Tanzania - Dairy Development Project

TDue ct The World Bank FOX OMCAL USE ONLY Rewi No. 5747 PROJECT COMPLETION REPORT TANZANIA DAIRY DEVELOPMENT PROJECT (CREDIT 580-TA) June 28, 1985 Eastern Africa Projects Department Regional Mission in Eastern Africa Thi iocmmt k a rutuckd dtLudem m_d y be used by uekdples Y mli the peufonmc of tbdr gUru d h Ias eomob my dM eise be d_da ed witha Wod BDak 2uihodimn  ABBREVIATIONS AI = Artificial Insemination CDL Coastal Dairies Industries DAFCO = Dairy Farming Company DANIDA = Danish Aid Agency ECF = East Coast Fever EEC = European Economic Community ERR = Ecovomic Rate of Return Fl = First Generation from Cross Breeding FMD = Foot and Mouth Disease FRR = Financial Rate of Return GDP = Gross Domestic Product GOT = Government of Tanzania IBRD = International Bank for Reconstruction and Development IDA = International Development Association KG = Kilogram KNCU = Kilimanjaro Native Cooperative Union KUC = Kilimanjaro Uremi Corporation LIDA = Livestock Development Authority MIFUGO = Livestock Development Division, Ministry of Agriculture MOA = Ministry of Agriculture NAIC = National Artificial Insemination Center NARCO = National Agricultural and Food Corporation NCCO = National Cold Chain Operation NMC = National Milling Corporation PCR = Project Completion Report RMEA = Regional Mission in Eastern Africa (World Bank) SAR = Staff Appraisal Report SIDA = Swedish Aid Agency TAC = Tanzania Audit Corporation TDL = Tanzania Dairies Limited TSA = Tanzania Sisal Authority TRDB = Tanzania Rural Development Bank WFP = World Food Programme  FOR OFICIAL USE ONLY TANZANIA DAIRY DEVELOPMENT PROJECT (CREDIT 580-TA) PROJECT COMPLETION REPORT Table of Contents Page No. Abbreviations Preface .............................. Basic Data Sheet ......... . .......... 00 6........................... (i Summary of Highlights ....................................... . (iii) I. INTRODUCTION ................... ..o.................. a............ I II. PROJECT IDENTIFICATION, PREPARATION AND APPRAISAL .o.. 2 III. PROJECT IMPLEMENTATION AND OPERATING PERFORMANCE ...... 4 IV. PROJECT IMPACT .......................... 19 V. FINANCIAL PERFORMANCE o . ............................... 21 VI. INSTITUTIONAL PERFORMANCE AND DEVELOPMENT ...- ....... 23 VII. BANK PERFORMANCE ........... ...... . ...... o.... 24 VIII. ECONOMIC RE-EVALUATION ............................... 25 IX. CONCLUSIONS ............... .......................... 26 TABLES 1. Credit and Project Summary 2. Summary of Annual Project Costs: (a) Project Components (b) Category of Expenditure (c) Summary 3. Estimated Schedule of Disbursements 4. Key Indicators - Targets 5. Production Co-efficients : Parastatal Farms 6. Fully Stocked Parameters : Actual vs. Appraisal 7. (a) Milk Prices (b) Farmgate Prices in Real Terms 8. Summary of TDL Plant 1976-81 : Fresh & Reconstituted Milk 9. Grant of Imported Raw Materials from 1975 to Date 10. TRDB Loan Portfolios : Smallholders and Ujamaas 11. TRDB : Loan Status of Parastatal Dairy Farms 12. Operating Statement Summary : Parastatal Dairy Farms 13. Analysis of Operating Costs (1981) : Parastatal Dairy Farms Map This document has a resticed distribution and may be used by recpients only in the performanoe of their ofricial duties. Its contents may not otherwise be dislosed without World Bank authorization.  - i - TANZANIA DAIRY DEVELOPMENT PROJECT (CREDIT 580-TA) PROJECT COMPLETION REPORT Preface This PCR is based on a review of the Staff Appraisal Report (No. 765-TA, May 23, 1975); the President's Report (No. P-1661-TA, June 18, 1975); the Development Credit Agreement and Project Agreement dated August 15, 1975; supervision reports; internal reviews of project progress by the Ministry of Agriculture (1980) and LIDA Technical Services (1982); relevant correspondence and information co-.tained in the RMEA files; and a mission to Tanzania in April-May .984 to collect additional data and discuss preliminary findings w%.ch officials in concerned agencies. The PCR was submitted to OED for review but the Project was not selected for audit. The PCR was sent to the Government for comments on March 4, 1985 but none were received.  PROJWCT PERF0r _C5 AUDIT RSC DATA SWLET DAIRY DWVtLOAPEM PROJECT (CCDIT 540-TA) KEY PROJZCT DATA Appraital Actul or Actal an of C tiate Litiated Actual *l Aspraiml Betilate a/ Project Coats (USS million) 15.3 n.&. - Cre4it Aount (MO mill. on) 10.0 h.8 an Dieburmed ) 10.0 8.6 8sx Repaid ) to March 31. 1963 Outstanding ) Cancell.d 1.2 122 Date N gotiatione Date Board Approval 6/30/71 6/30175 Date Effactivana. 9130175 11113175 Date Physical Coopow.nte Completed 4130/J1 3/31183 Closing Date 4130/61 3/31/83 Proportlnn then completed (E) Economic Rate of Return (2) a/ 18S probably negative CUMULATIVE DISBURSSEErNTS FY75 FY76 nn7 FY73 n 79 FY80 nu FY82 FY63 ApDriesal Eatiqate (USS million) 1.0 4.7 7.9 9.1 10.0 10.0 10.0 10.0 10.0 Actual (IlLS million) - 0.1 1.3 2.8 3.4 4.0 4.4 5.4 8 Actual me 2 of Appralmal (S) Ox 2S 162 312 342 ns 44 542 844 Date of Final Diabursauent: March 4th, 1963 MISSION DATA Month/ eNo. n Kandays Speciallzationo Porfar _ Type of Miasion Year Per on i n ield pay-rcntd b/ Ratiag C/ Trand dA Problem I Apuraioal 09/10/74 6 168 FA(2)EC. LIV(2). EKC Superw1sion I 11/75. 06/76 2 19 LnT/rC 2 2 M. T. P Supervision 1I 01/77 3 is LII/DAI 2 2 P. F Supervielon III 06-08/77 2 Is LITADU 2 2 P. M Supervision TV 02-03/78 1 9 LIV 2 1 P. M Supervision V 11/78, 01/79 3 23 LIV/DAI/FA 2 3 F. M Supervision Vl 04-05/79 2 24 LIT/lWI 3 3 F. M. T Sunerviajon VII 12/79 2 14 ACIR/D& 3 1 F. H Supervicion VIII 05/80 1 14 am 3 1 F, M. T Supervidson IX 11/80 2 24 ACR/FA 2 1 F. M Supervislon X 07161 2 28 AOl/IA 2 1 F. II Supervision XI 01/82 2 28 ACR/IC 2 2 F. N. P Supervision Ul 07-08/82 2 36 "IX/IC I 2 P. M. T. P Supervision X111 01/83 2 14 ACIVDC 2 I F. K. T. P CTM PROJECT DATA BorroCer Government of Tanzania Execucing Agecies 'Unietry of Agriculture. Tanzania Imral Developmmnt Rank tTRDII and the Datry Parring Coeany Fiscal Year of Borrower April I -Mrch 31 Name of Carrency (Abbreviation) - Mob. Currency Exchange ates: OSSI.00 7.1 Appraisal Year Average USSL.00 - 6.2 Inervening Years Average USS1.00 - 12.4 Pollow-on Project: a/ Me 1-iformtion available from COT on final project eAet-; It baa therefore not been poasible to re-calculate the financial and economic rates of return. b/ ACR - Agricultualist; EC - Agricultural Economist; 5A - inancial Amalyat; LM - Livestock SpecIaliet; DAI - Daisy comealtant: ENc - Dairy Proceaaing Engineer. cl I - Problem free: 2 - !arate problems; 3 - Major peoble. d/ I - Improving; 2 - Stationary: 3 - Detertorating. el F - Financial: M - Management: T - Technical; P - Political.  -iii - TANZANIA DAIRY DEVELOPMENT PROJECT (CREDIT 580-TA) PROJECT COMPLETION REPORT Highlights 1. The Project was the tenth Bank Group supported operation to assist agricultural development in Tanzania and the third in the livestock subsector. It was conceived as an integral part of the Government's long-term program for the improvement of dairy farming in Tanzania. The Project, to be implemented over a five year period, comprised rehabilitation and expansion of commercial dairy farming on nine parastatal farms, improvement of milk collection, processing and distribution facilities, development of the traditional dairy sector in 50 ujamaa villages, breeding of improved dairy heifers, technical services and applied investigation work for animal disease control, and future project preparation. It was an Import-substitution project designed to reduce the increasing amount of foreign exchange being spent by Tanzania in importing milk and dairy products to meet a rising urban demand. 2. The Project, prepared by the Government, was appraised in the field in September-October 1974, and negotiated in Washington in May 1975. The IDA Credit of $10 million was to finance 65% of total project costs ($15.3 million), including all foreign exchange costs and 33% of local costs excluding taxes and duties. The balance was to be financed by GOT, parastatal corporations and ujamaa villages. 3. The Project encountered major problems throughout its implementation. It was executed against the backdrop of a rapidly deteriorating economy, where the general climate for increased agricultural production was extremely poor (para. 3.28[g]). Parastatal farms under tl.e Project consistently operated at a loss, with high overhead and administrative costs (para. 3.27). As the supply of concentrates, food, fuel and spare parts worsened, farm managers often spent considerable portions of their time away from their farms searching for these essential items. This distracted them from day-to-day farm management and contributed to low productivity. Moreover, with increasing scarcity of commodities in an inflationary environment, the only way parastatal farms could make a profit was by selling milk on the parallel market. This, in turn, provided opportunities and inducements for misappropriation of official funds. In fact, the parastatal farms' financial records were so poor and questionable and the quality of their management so deficient that an RMEA supervision mission in 1979 seriously considered recommending termination of the Project, and finally in 1982 an independent financial analyst had to be brought in (at IDA's insistence) to improve on-farm record keeping and the general standard of accounting (para. 3.25). The state of accounts, however, continues to be poor: no financial statements have been prepared so far for 1982 and 1983, and it is impossible to obtain an accurate breakdown of total Project costs.  iv - 4. At the same time, the Project's ujamaa component never really took off. Peasant producers, demoralised by the villagization policy which tried to force them to produce officially-controlled cash crops under unfavorable conditions, had retreated into food production. They were not receptive to any further government intrusions into their economy, and it is hardly surprising that the ujamaas did not show much enthusiasm in taking up loans for dairy development (para. 3.28[e]). Once this was apparent, the Project was redesigned to allow TRDB, as the on-lending channel, to support a number of i-adividual smallholders, but again with limited success. Other improve,rtents in the collection, processing and supply components of the Project were effected smoothly hut became less important as there was little increase in commercial milk production. 5. IDA supervision missions from 1976 to 1983 correctly identified technical and other constraints impeding project implementation and repeatedly suggested corrective actions- (paras. 3.06-3.27). The constraints included: a lack of locally available cattle to stock dairy farms, poor husbandry which reduced the quality and output of available stock, low milk prices, poor on-farm training, improper record keeping and cost budgeting on farms, non-existent financial controls, few incentives for staff motivation, ineffective technical assistance and high overhead costs, among other factors. However, many of these constraints were also symptomatic of general government policies and of larger cAses in the Tanzanian economy and hence were beyond the scope of the Project to resolve. Even where solutions were applied, such as four successive increases in milk prices, they were conceded with reluctance by GOT and were too little and too late to have much of an impact. Given the consistently higher price of milk on the parallel market over that paid by TDL (Tanzania Dairy Limited), the milk-buying parastatal, there was little incentive for producers to sell at official prices. Consequently, the financial viability of the ProJect remained precarious and the parastatal dairy farms showed few signs of being able to support themselves. 6. Under the circumstances described above, it is hardly surprising that the formal market sector could not capture significant portions of any food produced, including milk, for distrihution to urban consumers. Tncreases in commercial milk production during the period of Project implementation were much smaller than targets and are prohably not sustainable. The appraisal goal was to iacrease the contribution to domestic commercial milk supply from 3 million liters in 1976 to 5 million liters by 1985, with an annual incremental project production by year seven of 9 million liters. As of 1982, the parastatal farms were contributing only an incremental 3.1 million liters to the domestic commercial supply. TDL's total intake of fresh milk actually decreased by 20% from 1976 to 1983, while imported milk products as a percentage of total throughput increased by 7% over this period (para 4.01).  -v - 7. In the absence of reliable accounts, it is, of course, difficult to say what part of milk production on parastatal farms was understated and heifer mortality over-reported to hide sales on the parallel market. The &ignificance of these factors in comparison, say, to technical constraints, husbandry and poor management, is difficult to quantify. Nevertheless, it is probably fair to say that the Project has not had a significant impact on smallholder incomes and nutrition. Project costs were higher than originally estimated and any benefits that are likely to accrue will come over a much longer development period than anticipated at appraisal. The economic rate of return on project investments is therefore low at best and possibly even negative. 8. It is possible to argue that: (a) the Project turned a number of run-down dairy farms into functioning entities; (b) there has been a spillover of dairy cattle from the parastatals into the smallholder sector; and (c) the existence of large-scale farms allowed for the importation of cattle and an upgrading of the domestic stock wnich could not have been undertaken by smallholders and may improve dairying over the long run (para. 4.03). Nevertheless, it remains true that dairy development through parastatals has proved to be too expensive and cumbersome to justify the costs, especially given that most of Tanzania's domestic milk supply is still produced by smallholders. If smallholder production near urban areas had been supported and the price had been right, it is possible that the commercial supply of milk to the cities could have been increased with fewer costs, lower overheads, and with a minimum of management problems (though the difficulties of providing extension, health and marketing services to small farmers should not be minimized). 9. A major lesson of the Project is that attempts to appraise or implement projects without carefully taking account of the likely effects of the general economic environment may lead to false assumptions, unrealistic expectations and ultimate failure. The Bank shauld be particularly cautious about developing large-scale, technically demanding projects in situations where the absorptive capacity of the economy is low (para. 7.01). A more feasible approach under such circumstances might have been: (a) to have developed a small-scale pilot project in an area or areas where the supporting infrastructure was available; and (b) to phase further project development so that heifers were available, pasture was developed, and management was trained, before expecting the dairy farms to operate as individual profit units. Had the Project been approached more cautiously, it is unlikely that the appraisal report would have assumed that sufficient heifers were available locally to stock the dairy farms or would have expected them to attain so many challenging targets (para. 9.01-9.03). Moreover, though the appraisal report correctly identified the -principal risk of the Project as management, it did not offer much in the way of remedies. More attention to this issue at appraisal could at least have led to closer examination of the existing management structure (including the role of parastatals) and more specific proposals for allowing project managements reasonable autonomy in operational decision making and staffing policies.  TANZANIA DAIRY DEVELOPMENT FROJECT (CREDIT 580-TA) PROJECT COMPLETION REPORT I. Introduction 1.01 Agriculture and related activities constitute the largest single sector in the Tanzanian economy. At the time of appraisal, roughly 40% of GDP was derived from this sector, of which 50% constituted subsistence production. Agricultural products accounted for 80% of total exports. About 94% of the population lived in rural areas, and 90% of the economically active people were engaged in agriculture. 1.02 Tanzania's development policy as it evolved in the late 1960s and early 1970s aimed to provide increased support for agricultural production, with emphasis on rural development and equity. Policies intended to improve agricultural production during this period included, directly productive investment to develop parastatal farms, compulsory villagization of the rural population, administrative decentralization, and changes in the marketing of crops through government parastatals. The overall result was a profound dislocation of rural life, with resulting declines in output. 1.03 Within the agricultural sector, dairy development was accorded priority for economic, social and nutritional reasons. The dairy industry was relatively underdeveloped, but was viewed as having high potential due to geography, climate and past experience with cattle husbandry. A small commercial dairy sector was engaged in modern dairy farming, using almost exclusively non-indigenious dairy cattle. A few farms were privately owned, but most were run by parastatal farms, serving commercial dairies primarily in the Arusha-Kilimanjaro-Dar es Salaam areas. Commercial dairy production had declined to approximately 10 million liters in 1973 due to the departure of non-African farmers and low milk prices, with many farms in need of urgent rehabilitation. The largest volume of milk (approximately 475 million liters) was produced by traditional cattle owners, with half going to cattle and the remainder for human consumption. A few traditional cattle owners and a limited number of ujamaa villages also produced milk for sale. 1.04 Despite rapidly rising urban demand, raw milk sales to the two major dairies declined from 8 million liters to 5.5 million liters between 1969 and 1973. At the same time, imports of dairy products had increased from TSh 19 million (US$ 2.7 million) in 1962 to TSh 70 million (US$ 9.8 million) in 1973, and accounted for 24% of all food imports. The Government's strategy for dairy development sought to reverse these trends by developing large scale parastatal dairy farms to substitute for imported milk products and, over the longer term, developing dairy units in villages throughout Tanzania to provide milk for the rural population. The Project was designed to be in line with this strategy and, at full  -2- development, was expected to save the country TSh 20 million (US$ 2.7 million) annually in foreign exchange spent on imports. 1.05 Responsibility for implementing the dairy development program rested in the Ministry of Agriculture (Kilimo), whose Livestock Development Division (Mifugo) had policy responsibility for all matters relating to livestock and dairy. Mifugo carried out its policies through its research, training, husbandry, extension and veterinary departments and through parastatal corporations these parastatals were under the management of the Livestock Development Authorith (LIDA) the National Ranching Company (NARCO), the Dairy Farming Company (DAFCO), and Tanzania Dairies Limited (TDL). DAFCO had responsibility for developing large-scale dairy farms, and TDL for developing and managing commercial milk collection, processing and distribution. NARCO operated beef ranches and breeding ranches. In addition, the Tanzania Sisal Corporation (TSC) operated beef ranches and dairy farms in the Tanga region as part of its diversification program. The Kilimanjaro Native Cooperative Union (KNCU) and the National Agricultural and Food Corporation (NAFCO) also operated dairy farms. The Tanzania Rural Development Bank (TRDB) was the primary financial vehicle for promoting agricultural credit. Operating on commercial principles, TRDB handled short and medium-term credit to farmers, provided some technical assistance to farmers, and had historically served as the main channel for IDA funds in previous Credits. II. Project Identification, Preparation and Appraisal IdeniAfication and Preparation 2.01 The project was prepared by the Government, which submitted a loan application to IBRD in june 1974. The Government's original project proposal included 41 large-scale dairy units, clie purchase of 5,000 imported friesian dairy heifers, and the estab_ishment of a pilot Disease Free Zone to control foot-and-mouth disease (FMD). Appraisal 2.02 The Project was appraised in September/October 1974 by a team comprised of five Bank staff and a consultant from FAO. During appraisal, it was agreed that for technical, financial and management reasons, the first phase of the Project should develop only 17 large-scale units. The mission recommended against any cattle importation, because of the high death risk involved. It was assumed that the proposed heifer breeding unit would be able to provide an adequate supply of upgraded cattle within a few years. The mission also concluded that the establishment of a Disease Free Zone was not justified in economic terms at the time, due to lack of evidence on the impact of FMD on productivity in the traditional herd. The program of applied investigation work included in the Project was designed to address this issue.  -3- r 2.03 The Project was to represent the first stage of a long-term program to develop dairy farming in Tanzania. It would, over five years, support dairy production on parastatal farms, improve milk collection, processing and distribution facilities, and provide the basis for dairy development in the traditional sector by supporting a pilot ujamaa dairy program and by producing upgraded dairy heifers. Specifically, the Project included the following components: (a) Large Dairy Farms. Seven new dairy units were to be established and ten existing units rehabilitated and expanded (each with about 350 cows) at a total of nine locations near existing or planned dairy processing plants. The farms would be run as commercial ventures by DAFCO and other parastatals (see para. 2.05), following modern intensive dairy farming practices with predominantly non-indigenous cattle and a high degree of mechanization. The farms were to serve as a rucleus for development of farming techniques applicable to different regions and geographic conditions, and for training managers and farmers in dairy farming. Items to be financed included buildings and installations, water development, equipment for pasture development, fodder production and preservation, milking machines, cooling equipment, vehicles, cattle, and working capital. (b) -Ujamaa' Dairy Farms. 20 small dairy units (each of about 20 cows) -were to be developed on a pilot basis in 50 ujamaa villages. The units would be based on upgraded cattle from the heifer breeding unit and run as commercial ventures. TRDB would be responsible for preparing investment plans and supervising the units. Since the development entailed the risk of animal death, alternative sources of cash revenue were to be used to service the loans. (c) Heifer Breeding Unit. A breeding unit was to be established to produce 1,800 dairy heifers for sale to parastatal dairy farms, individual farms and ujamaa villages. The breeding program would be based on artificial insemination with imported semen from pedigree dairy bulls. (d) Milk Collection and Processing. Milk collection systems in Dar es Salaam and Arusha were to be improved by locating collection centers on dairy farms which would also serve small producers. Processing and distribution facilities were to be expanded, obsolete equipment replaced, production bottlenecks removed, and vehicles provided to improve milk distribution. (e) Technical Assistance. A total of 30 manyears of technical assistance were to be provided, including: (i) within LIDA, the services of dairy farm group managers, a manager for the heifer breeding unit, specialists on machinery maintenance, dairy husbandry and pasture improvement, a dairy economist, and consultants on irrigation, processing and farm structures; (ii) within TRDB, a project preparaEion team consisting of a livestock specialist, an agriculturalist, an animal resources and extension specialist, a pasture specialist and other experts; and (iii) within the Ministry of Agriculture's Livestock Development Division, a veterinary economist to conduct a program of  -4- applied investigational work for animal disease control. Specialists recruited under (i) above were to be specifically charged with training Tanzanian managers and developing management systems and administrative procedures. 2.04 Detailed lists of Project components and costs are presented in Tables 1 and 2, and the estimated disbursements schedule is shown in Table 3. Total Project costs were estimated at US$ 15.3 million equivalent. The IDA Credit of US$ 10.0 million was to finance 65% of the total Project costs, including all foreign exchange costs and 33% of local costs excluding taxes and duties. The balance was to be financed by GOT, parastatal corporations and ujamaa villages. 2.05 The Ministry of Agriculture had overall responsibility for Project implementation. The proposed dairy units were to be owned by DAFCO (8 units), KNCU (3 units), TSC (5 units) and NAFCO (1 unit), while LIDA was to be responsible for coordinating their management. TRDB with the regional livestock staff was to be responsible for the ujamaa dairy program, NARCO would establish the heifer breeding unit, and TDL was to be responsible for the milk collection and processing component. The Ministry of Agriculture, L!DA and TRDB would direct and coordinate technical services. LIDA would oversee and monitor project implementation through a iroject Coordinator, and LIDA's Board of Directors, which had members from all agencies involved in the Project, was to serve as the Project Coordination Committee. Agreements Reached 2.06 During negotiations in Washington in May 1975, assurances were obtained that all the parastatals involved in the Project would maintain separate financial records and accounts for Project-related activities. Each dairy unit was to be treated as a commercial enterprise or profit center. Assurances were also obtained that farm investment plans and the development plan for the heifer breeding unit would be subject to IDA approval, that detailed plans for ujamaa dairy development and for applied investigational work on animal disease control would be submitted to IDA for comment, that LIDA would consult with IDA on the appointment of group managers and farm managers, and that consultations on milk price levels would be held with IDA on at least an annual basis. Execution of a subsidiary loan agreement between GOT and TRDB and recruitment of at least three group managers for dairy farms were conditions of Credit effectiveness. III. Project Implementation and Operating Performance Effectiveness and Start-Up 3.01 The Credit was declared effective on November 13, 1975 and a start-up mission visited Tanzania in November-December 1975. Four farm managers who had been identified were interviewed and found suitable.  -5- LIbA had prepared plans for the development of various dairy farms which were criticized by TRDB as 'desk exercises" on which the parastatal boards and farm managers had not been consulted. TRDB also complained that LIDA was not complying with many of its lending procedures, including a clause requiring proof of land ownership by the parastatals. This was the beginning of a major bottleneck to disbursements and also reflected the difficult relationship between the two major parastatals handling the Project. LIDA was making plans to import a large number of cattle possibly from Kenya, Zambia and Australia, but indicated that it was not convinced that cattle could not be procured locally. There was also some disagreement over the need for milking parlors and milking machines: LIDA's Managing Director wanted them on all 17 farms while the mission felt they were suitable for only four farms. As a compromise, it was agreed to procure the original eight included at appraisal. LIDA had not yet prepared plans for the heifer breeding unit, the artificial insemination unit (on which the Project depended) financed by SIDA was behind schedule, and little other progress had been made. In retrospect, the start-up mission's findings were a microcosm of many problems which later came to haunt the Project. Implementation 3.02 Summary. Tables 4-6 summarize actual physical achievements every year under the Project in comparison with appraisal targets. Farm development proceeded satisfactorily but most of the other Project components lagged well behind appraisal targets. Of the 17 large-scale farms to be established at nine sites, 15 units (representing 12 farms) were actually developed at eight sites. However, for a long time TRDB did not disburse f=nds to DAFCO because the latter did not have title deeds as security for loans to its dairy farms. Two further units were also approved but were not financed by the Credit. The original heifer breeding unit was considered financially non-viable for a TRDB loan and dropped from the Project (plans for a smaller breeding unit were later approved in 1981 and implemented through bilateral aid). Of the 50 proposed ujamaa village units, only six were eventually implemented. Changes in the Credit Agreement in 1981 added a smallhold'- component of 80 beneficiaries. By January 1983, TRDB had approved 57 loans for 87 cattle, but only 45 cattle had been supplied to 32 farms (a further 26 loans were approved by November 1983). TRDB also approved a loan in March 1976 of TSh 4,286,000 to TDL to rehabilitate the Dar es Salaam milk plant and double its daily processing capacity to 90,000 liters. Additional loans amounting to TSh 790,700 were approved to purchase a new ice water plant in 1979 and to install a new boiler in 1982. A further loan of TSh 4,162,000 was approved for the Arusha plant. Twelve of the 16 technical assistance posts were filled but were all financed by bilateral sources outside the Credit. Funds originally intended for technical services were used to finance accounting services and vehicles. Three short-term courses of 3-4 months each and one long-term course were also funded. A team of four veterinarians, including two Tanzanians, examined the possibility of establishing a Disease Control Zone in the Lukwa region. The fieldwork was complemented by training two Tanzanians abroad.  -6- 3.03 Progress in Disbursements. At all stages, disbursements from the Credit lagged well behind the planned schedule. Initial reasons for this included delays of up to two years in preparation and approval of farm plans and in procurement and distribution of farm machinery, inadequate security for TRDB loans, delay in completion of management agreements between farm beneficiaries, change in ownership of KNCU farms to KUC (see para. 3.05 below), lack of budgetary provision for working capital for DAFCO in 1976 and 1977, and IDA's initial reluctance to approve the external purchase of cattle and subsequent delays in importation (especially due to closure of the border with Kenya). From 1980, non-payment of interest by the parastatal dairy farms on TRDB loans led it to suspend further disbursements on these loans. Finally, between June and November 1982, the Project was affected by the temporary suspension of all World Bank Group disbursements to Tanzania. 3.04 Due to the delays in release of Credit funds for one or more of the above reasons, GOT grants and World Food Programme (WFP) funds were often used to prop up the Project through the financing of not only operating and maintenance expenses on parastatal farms but also a good deal of investment in farm infrastructure (staff housing, farm building, and water supply). This reduced the amount ultimately contributed by the Credit. Thus, only US$ 4.8 million (48%) out of the Credit had been disbursed by the original Closing Date of April 30, 1981. Additional procurement (mainly of heifers and spare parts) and followup on disbursement applications from TRDB and other agencies increased total disbursements from the Credit to US$ 8.8 million by the extended Closing Date of March 31, 1983. The major share of Project investments went into farm machinery and the purchase of cattle. 3.05 Amendments to Legal Agreements. A number of amendments were made to the Credit and Project Agreements in the course of implementation. Originally, IDA had agreed to finauce 80Z of all subloan disbursements by TRDB. Because of Tanzania's serious foreign exchange constraints, the Credit Agreement was amended on June 24, 1976 to allow IDA to finance 100% of TRDB's disbursements in foreign exchange (and 60% in local currency) so that procurement of imports would not be held up. An amendment to the Project Agreement at the same time required each of the five dairy farm group managers to report directly to the management of the respective parastatal beneficiary. Later, the functions of the Kilimanjaro Native Cooperative Union (KNCU) were taken over by the Kilimanjaro Uremi Corporation (KUC), which also assumed responsibility for two dairy farms. The Project Agreement was amended on July 5, 1977 to reflect this change. On April 12, 1978, the Credit Agreement was again amended to reflect the need for importation of dairy cattle. Another amendment on August 15, 1980, recognizing the financial plight of the parastatal farms, allowed them 20 rather than 15 years to repay their loans, with a grace period of five instead of four years. At the same time, smallholder dairy units (in addition to ujamaa villages) were allowed to become recipients of TRDB loans and two new dairy production units at Kitulo and a milk plant at I4beya were brought under the Credit (subject to satisfactory farm plans and feasibility study, respectively). A final set of amendments on January 14, 1982 extended the Closing Date from April 30, 1981 to March 31, 1983, phased out three problem farms (at Gararagua, Bagamoyo and Ngerengere) following a review  -7- of their performance in April 1981 and alleged financial mismanagement, reduced the total number of parastatal farms from 17 to 12, and redirected the funds allocated for the Mbeya milk plant to the replacement and renovation of equipment at the Arusha milk plant. Chronology of Implementation 3.06 A -farm evaluation' mission in June 1976 was followed by a supervision mission in January 1977. By this time, eight groups of parastatal dairy farms had been established (compared to four estimated at appraisal), milk production was in line with expectations, and satisfactory progress had been made in building fences, water reticulation, land clearing and fodder conservation. However, most other components of the Project were experiencing serious problems. The major difficulty was a lack of adequate working capital for which no provision had been made in DAFCO's budget; TRDB, on its part, was refusing to disburse until rights of occupancy were established for each beneficiary farm. The ujamaa component was also not doing well: no projects had been submitted to IDA for approval, neither villagers nor regional livestock staff seemed interested, and regional authorities were using subjective criteria to identify potential project villages. There were already doubts whether, in the absence of understanding of the basic principles of milk production under sound management and the lack of essential inputs and the funds to buy such inputs, the villages were capable of handling investments under the Project. The mission therefore suggested that alternative forms of village dairy development should also be explored. 3.07 Progress on the dairy processing component in Dar es Salaam was also slow: the local contractor had difficulty obtaining cement, the t-wo milk trucks which had been ordered had not yet arrived, and the Dar plant continued to rely on skim milk powder and butter oil purchased from WFP and EEC funds. The planned TRDB positions had been only partially filled and the employment of another livestock specialist was recommended as TRDB's livestock unit was understaffed. The veterinary investigation component had not begun, though there were plans to contact Reading University. A proposal for the heifer breeding unit was rejected as financially unviable. Additional problems concerned the supply of inputs. The national shortage of cement was holding up civil works on the farms, and suitable cattle were found to be in short supply. Some farm machinery had been procured, though clearing, assembly and transport to remote farms were causing delays. The mission was unhappy with DAFCO's accounting system for its farms, but DAFCO insisted that anything more sophisticated was beyond the capacity of local bookkeeping. It was suggested that to improve farm management, the herd size should be broken down into units of 100-125 cows with managers responsible for day-to-day operations of milking, ploughing, cropping, grazing, calf rearing, etc. These units would, in turn, be supported by a group 'service unit- handling technical problems such as artificial insemination (AI), water supply, major repairs, marketing and finance. 3.08 By the time of of the July-August 1977 supervision mission, a number of problems pertaining to the role of parastatals were becoming increasingly apparent. These included coordination among parastatals,  -8- large overhead and administrative costs, and poor relations between DAFCO and individual dairy farm managers. TRDB had stopped disbursing to the parastatal farms because occupancy/title rights had not been obtained for seven out of the 10 areas, involving 12 out of 17 planned farms. TRDB's unwillingness to bend its rules on this matter soured its relations with DAFCO. Delays in payments due to DAFCO by the National Cold Chain Operation (NCCO) also reduced DAFCO's working capital and raised doubts about the marketing prospects at Mbeya, where NCCO had a factory and was the main purchaser of milk. Finally, some parastatal farms were having difficulty in obtaining a regular supply of feed concentrates from the National Milling Corporation (NMC), which was adversely affecting milk production. Though alternative supplements were available in Tanzania, the farm managers did not have an adequate knowledge of the nutrient value of available concentrates and received no assistance in this matter from DAFCO. 3.09 Progress in development of farms was still in line with expectations, but their operating procedures and costs were showing up as problems. The average deficit per farm was TSh 252,000 compared with a surplus of TSh 17,000 per farm expected at appraisal. While the general level of farm managers was satisfactory, there was little on-farm planning and costing of different activities. The mission recommended much more emphasis on training and fewer transfers of staff: in the previous 18 months, only two of the nine farms had the same management they started with. On the ujamaa component, TRDB had circulated guidelines to all its officers and other interested parties at the regional levels. It required that basic infrastructure (i.e dip, handling facilities and fences) should first be supplied by the village and then reimbursed by TRDB. However, the basic commitment to do this was lacking; almost all villages nominated by regional officials were not enthusiastic. 3.10 By March 1978, the ujamaa component had had its first beneficiary, though prospects were still considered poor and TRDB was investigating alternative ways of assisting smallholders. The Mara region had formed a regional and district task force in March 1977, with six ujamaa villages being planned and in their early implementation stage. However, other regional authorities were still inactive. Thirteen of the planned 17 parastatal farms had been started and the dairy development program was more or less on track. Production coefficients continued to improve with fewer mortalities, higher milk production per cow, but low calving figures. DAFCO's overhead costs of TSh 300,000 per farm were, however, imposing an undue burden on farm finances. Lack of occupancy rights and a shortage of working capital had been resolved for DAFCO farms, although some prs.blems still existed for other beneficiaries. TRDB had begun disbursing against the security of DAFCO promissory notes. However, KUC and NAFCO had still not signed promissory notes and TRDB responded by halting disbursements to NAFCO's farm at Bagamoyo and refusing to disburse loans to the KUC farm. 3.11 The March 1978 supervision mission noted two main problems: (a) a shortage of suitable cattle which was adversely affecting production, and (b) a lack of adequate milk pasturizing facilities at Mbeya to handle the excess of production over local demand. Though  -9 - cattle were expected from New Zealand, they could only be used at a higher altitude. Five hundred additional cattle were therefore still needed for the Coast. Procurement of cattle from Kenya posed a problem because of the border closure. The issue arising from Mbeya's surplus production was to decide that if the pasturizing facilities were expanded, who would finance it and who would carry the financial losses as the costs of production (i.e. purchase, pasturizing and transport to Dar es Salaam) were expected to be higher than the fixed wholesale price of milk in Dar. Otherwise, the milk processing and disease control components were progressing satisfactorily, though both had started somewhat behind schedule. Three Tanzanian veterinarians had started training at the University of Reading, while a number of technical positions were being funded by bilaterals. GOT had increased producer prices: the new farmgate prices ranged from TSh 1.50-1.80/liter (versus TSh 1.25 used by the appraisal mission). In fact, although the farmgate price had nominally increased by about 32Z on average, it had actually declined in real terms. 3.12 By the end of 1978, there had been a marked deterioration in Project implementation, with some major problems, including poor financial performance and prospects, inadequate farm management, and a shortage of suitable dairy cows which restricted development on eight of the 15 units. The performance of the only one ujamaa village involved (Isana) was disappointing: cattle purchased were not pregnant and milk production was consequently delayed; also, for want of transport, milk was either being consumed locally or spoiled. The alternative of lending to other smallholders was not being seriously pursued. GOT also appeared to be procrastinating on the contract for disease control. The Dar es Salaam milk collection and processing component was completed somewhat behind schedule. Finnish aid had been arranged for a pasturizing unit at the TDL's Mbeya plant. 3.13 Financial projections indicated that only four of the proposed 17 parastatal farms would have a positive cash flow, while the other 13 were expected to have large sustained negative cash flows. Record keeping was poor, problems were not being followed up, farm managers were generally apathetic and lacking a sense of direction, and two of the General Farm Managers' positions were vacant. Understanding of all aspects of farm management, including grazing management, animal husbandry and production economics, was poor. LIDA's on-farm supervision was inadequate and the overuse of project vehicles by LIDA administration was at the expense of the Project. Total milk sales revenue was expected to be 27% higher than appraisal estimates, but this was more than offset by 91% escalation in total farm costs (mainly overheads, salaries, pasture and crop input costs). Despite higher inputs into pasture crops and concentrates and higher numbers of cows (4,490 vs. 4,272) than estimated at appraisal, the yield per cow milked was lower (8.6 vs 10.1 liters/day) and the total yield per lactation had been reduced from 3,032 liters to 2,487 liters. LIDA recommended a series of remedies including an increase in the price of milk and a provision for Government subsidies. The supervision mission however argued for a much more concerted effort to reduce the operating costs of the farms and bring them closer to cost levels of similar operations in Kenya. It also felt that the rate of development (i.e. building, fencing and water supply)  - 10 - should be scaled down to enable the herd size to build up rather than purchasing more cattle. 3.14 The May 1979 supervision missiun came close to suggesting that the Project be abandoned. The financial prospects for large-scale parastatals looked bleak: in 1978, DAFCO and its eight farms had accumulated a total deficit of TSh 3.3 million. GOT's decision to turn the Kitulo plateau into a major milk producing center under a new parastatal threatened to take away DAFCO's two most profitable units, thereby ensuring its bankruptcy. TSA and KUC also faced serious liquidity problems. Both technical parameters and financial status had further deteriorated on the large-scale farms, with low production, high costs and subsequent financial difficulties. The mission observed declining milk yields, despite high concentrate feeding and genetically improved cattle, unacceptably high average mortalities, poor grazing management, and a high percentage of cows which were not in calf. It was estimated that 300 females had -died' in 1978 inspite of available full-time veterinarians. The possibility that mismanagement and unauthorized diversion of resources (such as theft of feed and milk) were at least as widespread as poor farm management made the mission somewhat hostile to the Project Coordination Committee's suggestion of solving the problem by increasing the milk price, injecting Government equity, and restructuring the parastatal farm loans. While not opposed to an overall increase in milk prices, it felt that the priority should be to correct gross production inefficiencies: lowering on-farm operating costs and tightening up on farm management and controls. There was evidence of substantial overstaffing, over-use of concentrates on some farms, and lack of controls over vehicle use and farm office overheads. The mission recommended cancelling the uncommitted funds if GOT insisted on creating another parastatal. The Isana ujamaa was experiencing operating difficulties: 7 of the 25 cattle had died, milk yields were low, cattle prematurely dry, transporting milk to sell in Mbeya was difficult, most milk was consumed without charge in the village, and overstocking was becoming a problem. The mission approved a TRDB plan for individual smallholder loans. The agreement with Reading University to conduct the veterinary disease control investigation was finally signed and work was expected to begin in July 1979. 3.15 The December 1979 supervision mission felt that, while financial prospects for the parastatal farms remained poor, there was an improving trend due to stricter budget cost controls. There was some improvement in the physical parameters and production costs per liter of milk. However, there were still 10 fewer cows, a 20X lower conception rate, and a 67% shorter lactation period than had been expected earlier. The mission pointed out that if the same number of cows had achieved a 77% calving rate (instead of 61%), a 300 day lactation period (instead of 283 days) and a daily milk yield per cow of 8.7 liters (instead of 7.3 liters), total production would have been 8 million liters worth TSh 16 million; the extra TSh 3 million would have solved the farms' liquidity problems. There was thus no technical, genetic or environmental reason why the appraisal estimates could not be achieved with good management. The mission also recommended that more attention be paid to pasture control and fodder conservation, and that bulls be used as a backup to Al. GOT planned to review the milk price situation and was seeking ways  - 11 - to provide adequate working capital through equity financing and the restructuring of loans, so as to get around the vicious circle of diminishing production and income ensuing from a lack of funds for essentials such as fertiliser and feedstuffs. The mission recommended a capital grant to each farm to lower its debt burden and noted that, adjusted for inflation since appraisal, the farmgate price of milk should "ave been TSh 2.10/liter (instead of TSh 1.80). GOT also requested a reduction in ujamaa village lending and an increase in lending for smallholder dairying. At this stage, TRDB had approved six loans to ujamaa villages but disbursed to only two of them, as there was a constraint on the availability of heifers. It had already approved loans to four smallholders and another eight were under study. 3.16 The May 1980 supervision mission reported a persistence of most of the above-noted problems. Contrary to Section 4.02 of the Project Agreement, no acceptable audited accounts had been received from DAFCO or any other project beneficiary. Because of structural changes in the Ministry of Agriculture, the responsibility for the Project had been changed to the Ministry of Livestock and Natural Resources; this had delayed consideration of issues in farm management and milk pricing. A milk price review was long overdue. With the milk price stagnant at TSh 1.80 per liter since 1976, all farms were running at a deficit, notwithstanding decreases in production costs per liter from TSh 3.88 in 1978 to TSh 2.43. Although the overall volume of milk production had increased, a large proportion was fed to calves rather than sold. Milk production reported on all parastatal farms had increased from 3.0 million liters in 1976 to 6.4 million liters in 1980. The mission felt that, notwithstanding the managerial, technical and financial inefficiencies of the parastatals, milk production could further increase if the farms were fully stocked and substituted high yielding cows for lower yielders. 3.17 There were still only six villages under the ujamaa component Isana, and five in Mara district - with 65X of disbursements completed. In addition to the loans, all villages had received grants of TSh 50,000 each to help initiate work and boost their equity in the dairy enterprise. The Mara villages did not seem overly keen, since most owned cattle on an individual basis, extracted only a liter a day from selected cows, and did not wish to lose ownership or incur debts to embark on cooperative ventures. Isana was the only -truly ujamaa operation". However, while village cows produced only three liters a day, individual cattle produced up to 12 liters. Poor water development, weak veterinary and feed supply services, shortage of qualified dairy workers and of suitable cattle, and lack of transport for Ministry technicians were additional constraints. 3.18 At this stage, it was realized that no spectacular increases in commercial milk production could be expected from either the parastatal sector or the ujamaas, and the mission argued for a long-term strategy to support the smallholder sector. TRDB had so far approved 55 smallholder loans in the Arusha/Kilimanjaro area. However, only 28 out of 70 cows had been procured and TRDB felt that the Project had reached its limit in the area. Shortages of suitable cows, poor communication and lack of support services such as extension, AI and veterinary assistance were  - 12 - uajor constraints holding up development of the smallholder dairy sector. The milk collection component at the Dar es Salaam and Arusha plants had been dropped and funds were reallocated to install an additional yoghurt making unit and to order equipment, two trucks and 400 additional milk cans. A GOT request to reallocate some of the Credit for the relocated Hbeya milk plant was held up, pending a feasibility study. An interim report on the animal health study had been received and recommended against establishing a disease-free zone. 3.19 The November 1980 supervision mission supported GOT's request to extend the Credit Closing Date by two years to March 31, 1983. TRDB interest arrears had improved to TSh 940,000 from TSh 1.6 million at the end of 1979. The July 1980 increase in farmgate milk prices from TSh 1.80 to TSh 2.70 per liter was estimated to boost farm incomes by 67Z, though operating costs were also soaring. During 1980, GOT had also injected TSh 12.8 million of WFP funds into parastatal farms. Apart from Ngerengere, the DAFCO farms showed prospects of breaking even in 1981 and could have done even better with additional cow and heifer purchases. However, although three problem farms among the parastatals had been identified, nothing had been done to remedy the adverse financial and technical situation on these farms. The mission advised strict controls over farm labor and vehicle operational and maintenance expenses, elimination of low-yielding cows and breeders, disposal of most bull calves, and adequate budgetary provision for fertiliser for pasture and bought feeds for high yielding cows. 3.20 NARCO had objected to the last three supervision missions' suggestion that its ranches be used for the production of heifers for Project farms, on the ground that its own ranches were still understocked. However, three heifer breeding units had been established with bilateral funding capable of producing 2,000 heifers a year at full development. All items for the new TDL operations at Arusha and Dar es Salaam had been purchased and installed. The proposal to establish a new milk plant at Mbeya was still under consideration but GOT was now seeking bilateral financing for this and requested use of Project funds for the replacement and renovation of worn-out plant and equipment at the Arusha milk plant. A further Government reshuffle removing the Project from the Ministry of Livestock and Natural Resources and placing it with the Ministry of Animal Husbandry had resulted in many critical delays in the endorsement of the reallocation of Credit proceeds, increased GOT equity financing, higher milk prices, and a host of other decisions. The consequently weak and delayed decision making was manifested in the slow rate of disbursements: as of October 31, 1980, only US$ 4.2 million of the Credit had been disbursed. 3.21 The July 1981 supervision mission had more cause for optimism about the status of the Project as farmgate milk prices had just been raised to TSh 3.70/liter thus improving the parastatals' cash flow position. WFP had provided further equity financing, the problem farms identified earlier had been eliminated from the Project, and there had been some improvement in parastatal farm management. However, DAFCO and TDL audited accounts had still not been received and the audited accounts of other beneficiaries (KUC, TSA, NAFCO) were virtually useless as their dairy activities were lost in their overall budgets. Moreover, since the  - 13 - real producer price of milk had actually decreased since appraisal and was 2-3 times lower than the price in the parallel market, the most recent price increase was unlikely to provide much of an incentive for marketing through the formal system of TDL. Thus, there remained a question as to how much the Project had really improved since its worst low of 1978 and how sustainable were the improvements given the utterly dire state of the Tanzanian economy. 3.22 TRDB had so far approved 40 loans to smallholders for 60 cows and disbursed TSh 179,000 (out of TSh 366,000) for 34 cows. It was finding the exercise costly. There was no assistance from the local extension staff. AI and veterinary services were also scarce as were reliable sources of supplementary feedstuffs. TRDB had already d[sburSed TSh 4.5 million for the Dar es Salaam milk plant and the balance of rSR 5- million was being used for the installation of recently arrived equipment. The shortage of tetrapak material was limiting the plant's utilization to only 50% of its full capacity of 90,000 liters/day. GOT's proposal to renovate the Arusha milk plant was judged to be feasible. A report was received on the animal health study and it was agreed to further extend Reading University's contract by one year. 3.23 The January 1982 supervision mission noted that average milk per cow had fallen from 7.8 liters to 6.9 liters/day. The new farm managers appointed in 1981 had made some improvements but were still not using their records for decision making. Unfortunately, even when farm and project accounts were well kept, there was no standardized format and hence the results were not comparable - furthermore, the Tanzanian Audit Corporation continued not to produce timely audits of project accounts as it was required to do. Improvements could have also be made in the technical aspects of pasture improvement, maintenance and dry season fodder conservation. Slow procurement and a centralization of decision making in LIDA, DAFCO and other parastatals were continuing to hinder day-to-day management and long range planning by farm managers. Government policy on milk marketing and its insistence on retaining male calves, overstaffing and limited powers to hire and fire, and lack of incentives for better productive performance also reduced the efficiency of the parastatal farms. Farm managers complained of being forced to attend political rallies and much time was taken up in court over staff litigation. The employment of new group farm and farm managers had led to suppression of certain abuses and significant rises in milk yields on some farms. However, some farm managers were still selling milk on the open/parallel market and generally attempting to improve their cash flow by not conforming to GOT policy that all milk be sold to TDL at the official price. This tendency was also encouraged by TDL's reluctance and/or inability at times to collect milk. DAFCO farms as a group still generated a surplus of only TSh 604,000 excluding debt service, even though the increased milk price had been expected to increase revenues by about TSh 2.6 million. TRDB was advised by the mission to conduct more frequent field supervision of borrowers, especially those in arrears. However, it had not yet hired adequate technical staff to do this effectively or to pursue the development of smallholder dairying. The Coordination Committee still met infrequently and the overall leadership of LIDA and TRDB left a great deal to be desired. There had been a general reluctance to use Credit funds for technical assistance, relyirg instead on FAO or bilaterally financed staff who had the drawback of rapid turnover.  - 14 - 3.24 The July/August 1982 supervision mission did not give a particularly promising view of the Project. Milk output in 1982 was likely to be even lower than the 1981 production of 5.8 million liters due to continuing management and technical problems. These included fewer cows in milk, extremely high apparent calf mortality, and lack of fuel and spare parts to operate machinery for fodder conservation, which had held down the stocking rate to 60X of the appraisal expectation. Farms were ill-prepared for the forthcoming dry season as they were overstocked relative to feed/fodder supplies. Reduced stocking through heavy culling of unproductive animals was therefore necessary, especially given the unreliable supply of concentrates which otherwise could have been used to sustain milk prod-ction during emergencies. Poor herd management on farms was manifested in slow breeding, high death rates, theft and other diversions of resources. The increase of the official price of milk to TSh 6.20/liter in July 1982 (the first increase which put the real price back at the 1975 level) could have offset financial losses only if costs had been held down and tight financial controls imposed. Procurement of cattle, fuel and spare parts was falling further behind, due in part to the general suspension of IDA disbursements to Tanzania in June 1982 and also to TRDB's reluctance to finance investments on farms with debt servicing problems. The mission suggested giving incentives to managers under DAFCO and LIDA, authority to match their responsibilities, and required adherence to internal controls and reporting procedures. Only one ujamaa farm out of six was doing well, while smallholder development was being held back by a shortage of dairy animals. 3.25 The parastatal farms' financial records were so poor and questionable that an independent financial analyst had earlier (March 1982) been hired at IDA's insistence to report on the situation and conduct a 'performance' audit of the farms' operations. His report, which was received at RMEA in September 1982, concluded that mismanagement, poor farm management and low milk prices had all contributed to high farm losses and relatively small overall increases in milk production and sales. The financial analyst's recommendations covered budgetary and cost controls, deterrents against various forms of fraud and unauthorized sales of animals and milk, inventory controls, disposal of unproductive animals, excessive labor costs, overheads, regular and uniform financial reporting including timely preparation of accurate farm income and expense statements, timely payment for milk by TDL, improved debt servicing plans for the future, and less reliance on expatriate farm managers. These recommendations were fully supported by IDA. The financial analyst was subsequently retained for several mcnths to implement his recommendations and to train farm accountants and DAFCO's principal accountant (also see paras. 5.02 - 5.05 'later). 3.26 The final supervision mission, in January 1983, still classified the Project as having major problems, though with an improving trend. A further extension of the Credit Closing Date was not -ecommended and the mission actively discouraged proposals to prepare a second phase of the Project. Audits had not been brought up to date, and 1981 and 1982 accounts of the dairy parastatals still needed to be sent to IDA. Continuing technical problems included an inadequate number of breeding animals available to the Project for herd expansion, lack of pasture improvement, and mediocre husbandry practices. The estimated total milk  - 15 - production for 1982 was 6.139 million litres, 5% higher than earlier forecast and up from 5.8 million liters in 1981, as shortages in fuel hind spare parts for fodder conservation were temporarily overcome. Production was also boosted by the importation of 250 in-calf heifers from Zimbabwe. The number of cows in milk increased. Reported daily milk yields/cow on six of the 11 parastatal farms also went up, due to better controls and/or better reporting. The results of the 1980 audit and the financial analyst's report (see above) had apparently raised the awareness of DAFCO, TSA and KUC management of the need to hold down costs and impose stricter financial controls at the farm level. However, in spite of the greater volume of milk output, an increase in the milk price and sale of excess cattle, debt servicing to TRDB was still in arrears for six out of 11 farms. TRDB performance of debt management was quite inadequate, it had no data on the aging of debts in arrears, and its accounts on debt servicing still required reconciliation with those of the parastatals. 3.27 The 11 parastatal farms as a group expected to earn a small profit of TSh 500,000-800,000 in 1982 1/, after consecutive losses in all previous years, including a loss of TSh 8.8 million in 1981. Sales of cattle were estimated to have increased from TSh 804,000 to TSh 4.2 million. Operating costs were about TSh 25 million, or about TSh 1 million less than in 1981. However, even with the forthcoming acquisition of 1,000 heifers, the mission felt that neither the stocking rate nor milk production would reach appraisal targets. Consequently, project benefits would be reduced and the rate of return was expected to be lower than estimates or even negative. The mission estimated that if the parastatals had been fully stocked under the actual (versus the appraisal) parameters, milk production in 1982 could have been 9.7 million liters instead of 6.1 million liters. However, though the technical qualifications of farm managers were uniformly high, their technical performances were highly variable. A particular concern was the high percentage of dry or non-productive cows to those in milk. Owing to understocking and low toLal milk production, farms tried to maintain maximum milk output by retaining low yielding cows. Ideally, the milking/dry cow ratio, based on a 300 day lactation and 100% calving rate, is 82.2%. In both 1981 and 1982 the average for all parastatal farms was 50.4Z (with a range of 39Z-71Z). The mission felt that with a better use of records and culling of slow breeders, a rate of 66% could be attained (with 80% calving). Issues in Implementation 3.28 The following were the main issues which arose in the implementation process; (a) Milk Prices and Marketing. A critical assumption underlying the Project was that the price of milk would be raised sufficiently to stimulate greater commercial production of milk, thereby also justifying 1/ Profit and loss statements for parastatal dairy farms for 1982 and 1983 are still not available. Hence, these figures are only estimates.  - 16 - the use of project funds for improving the collection and distribution facilities in Dar es salaam and Arusha. In fact, as shown in Table 7, official increases in milk prices until 1982 were always too little and too late to maintain the price in real terms even at t.'e 1975 level; the average decline in milk prices in real terms during 1975-82 was 21%. This adversely affected the profitability and cash flow situation of the parastatals: some half of the shortfall in revenue in 1981 as compared with appraisal estimates was due to this factor (the remainder being due to lower production levels and higher operational costs). Had the milk price been adjusted in real terms, DAFCO would have earned an additional TSh 13.6 million in 1976-81, which would have partially offset its losses totalling TSh 25.1 million over the same period. Low milk prices also reduced the incentive for both parastatal and other producers to sell milk legally, particularly in the generally deteriorating economic environment in Tanzania with growing shortages and emergence of a large parallel market for food items such as milk (with 2-3 times higher prices). Moreover, TDL was not always a reliable buyer or a timely payer, further decreasing the incentive to sell legally. Throughout implementation, supervision missions thus found themselves having to address the problem of illegal milk sales. In fact, the only parastatal farm to make a profit in 1981 sold part of its output on the parallel market. Profitability could apparently be best achieved only at the expense of the Project's objectives of import substitution and increased milk supply to urban areas. Since the formal market sector was unable to attract major portions of the milk produced, improvements under the Project's milk collection and processing component also became somewhat less significant. (b) Role of Parastatals. The number of parastatal institutions involved in the Project generated severe problems of coordination, which were accentuated by the widely dispersed location of many parastatal farms far from good roads and supporting services. The idea that parastatals might actually impede good farm management and unnecessarily add to overhead costs (as with other crop-buying parastatals in Tanzania operating at a loss) was apparently not fully appreciated at appraisal. In the event, decision making by Project parastatals was unusually cumbersome, thus slowing down all levels of implementation and reducing the initiative of individual farm managers. Spare parts could not be obtained without going through LIDA, and regularizing the supply of concentrates depended on NMC. The conflict between IfDA as the Project's technical arm and TRDB as its financial (and somewkat restrictive) arm continued to hinder achievement of the Project's aims throughout the implementation period. No project management unit was set up as originally envisaged. This left the work of running the Project with LIDA's overtaxed Director and the Coordination Committee, which did not meet regularly or take the initiative expected of it. Implementation was also affected by the lack of continuity in Group Farm Managers, in part the result of two-year contracts under bilateral aid. GOT's decision to transfer the Project midstream to another Ministry also decreased the momentum of implementation. There was, in addition, an unsuccessful attempt to create a new parastatal for another GOT project on the Kitulo plateau. This tendency to -solve' problems by rearranging the administration or creating new organizations only resulted in multiplying  - 17 - existing problems, generating new ones, and adding to overhead costs of operation. (c) Farm Management Performance. The appraisal mission had correctly stressed the highly technical and managerially demanding features of dairy development: this was the reason GOT's original project proposal was substantially scaled down. It was probably still optimistic, however, to assume that the gap in critical management skills could be filled easily and quickly enough during the course of Project implementation to ensure well run and economically viable dairy farms. By 1981-82, there were apparently sufficient numbers of technically well-trained (though inexperienced) farm managers, but supervision missions continued to report a lack of attention to detail and a lack of motivation to improve operating performance. Farm records were poorly kept and not used to diagnose problems, cows were not always properly fed or monitored, lax controls created opportunities for theft and waste, machinery was allowed to run down, and so on. Since farm plans had to be prepared and approved before ordering machinery, there was a 9-12 month delay in farm development (in addition to 6-9 months needed for tendering, ordering and delivery): these delays were not anticipated in the original Credit disbursement schedule. The effective annual calving rate during the Project was also well below expected levels. In 1981, only 63% of mature cows were milking. This low performance (which stemmed from low submission rates of cows to mating) slowed herd build-up, the number of lactations and total milk production. Supervision missions constantly advised that bulls be used as a back-up to the AI service to improve calving; however, this appears not to have been done systematically. The extremely high apparent death rates of young stock and adults (22Z and 9.5% respectively in 1981) also contributed to a slow build-up of herd numbers. Poor calf rearing management (exarcerbated by the GOT requirement that all bull calves should be reared for eventual sales to smallholders and ujamaas), inadequate pasture grazing, too few veterinary diagnosis facilities, and shortages of health supplies were among the factors cited as causing this problem. Finally, staff/cow ratios on farms were some three times appraisal estimates. Due to poor motivation, inadequate supervision and an inability to fire workers, labor productivity was particularly low among semi-skilled workers. In addition, outside pressures forced farm managers to be away from their farms much of the time. As DAFCO's head office expenses were levied against individual farms, this also hindered the parastatal dairy farms from operating as independent profit units as anticipated at appraisal. (d) Availability of Cattle. The appraisal mission had assumed, incorrectly as it turned out, that sufficient cattle would be available within Tanzania to stock the parastatal dairy farms, and also did not make any provision for pasture development on these farms. It was assumed that the West Lake heifer breeding unit would produce about 1,800 upgraded heifers a year. However, there was in fact insufficient lead time for this unit to contribute significantly to the stocking of Project farms. Soon after Project start-up, the unit was abandoned as financially unviable; as it was located far from most of the parastatal dairy farms and ill served by transport, it was in any case not sited optimally from the Project's point of view. Bilateral aid financed three  - 18 - heifer breeding units in the late 1970s, but heifers from these units did not become available until 1981. As of 1977-78, the large-scale dairy farms were short of approximately 2,000 cows. Some cows from the U.S.A. were provided for Kitulo in 1977, and 2,517 additional cattle were imported under the Credit from New Zealand, Kenya and Zimbabwe during 1977-83. In addition to the inadequate availability of cattle, the lack of initial attention to pasture development left the Project unnecessarily vulnerable to the supply of feed concentrates. This again contributed to the slow herd build-up and poor operating performance of the parastatal farms. (e) Role of Ujamaas. The ujamaa component never really took off during the Project because it was unviable from its inception. At the time of appraisal, there were no real 'ujamaaC villages. Communal production had never been a common feature of village life in Tanzania. Instead, the population was being relocated by the Government into "development" villages, thereby causing a major disruption of the agricultural sector. Peasants for the most part were tired of official intrusions, which brought low prices, poor marketing, and debts. Also, few villages could meet both the loan and other criteria for dairy projects, such as previous experience with livestock, suitability of location for dairying, capacity of regional livestock officials to provide extension and veterinary services, and alternative sources of cash income to pay TRDB charges. The few villages which did meet these criteria did not perform very well due to lack of genuine commitment by villagers and extension staff, shortage of qualified manpower and training facilities, difficulties in obtaining and transporting cattle, and unavailable or declining infrastructure such as water supply. In fact, it is questionable whether TRDB would have considered the funding of village dairies as viable if WFP had not given a grant to finance half of the investment cost. While the appraisal mission had not fully appreciated the impact of villagization on the proposed development of dairy production, the early supervision missions also continued to believe for too long that this component could be successfully implemented. When ujamaa were finally given up as a lost cause, the emphasis was switched to smallholder lending. (f) Smallholder Dairy Schemes. By the end of 1983, 100Z of projected loans to smallholders had been approved and close to 100% disbursed, with most recipients concentrated in the Arusha/Kilimanjaro area. However, their operating performance and impact have so far been limited by the general economic constraints and the non-availability of inputs and technical advice. Traditional husbandry practices are inadequate to keep highbred dairy and there have been no means of providing training for prospective borrowers. TRDB was also unable to collect any technical or financial records from smallholders, and did not hire adequate technical staff to appraise or supervise loans to smallholders. Livestock extension staff lacked transport and other services necessary to support the individual dairy farmer. Concentrates, medicines, inputs and breeding bulls have been in short supply, leading to prolonged calving intervals and dry non-pregnant cows. While smallholder dairying may ultimately offer a more cost-effective means of increasing the domestic supply of milk than parastatals, it can only succeed if sufficient pasture, clean water, dipping facilities, good roads and transport to supply feeds and other :nputs, nearby veterinary services and  - 19 - good marketing facilities are available. These conditions are closely linked to the general state of the Tanzanian economy. (g) General Economic Environment. The appraisal mission had identified a number of constraints which if improved or minimized would allow Tanzania to develop an indigenous dairy industry to meet its milk requirements. These included tae location of some high potential areas far from major markets, low milk price levels, limited milk collection services, low milk yields of traditional cattle, shortage of feed during the dry season, weak dairy extension services, shortage of inputs, lack of trained dairy managers, etc. In practice, however, the impact of additional funds, improved cattle, inteinationally-recruited manpower and machinery on the dairy system was severely limited by the general deterioration of all services and sectors in Tanzania over the Project implementation period, the full extent of which could not have been reasonably anticipated in 1974. Some of the major constraints on dairy production (such as availability of inputs and transport facilities) were really features of the wider environment and as the economic crisis worsened so did these constraints: it was impossible ;or dairy farm performance to improve in isolation when everything on which it depended was deteriorating. Moreover, the IDA Credit was supposed to primarily finance investments, with GOT paying for most recurrent expenses and capital replacement. The growing general scarcity of foreign exchange caused a continuous shortage of veterinary medicines and spare parts for farm machinery, with consequent delays in farm development work and high rate of animal deaths. Thus, the many shortfalls between appraisal estimates and actual results can be explained partly by severe shortage of foreign exchange available to GOT for the purchase of recurrent inputs and spare parts and, in a larger perspective, the rapid decline of the general economy over the Project implementation period. (h) Financial Controls. There are many indications that loose financial controls and unauthorized diversion of resources substantially affected the profit and loss situation on the parastatal farms and hence delayed or impeded Project implementation. The supervision missions regularly pointed to missing milk, missing feed, misuse of vehicles, nepotism, and poor or doctored accounts. It appears that the more the larger economy deteriorated, the more such malpractices became a feature of the landscape. The question naturally arose as to whether implementation was proceeding better than appeared to be the case, if one could factor oit unofficial diversions of resources. In short, was the problem of the Project technically poor farm management or mismanagement (and manipulation of data)? This question was important in assessing whether or not it was at all possible to implement the Project successfully and profitably. Unfortunately, there is no easy answer to this question, and the effect of these elements on the Project as a whole must remain a conjectural point. IV. Project Impact 4.01 The main quantifiable impact of the Project has been the increased domestic commercial supply of fresh milk from the dairy parastatals. By 1982, commercial milk production on the parastatal farms  - 20 - had risen to around 6 million liters/year; additional quantities being sold illegally on the parallel market are impossible to estimate. The Project was undoubtedly instrumental in turning a number of run-down dairy farms into functioning entities. Nevertheless, the incremental production figures were still less than a third of appraisal targets. Moreover, TDL's reliance on imported milk products actually increased over the Project implementation period, while its intake of fresh milk decreased by 20% (see Table 8). During the same period, imports of milk products rose 53% and of butter oil by 71% (Table 9). The impact of the ujamaa and smallholder components on milk production was also negligible. Overall project impact on import substitution was therefore not nearly as significant as anticipated, and the impact on nutritional levels of the population was limited at best. 4.02 There also remain serious questions about the long-term sustainability of the milk production levels already achieved. Milk output and the ratio of milking cows on DAFCO farms actually dropped in 1983. Much of the equipment is now run down and in need of replacement, for which there is no foreign exchange. Without replacement of this machinery on a regular basis and its maintenance in good working order, the farms will deteriorate and production will fall. It has been estimated that the foreign exchange equivalent of some TSh 5.3 million (in 1982 prices) is needed annually to replace machinery and farm vehicles, maintain the supply of spare parts, import seed for pasture development, and continue sending farm managers on training courses. At present, much of this assistance is being provided by the Netherlands, but there is no assurance of its continuation over the long term. 4.03 An important non-quantifiable benefit of the Project expected at appraisal was the development and demonstration of dairy farm techniques, which were to lay the basis for future dairy development in Tanzania. In fact, farm management left a lot to be desired throughout the Project implementation period. It is therefore difficult to say how much, if any, spillover effect will come from dairy farming techniques that are Project-related. The upgrading of the domestic stock through importation of cattle could well improve deirying over the long run. Perhaps the most important positive impact from the point of view of benefits to individual producers has been the unintended spillover of some upgraded dairy cattle from the parastatal farms to the smallholder sector. When cattle disappeared from the farms, some were slaughtered while others were thought to have gone to interested smallholders. Nevertheless, while the Project as a whole had been expected to produce 2,850 dairy heifers, the need to procure cattle from abroad after the heifer breeding unit was dropped actually resulted in unanticipated costs rather an increased output of local cows. 4.04 One clear positive impact of the Project was to initiate an annual Government review of the milk price. Indirectly, this encouraged systematic analysis and announcement of regular producer price increases. These increases in turn have demonstrated that the demand existed to sustaia higher prices for both the producer and the consumer. It is possible that this could persuade the Government to decontrol milk prices in the future.  - 21 - V. Financial Performance 5.01 The appraisal report had estimated financial rates of return (FRRs) on incremental investments at 18Z for dairy farms, 28Z for the ujamaa dairy units and 33% for the heifer breeding unit. The Project was expected to result in a net cumulative surplus of TSh 13.6 million by year 5 and TSh 123.7 million by year 15, thus adding to the Government's revenues. However, for all the reasons mentioned in earlier paragraphs it is clear that the financial targets and projections made at appraisal will not be achieved. Improvement of processing facilities at Dar es Salaam and Arusha assisted in raising TDL's net profits from TSh 8.9 million in 1977 to TSh 28.5 million in 1982. But this performance was an exception. Calculations by LIDA's Dairy Advisor in 1982 indicated FRRs for the parastatal farms ranging from 2% to 9Z, while its consultant financial analyst estimated a negative FRR of 12% for all parastatal farms combined in 1981. Unfortunately, poor accounts, record-keeping and differences in accounting systems on Individual farms, parent parastatals and other implementing agencies make it impossible to realistically assess the financial performance and FRR on Project investments eith r by farm or for the Project as a whole. Production data is also generally inadequate or inconsistent, and it is not possible to make causal linkages between Project investments and production. Tables 10 and 11 show the latest avallable status of TRDB's loan portfolio and arrears. The last supervision mission in January 1983 recommended that a case-by-case review of TRDB's credit accounts be carried out and a report reconciling these with the parastatals' accounts be submitted to IDA. To date, this report has not been received. 5.02 In 1982, a consultant financial analyst was hired by LIDA at the Bank's request to work in Tanzania for six months, focusing on the causes of the parastatal farms' inability to generate sufficient income to cover costs and their growing arrears in repaying TRDB loans. He found that most of the parastatal farms did not pcssess up-to-date accounting records, and reported systematic and sustained losses by the farms and all four parastatals (NAPCO, DAFCO, KUJC and TSA) since 1976. In 1981 alone, tne parastatal farms lost TSh 8.8 million, or 48% on a turnover of TSh 18.1 million. Only four of the farms were able to service their debts to TRDB and only two of them from internally generated income. Cumulative losses since 1976 had eroded almost half of the farms' investment costs; without substantial capital grants from WPP, the situation would have been even worse. Tables 12 and 13 present an analysis of the parastatal farms' operating costs and income during 1976-81. (As of March 1984, only two parastatal farms had no arrears on loans from TRDE; accumulated arrears for the other farms were TSh 4.97 million. In addition, all but one ujamaa village were in arrears on their loan repayments, and the majority of smallholders were also unable to meet their repayment schedules). 5.03 The financial analyst identified two main reasons for the parastatal farms' poor financial performance: inadequate income, and poor cost performance and structure. In 1981, a decline in both milk output and the selling price of milk in real terms had reduced farm income: milk sales declined 8% with an lIZ drop in yields/cow, 4% reduction in number  - 22 - of cows in milk, and maintenance of a large number (40%) of unproductive cows. The average farmgate milk price of TSh 3.20/liter compared unfavorably with the appraisal price assumption of TSh 5.21/liter (TSh 1.80 expressed in 1982 shillings). Including non-milk income, all farms consequently lost an average of TSh 1.70/liter of milk production in 1981. The operating cost structure of the farms showed a definite bias in favor of unproductive (45%) versus directly productive (28%) expenditures. On DAFCO farms alone, head office costs ranged from 10% to 40% of total farm income, mainly as a result of overstaffing: as just one example, five managerial posts in DAFCO were found to be functional duplicates of those in LIDA. Labor was the highest single cost item, amounting to 21% of total farm operating costs and 24% of the average milk price received (vs. 15-16% standard). Excess labor costs were reflected in high overtime payments and the employment of a large number of casual laborers: both excessive in relation to farm activities and requirements. Vehicle running costs were similarly excessive. A major conclusion of the financial analyst was that the lack of internal stock control and budgetary monitoring were major factors in explaining both the poor income and the high ratio of unproductive to productive costs. The lack of accounting procedures for the issue of stocks and a failure to reconcile inventories with stock purchase and issue had resulted in an accumulated stock shortage for DAFCO farms of TSh 1.6 million which had never been investigated and was finally written off in 1981. 5.04 Only three of the parastatal farms were able to produce up-to-date and accurate accounts for 1981. Hence, income and expenditure accounts for management decisions, particularly cost controls, were non-existent. This made it impossible for the parent parastatals to monitor the farms' financial performance, especially sales of animals and revenue from milk sales to non-TDL outlets. The financial analyst estimated that at least one half of all milk produced by the farms was not sold to TDL but to private buyers at a free market" price higher than the official price. Since there is no data for 1982 and 1983 comparable to that collected by the financial analyst for 1981, it is difficult to say whether the situation has substantially improved in the last two years. 5.05 The financial analyst's recommendations included: (a) budgetary and cost controls, with head offices requiring uniform monthly and quarterly operating statements and setting standard cost coefficients (per animal and per liter of milk) for monitoring performance; (b) inventory controls; (c) a reduction of the permanent labor force; (d) a limit on private sales of milk to 10% and a maximum quota of 100 liters/month to farm employees; (e) sale of surplus male calves and unproductive females; and (f) assessing charges of DAFCO overhead at a flat rate of 11% of total sales per farm. While DAFCO, LIDA, TSA and KUC all agreed to issue written instructions to their parastatals to comply with these recommendations, in the absence of up-to-date statements of accounts for the last two years, it is difficult to gauge how far they have actually been implemented or to what extent parastatal financial management has improved as a result.  - 23 - VI. Institutional Performance and Development 6.01 At the time of appraisal, the quality of management (particularly on farms) was identified as the principal risk of the Project and the pivotal factor on which the economic and financial viability of dairy farms hinged. The Project had no institution-building objectives per se: all the parastatal institutions involved were already in place and the Project did not envisage setting up any new institutional arrangements. The Project basically aimed to turn run-down dairy farms into productive and profitable units. Adequate provision for technical assistance (30 manyears) was made for LIDA and TRDB in the original Project design, including internationally-recruited financial controllers, group farm managers and a training component. In the event, the major part of the Project's technical assistance component came from bilateral sources and the provision for a preparation team in TRDB for ujamaa villages never materialized. Since the technical assistance personnel who were hired could only play an advisory role, they did not have any operational authority and therefore remained ineffective in practice. Three short-term courses of 3-4 months each and one long-term course were also funded under the Project and overseas training was provided to two veterinarians: these can be expected to yield some benefits over the longer term. 6.02 An implicit assumption at appraisal was that the Government had the capacity to absorb the administrative, manpower and managerial demands of the Project, with appropriate strengthening through technical assistance and training. However, certain features of the Project, including the large number and remoteness of dairy farms, made Project management more difficult. This was exacerbated by the centralization of decision-making in the parent parastatals located in Dar es Salaam, the lack of identification of TBDB with the Project objectives, and the increasing difficulties caused by the general deterioration in transport and input supply services. Although the Project had sufficient numbers of technically trained farm managers towards the end of the implementation period, poor farm management performance continued under these constraints. Project management problems could not really be dealt with outside of their wider context of the overall economy. As just two examples of the external policy constraints, low milk prices and soaring administrative overheads of the parent parastatals both contributed to demoralization, poor staff motivation, mismanagement, high non-productive operating costs and poor bookkeeping practices. The institutional impact of the Project was thus severely limited. 6.03 GOT's response to problems in Project implementation and suggestions made by the supervision missions war often disappointing. It did not comply with the clauses in the Credit and Project Agreements requiring audited accounts, did not hold regular meetings of the Coordination Committee, did not seriously investigate charges of mismanagement, and did not produce final 1982 Project accounts by March 31, 1983 or final parastatal accounts by June 30, 1983, as jointly agreed in January 1983. Supervision inputs by 2MEA often went beyond simple advice and bordered on direct management intervention, raising again the issue of the Project's long-term sustainability, even at the  - 24 - managerial level. It was only at IDA's insistence that the Government finally called in a financial analyst (see para 5.02 above) to ascertain the financial and management status of the parastatal dairy farms. Prior to this, the Government had persistently maintained that there were no problems except occasional sloppy accounting procedures, which would be rectified. When this ultimately proved not to be the case, the Government was verbally responsive to all the financial analyst's suggestions including the need for standardized budgetary and accounting procedures. The financial analyst's contract was extended to enable him to implement his recommendations and to train farm accountants. The contract of DAFCO's Chief Accountant was also extended for the express purpose of providing up-to-date 1982 accounts. Nevertheless, as of July 1984, the initial response had not been translated into concrete action either on the parastatal dairy farms or at the general manager'al level: final accounts have not yet been prepared for 1982 and 1983. VII. Bank Performance 7.01 Although the scope of the Project was substantially cut back at appraisal from the original proposal by GOT, in retrospect it appears that it might have had greater chance of success if it had initially concentrated on even fewer (and more accessible) dairy farms and/or pilot smallholder production near urban areas, given that this was Tanzania's first experience with large-scale dairying and trained management was not initially available. Indeed, the major constraints in establishing 600-700 dairy cow units in developed countries with a history of progressive dairy farming, is usually the shortage of capable farm managers. The institutional focus on both parastatals and ujamaas was misplaced, since the former actually impeded good farm management while the latter were a new and untried concept that proved totally unsuitable for dairy management in the Tanzanian context. The appraisal mission had rightly stressed the technically demanding aspects of animal husbandry and dairy development. However, it had apparently not sufficiently researched the local availability of suitable cattle, focused almost exclusively on feed concentrates at the expense of pasture development, and probably set the Project's targets (including production coefficients) in the context of parastatal and ujamaa management which restricted the authority of farm managers in relation to their responsibilities. 7.02 The Project was supervised regularly from RMEA, wih a reasonable degree of staff continuity. Supervision missions consistently identified the technical, management and other problems impeding Project implementation, though there were limits to what they could achieve in the narrow Project context while the general economic environment and infrastructure were rapidly deteriorating. In retrospect, the inclusion of a financial analyst on some of the earlier supervision missions (one was included, in 1980-81, on only two of the 13 supervision missions during the seven years of Project implementation) and enforcement of the Project's reporting requirements might have forced the issue of poor accounting and financial performance to the forefront of management attention much earlier than 1981, when matters had really gotten out of hand. Although a lack of suitable milking cattle was continually  - 25 - identified a serious problem, the decision to go ahead with the importation of 1000 dairy animals in the closing months of the Project could be questioned in the light of poor management and the prevailing economic conditions. More attention might also have been given to the training of key project staff from the beginning. Time was probably also lost in pushing the ujamaa component too long and too hard, when it was obviously beyond redemption. 7.03 Disappointment has been expressed by GOT at the Bank's general discouragement of a Phase II Dairy Project. Arguments have been made that the dairy parastatals have done "no worse' than other parastatals which the Bank is continuing to finance, and that as the farm infrastructure is now in place, cattle have been procured and there is a greater seriousness concerning animal husbandry, a follow-up project could be used to consolidate what already exists. There still remains the question, however, of how far the implementation of any particular project in the Tanzanian context can be isolated from the general policy and economic environment in which it operates. VIII Economic Re-Evaluation 8.01 The appraisal report had estimated the economic rate of return (ERR) on Project investments at 18% over 25 years, based upon an increase in average milk yields from 1,900 liters to 3,100 litres/cow and in annual commercial milk production from 10 million liters to 20 million liters by 1980 and 25 million liters by 1987. Better nutrition, disease control, management and genetic quality of cows were to be used to improve on the 1974 figures of a low calving rate of 50%, calf and adult mortalities of 10% and 20% respectively, and a low slaughter weight of 250 kg. 8.02 As documented in earlier sections of this PCR, none of the above targets was met in implementation of the Project. Management of parastatal farms continued to be weak, the ujamaa dairy component failed to take off, and the heifer breeding unit was shut down. Unfortunately, the lack of up-to-date and consistent accounts (see para 5.01) makes it impossible to obtain an accurate picture of the operations of individual farms or their parent parastatals (including WFP contributions). It is therefore not possible to re-evaluate the ERRs either on individual components or on Project investments as a whole from the available data. Only two conclusions can be drawn at this stage: (i) the ERRs on investments on parastatal farms (and also on the Project as a whole) are likely to be somewhat higher than the FRRs to the farms, given the substantive evidence of significant unofficial milk sales (probably at least 20-30Z of total production) and diversion of improved heifers to the private sector that were never documented; and (ii) probably the only part of the Project on which the ERR is l