATIONDENSITYAND < ~~~1NMAJOR ZONES ECOLOGICAL ~~Ecological Ni ~~~~~~~~~~~~~~~~~~~ ~ LL ~ ~ S -. / 0 ~--.-\Highlands zones: ~~~~~~~~~~~~~~~~~~Lakeshore --- ....... Regional boundaries Districtboundaries Intermediate ~ l District headquarters 60 r66 ° /., Ct\ ( 8~Miornbo Main roads 1j > \\ ~\ tk Mostlyuninhabited ~ Rivers f t \_< Populationdensityl Internaiional boundaries ~ co ~ ~~ ~ ~ ~ ~ ~~~~)~Each dot represents Di 1000 people 0- o ) 7,000 Urban popuilation LLJ / ' asedanthe 1967Census > \ / 20 40 60 60 100 KIL KOMETERS \\ *,,,/ 9 2p 40 60