Malawi - Shire Valley Agricultural Development Project

CIRCULATING COPY TO BE RETURNED TO REPORTS DESK Report No. 895 CONFIDENTIAL INTERNATIONAL BANK FOR RECONSTRUCTION AND DEVELOPMENT INTERNATIONAL DEVELOPMENT ASSOCIATION Project Performance Audit Report MALAWI SHIRE VALLEY AGRICULTURAL DEVELOPMENT PROJECT PHASE I (Credit 114-MAI) October 22, 1975 Operations Evaluation Department  PREFACE This report presents an audit of accomplishments under IDA Credit 114-MAI, for US$3.7 million, which was signed in February 1968 and closed, fully disbursed, in March 1974. The audit was based on information contained in the Project Completion Report (PCR) as prepared by staff of the Eastern Africa Region, on information contained in Bank files, and on discussions with Bank personnel in Washington and Nairobi and with Government officers in Blantyre and the Shire Valley. The cooperation of the Government at all locations is gratefully acknowledged.  PROJECT DATA Credit 114-MAI Amount of Credit US$3.7 million Amount Disbursed US$3.7 million Date of Credit Agreement February 5, 1968 Date of Effectiveness March 15, 1968 Original Closing Date December 31, 1973 Final Closing Date March 31, 1974 PCR Issue Date September 1974 Repeater Project Number Credit 363-MAI Amount of Credit US$10.5 million Date of Credit Agreement March 28, 1973 Land Measure Square Kilometer= 247 acres Exchange Rates (Kwacha) 1 Kwacha = 100 Tambala = 0.5 MA 1967-1970: 1 US$ = .83K 1 K = US$1.2 1971-1974: 1 K = varying rate between US$1.30 and US$1.17 Conversion Rate Used: 1 K = US$1.2 (unless otherwise stated) Abbreviations Used, PPAR - Project Performance Audit Report SVADP - Shire Valley Agricultural Development Project ARI - Appraisal Report SVADP Phase I ARII - Appraisal Report SVADP Phase II PCRI - Project Completion Report Phase I ADMARC - Agricultural Development and Marketing Corporation SBF - Sprayer Borrowing Farmer SOF - Sprayer Owning Farmer  TABLE OF CONTENTS Page Summary i-ii I. Introduction 1 II. Conception of the Project and the Appraisal Report A. Conception 2 B. Critical Assessment of the Appraisal Report 3 III. Implementation of the Project A. Organization 5 B. Disbursements, Credit Allocations and Project Costs 7 C. Development Work 8 IV. Impact of the Project A. General 12 B. Production 12 C. Farm Budgets 13 D. Equity 16 E. National Economy 17 F. Government Budget 18 V. The Bank's Role 19 VI. Concluding Remarks 23 Annex 1 - Sensitivity Analysis of ARI Projections 25 Annex 2 - Organization Chart of the SVADP 28 Annex 3 - Project Finances 29 Annex 4 - Physical Execution of the Project 32 Annex 5 - Projected and Actual Cotton Production 34 Annex 6 - Cotton Prices 36 Annex 7 - Value of Incremental Cotton Production 38 Annex 8 - Farm Budgets 39 Annex 9 - Social Cost Benefit Analysis 44 Annex 10 - Fiscal Analysis 47 Annex 11 - Development of Cotton Production in the Shire 50 Valley According to Different Bank Reports Annex 12 - Rate of Return Calculation, Phase II 54 Map  Project Performance Audit Report MALAWI SHIRE VALLEY AGRICULTURAL DEVELOPMENT PROJECT PHASE I (Credit 114-MAI) SUMMARY S.01 The Bank is giving financial support in Malawi, one of its poorest member countries, to three large regional agricultural programs, including the Shire Valley Agricultural Development Project (SVADP). In phase I, which is the subject of this audit report, it covered an area of just over 2,000 sq. km with some 17,000 smallholder families. Its principal objective was to induce farmers to adopt improved agricultural practices, in particular the spraying of cotton, through the provision of extension, credit, marketing facilities and. infrastructure. The project was supported with IDA Credit 114-MAI, signed in February 1968, amounting to US$3.7 million and covering 80% of total project costs. S.02 An analysis of the conception and plan of the project presented in the Appraisal Report suggests that the major project component, the intro- duction of the spraying technology into the Shire Valley - a traditional cotton area - was sound but that the other activities foreseen (settlement, maize improvement) lacked a realistic chance of success. In this respect, and by also assuming ever-increasing benefits for 25 years due to costless demonstration effects, the Appraisal Report's forecast of a rate of return of 21% was overly optimistic. A more realistic forecast would have put the rate of return between 10% and 15%. S.03 Implementation of the project benefitted from a strong expatriate staff in the initial phase. In the course of five years of the project most of the top positions were handed over to Malawians. S.04 Disbursements went according to schedule. Physical construction was satisfactory. Extension density was increased to one demonstrator per 150 farmers. Farmer loans were provided for the purchase of sprayers and insecticides. On average, repayment rates of well over 90% were achieved. The credit was closed, fully disbursed, in March 1974. S.05 Considerable savings compared to the appraisal projections were effected due, on the one hand, to wise purchasing and tendering procedures and general efficiency on the part of the management and, on the other, to a rather lavish financial schedule prepared at appraisal (attributable to a contingency allowance of 15% and credit requirements calculated by mistake on a cumulative basis instead of a revolving fund). Theseisavings were, with the c6nsent of 'the Bank, used to construct health facilities and to finance part of the second phase of the project. S.06 The project can:-be.credited with a general mobilization effect in a populace and an institutional setting that have been considered resistant to change and in an area which is described as the most disadvantaged in Malawi. - ii - S.07 Total cotton production in the project area developed erratically over the five project years due to significant weather-induced yield variations. Incremental cotton production followed projections closely up to year four, when over 4,000 short tons of seed cotton attributable to the project were delivered. In year five, however, production reached only one-third the expected level due to bad weather. Project management believes that (1) yields will stabilize well above traditional levels in unsprayed fields but below appraisal forecasts, (2) the area sprayed will fall considerably short of forecasts, and (3) the production attributable to phase I efforts would in the long run not significantly exceed 4,000 short tons instead of the 11,000 short tons forecast. The prospects of continued costless spread attributable to the phase I demonstration effect are thought to be low. S.08 The project provided to the farmers a way of increasing income by about 50%, though the new level remains below the national average and the return to a day's labor hardly improved at all. This means that alternative employment opportunities, such as offered by the expanding Sucoma Estate (irrigated sugar cane), would be more lucrative to the farmers. Not taking into account price increases, and excluding as project costs items unrelated directly to production (like the health facilities), the economic rate of return of the project is 15%, which is to no small extent attributable to the exceptional efficiency of management and to the savings that were thus effected. Taking into account the increased cotton prices on the world market, the internal rate of return goes above 25%. Considerable cost escalation could be borne before the rate of return were depressed below 15%. At full operation the project contributes over K 630,000 per annum of foreign exchange. S.09 It is typical for rural development projects involving large numbers of smallholders and social investments that the majority of the benefits generated stays with the rural population. Thus, in the case of the SVADP, the Government will probably not recover more than a minor part of its ex- penditures. This appears dangerous for small countries like Malawi with several rural development projects, since, regardless of the economic rate of return, these projects do not contribute sufficiently to the government capacity to finance development and to meet repayment obligations. S.10 ' Of most concern is the way in which out of a fairly modest but sound phase I project, a complex phase II project was developed. Based on overly optimistic benefit projections, a rate of return to phase II of 22% was predicted. A more realistic calculation of phase I and phase II strung together would have produced a rate of return of 12%. Apart from these arithmetics, there remains the fundamental question of whether phase I in its area has not already shown the long-term limits to the development of rainfed agriculture in the whole of the Shire Valley. S.11 Phase I of the SVADP can be called a good project, well-conceived and well-implemented. The fact that phase I, one of the Bank's first small- holder projects and situated in a desperate area, worked as well as it did, should encourage those who believe that no rural condition is hopeless. Project Performance Audit Report MALAWI SHIRE VALLEY AGRICULTURAL DEVELOPMENT PROJECT PHASE I (Credit 114-MAI) I. INTRODUCTION 1.01 Malawi became independenp in 1964. The total land area of the country is 94,000 square kilometers (roughly the size of Portugal or Jordan). In mid-1972 the population was estimated at 4.7 million, growing at a rate of about 2.5%. The population density of 49 per sq. km is considerably higher than in the surrounding areas. Some 90% of the population is engaged in agriculture. Slightly over 50% of the national income is still generated by agriculture (30% by subsistence agriculture). The per capita GDP in 1972 was estimated at US$87. Anrt from the lake and the agricultural land there are no important natural resources known. Malawi is one of the poorest member countries of the Bank. 1.02 The Bank is financing three large agricultural development projects in Malawi, including the Shire Valley Agricultural Development Project. The others are at Lilongwe (1968) and at Karonga (1972). In its phase I (1968- 1973) SVADP covered an area of 2,150 sq. km with some 16,500 smallholder families. Its principal objective was to induce farmers to adopt improved agricultural practices, particularly in cotton. 1.03 The Shire Valley is one of Malawi's most disadvantaged areas in terms of climate, disease incidence and agricultural productivity. The government of newly independent Malawi considered agricultural development in general and development of the Shire Valley in particular to be of highest importance and urgency. 1.04 In the mid-sixties, while an FAO/UNDP survey of long-term agricul- tural investment opportunities, especially of the irrigation potential in southern Malawi, was still underway, the Malawi Government requested assistance for immediate development of the Lilongwe area and of the Lower Shire Valley. The FAO/UNDP was to continue research into the possibilities of irrigation development, requiring intensive and lengthy preparation. A joint FAO/Bank mission was charged with the identification of projects that could ,be imple- mented without undue delay. This implied concentrating on the development of rainfed agriculture and on improvement of existing rather than the introduction of new land use systems. -2- II. CONCEPTION OF THE PROJECT AND THE APPRAISAL REPORT A. Conception 2.01 The conceptual phase began with the work of the above mentioned identification team, and was continued by the appraisal mission, staffed by the Bank and FAO, in May 1967. 2.02 The crop that lent itself to improvement in the area was small- holder cotton which, in spite of very low yields, had for a long time been the area's only rainfed cash crop of any importance. By the mid-sixties a cotton spraying technology had been developed whose introduction into the valley appeared promising. This, together with the necessary support by credit and extension, was identified as the main agricultural component of a development project. The second component that was designed for the project related to infrastructure and was considered largely complementary to the agricultural component. It was to consist of borehole construction (for the supply of water for both drinking and spraying), of the construction and improvement of roads for produce extraction and of the construction and equipment of produce markets. 2.03 A third quite separate component that was added to the project referred to a proposed settlement of 4,000 families on 20,000 acres. The primary cash crop was also to be cotton. The inclusion of the settlement component reflected a nationwide political program then in force. 2.04 The fourth component referred to staple food production. It was intended to increase maize yields by the introduction of improved varieties. 2.05 An area of 530,000 acres accommodating some 16,500 smallholder families1/ in the heart of the Lower Shire Valley was selected as the project area (see map). By applying the project measures to this area and to these people a range of objectives, which can be divided into two groups, were to be achieved: 1) Economic objectives (a) help improve the economic well-being of the people of Malawi; (b) increase the participation of farmers in the benefits of a cash economy; (c) contribute to the foreign exchange earnings of Malawi. 1/ It was not until 1974 that this rather reliable population figure for mid-1968 was worked out. The figure refers to the original project area. An extension in 1970 brought about an increase in project area and popu- lation of about 15%. - 3 - In relation to the first economic objective the other two can be considered more as instruments than as independent objectives. 2) Other welfare objectives (a) improve the nutritional standards of the populace particularly through the provision of increased maize supplies; (b) help combat disease by making available adequate water of potable quality. B. Critical Assessment of the Appraisal Report 2.06 The cash crop component, i.e. the introduction of the cotton spraying technology with the aid of extension and credit and the supporting infrastruc- tural component (boreholes, roads, markets) was well conceived, made sense under the given conditions and could be expected to contribute to the stated economic objectives and to some extent to other welfare aspects like health and communication. 2.07 The settlement component constituted a concession with respect to the political objective of settling young people in rural areas. Criti- cism should not be directed against the inclusion of a politically motivated component per se but against its poor preparation as part of an agricultural development project. Questions like where the settlers were to come from, how they were to be selected, who was to pay for the preparation of their plots and whether they should have a say in project policies were left open. It appears that it was not realized at the time that the settlement component could constitute a heavy political liability. The Salima project (Central Region Lake Shore Development Project), a regional development project also initiated in 1968 and supported by the Federal Republic of Germany, carried this burden in full. There, difficulties with the poorly planned settlement activity almost entailed the closure of the project. It may be credited to the management of the Shire project that the settlement component was tacitly removed from the project in 1970, when only 338 of the proposed 4,000 settle- ment plots had been occupied, thus keeping the economic costs and the poli- tical hazards to a minimum. 2.08 The planning of the maize component was not well founded either. In contrast with the details that had been worked out in connection with the introduction of the cotton technology, and the high degree of certainty that existed with respect to the success of this innovation, it,seems that the maize component was just "thrown in" at the last minute: high-yielding maize varieties were to be introduced from an unspecified source, and they would result in a "costless" increase of maize yields. The criticism is directed against the poor ex-ante elaboration of this component, not against the principle of its inclusion. On the contrary, the development project in the Salima area could show conclusively that the inclusion of a subsistence -4- crop activity with credit and extension programs can act as an important stimulus. The SVADP Appraisal Report did not, however, provide the necessary basis for.success (see 3.26). 2.09 Critical to the appraised design is the optimism underlying the projections into the distant future. After five years the project costs are assumed to drop to half. Adoption of the spraying technology by farmers is, however, assumed to continue until the last farmer in the area has been reached. A demonstration effect of unusual strength and longevity is assumed to take effect from year 6, but no explanation or justification is given. 2.10 The Appraisal Report showed an economic rate of return of 21%. It should be noted, however, that if the maize activity is excluded, the rate of return is reduced to 17%; and if the hazardous assumption about ever- increysing benefits from year 5 on is dropped, the rate of return goes down to 11%.1/ A forecast of a rate of return between 10% and 15% would have been more realistic at the time of appraisal. 2.11 It may be concluded that the basic concept of the cotton development strategy as presented in the ARI was a sound one, but that the overall projec- tions of costs and benefits were overly optimistic. 1/ For details see annex 1. If, in addition, the assumption about a 50% cost reduction in the future is dropped, the rate of return falls to 6%. The figure of 6% may be unduly low since the assumption of declining costs is compatible with that of constant benefits. Compare annexes 9 and 10. -5- III. IMPLEMENTATION OF THE PROJECT A. Organization 3.01 Within the Department of Agriculture the Agriculture Development Branch was established and became responsible for the implementation not only of the SVADP but also of the Lilongwe, Karonga and Salima Projects. Under this branch a project organization with headquarters at Ngabu was set up. The Project Manager was given full responsibility for the execu- tion of the project and for the accounting of all finances and stores. Under the Project Manager there were six sections/ (administration, training, research, extension, credit and development), each with a section head. All in all the project employed on average over 300 Malawians, of which 20 were senior staff, 50 industrial class and 230 junior staff. 3.02 Staffing difficulties were experienced as the pool of Malawian personnel from which to recruit was extremely small and since conditions that could be offered to local staff were not particularly attractive (government salaries have increased by only 6% in the last decade; tempo- rary officers at Technical Assistant, Senior Technical Assistant and Tech- nical Officer levels had to be held on a month-to-month basis). In this respect the location of project staff positions within the Malawian civil service had a negative effect, since it prevented the use of special salary incentives. On the other hand it meant that senior project staff could be held against established government posts, an important advantage with respect to stability and continuity of a development project. The simultaneous start of three regional development projects in 1968 may have overtaxed the Govern- ment's manpower resources to some extent, but the Project Manager minimized the extent of the problem and found some advantages in the competition for the better graduates when the issue was raised in the audit. 3.03 The positions of Project Manager, Chief Executive Officer and the research section head were filled by expatriates throughout the life of the project (the original Project Manager left the phase II project in July 1975). Expatriates headed the credit and extension sections for 2-1/2 and 3 years, respectively, and were then replaced by Malawians. All other posts, including Assistant Project Manager and the head of the training section, were continu- ously manned by Malawians. The expatriates are given considerable credit for the efficient start up and implementation'of the project, including the cost savings and institutional links. But the field program was largely in the hands of Malawians and it is to that staff that credit is given for the impact on cotton production. 3.04 Close cooperation between the SVADP and the Farmers' Marketing Board (FMB), in 1971 renamed the Agricultural Development and Marketing Corporation (ADMARC), was.envisaged from the start. 3.05 ADMARC is one of the most important parastatal bodies in Malawi. 1/ This disregards a later reorganization which was carried out with a view to phase II of the project. An organization chart is given in annex 2. - 6 - Its functions include increasing the volume of exportable crops and improving the standards of agricultural produc ion, supplying the agricultural require- ments to farmers of customary land,!' purchasing the crops produced by these farmers, marketing agricultural produce for export, promoting its consumption in Malawi and abroad and establishing and developing agro-industrial enter- prises. ADMARC announces cotton prices and guaranteed minimum prices for other crops at the beginning of the season and has sufficient financial reserves and political power to support those levels. 3.06 Within the SVADP, ADMARC was to expand its marketing facilities and to organize the supply of inputs, the credit system and the purchase of crops. Difficulties were encountered in that ADMARC had an unexpected disastrous financial year in 1967/68 and took two years to recover and meet its obligations. Here it proved to be an important advantage that the project tendered out and purchased inputs on its own account. Inputs were thus avail- able and the effect of ADMARC's malfunctioning in the early years was minimized. Later the project was through this arrangement able to dictate input selling prices to ADMARC, thereby ensuring that the farmers were supplied with inputs at lower prices than elsewhere in Malawi. 3.07 Dependence on other government departments, including ADMARC, proved troublesome with respect to construction works (road construction by the Department of Roads, the armoring of borehole surrounds by the Ministry of Community Development, and the construction of markets by ADMARC) and it is now considered that tendering out these jobs to contract firms would have been a preferable alternative. 3.08 In its dealings with ADMARC as well as with other government departments the project management showed a remarkable capability to main- tain excellent relations while still emphasizing the issue of efficiency. 3.09 The project management was able to master a related difficulty: agricultural development projects which have a large construction component in addition to the activities of direct smallholder development put two different requirements on the project organization. Construction works require an organization "from-top-down" which ensures a smooth flow of information and orders from the top in order to achieve highest efficiency. Smallholder development necessitates an organization "from-bottom-up" which ensures a quick transmission of grass root problems into top level planning. This was achieved through a suitable organization of extension, the formation oi farmer committees, the close cooperation with the local government and the frequent and extensive general contacts between the project and the local populace. The project management was successful in both tasks, i.e. in terms of efficiency as well as responsiveness. 1/ i.e., land over which the traditional land tenure system prevails. -7- B. Disbursements, Credit Allocations and Project Costs 3.10 'The credit amount was determined so as to cover 80% of the total project costs, which the appraisal mission had projected to be K 3,854,000 (US$4,624,800); the credit was thus US$3.7 million. 3.11 A comparison of projected with actual disbursements of the IDA Credit (annex 3, table 1) shows that except for a slow start in the first year it was possible to stay on schedule throughout the project period. Communication between the Treasury and the project constituted a bottle- neck, however, which is not reflected in these statistics. The project was required to go through most tedious exercises of documenting the various claims, but the Treasury would fail to inform the project about the acceptance of these claims, receipt of funds, exchange rates and allocation of these funds according to the reimbursement schedule agreed upon. Backlogs of up to 18 months occurred. Thus the project was never up-to-date with its financial planning basis. 3.12 In addition procurement suffered from the tedious tendering procedures. The time lag between preparation of tenders and the receipt of goods was between 11 and 14 months. This long delay meant that unneces- sarily large stocks had to be carried, tying up cash and making budgeting difficult. It has been suggested that these delays might have been reduced if the 90-day tender waiting period had been cut for some items, and if the project management had been enabled to dispose of smaller sums without prior consent of either the Government or Bank. 3.13 Actual loan allocation and total project costs as given in tables 2 and 3 of annex 3 reflect a number of revisions during the project life. These revisions related to the extension of the project area, to an aerial spraying program, to the construction of health facilities and to the allocation of funds to phase II of the project. To some extent these revisions merely reflect the fact that the project proved to have been more than adequately financed and was thus in a position to finance additional activities. The major reasons for the underrun of the original cost estimates were: - the successful input purchasing policy of the project management (bulk buying in the early years; thus lowest prices could be obtained); - the sizeable contingency allowance of 15%; - the overestimation of some cost items in the original projections (boreholes, buildings). 3.14 The resulting savings were in the order of US$350,000, 10% of the credit. In addition there had been a mistake in the cost projection,of the credit scheme: the requirements had been calculated on a cumulative instead of a revolving fund basis. Repayments from the farmers were deposited with - 8 - the Bank of Malawi every year. By 1974 an amount of over half a million Kwacha had been accumulated and handed over to the Government. 3.15 The accounting system was not satisfactory. The basic cost breakdown was based on the IDA categories, whose objective is not concerned with management and cost control accounting. The project therefore had no systematic records of, for instance, recurrent expenses under appropriate operating activities, such as extension, and could not control expenses by periodic reference to past experience nor follow adequate budgeting proce- dures. In spite of these shortcomings it must be conceded that the project has kept a remarkably close watch on its expenditures in terms of final aggregates. 3.16 On the government side there were also accounting problems. Thus the Development Credit Agreement stipulated that separate accounts be kept for the project and these, as well as those of ADMARC, were to be audited at least once a year by an independent auditor, acceptable to IDA. Certi- fied copies were to be forwarded to IDA not later than four months after the close of the financial year, which, in both cases, was March 31. These time limits could not be met. As regards the separate project accounts, these were reflected in the annual government expenditure publication. How- ever, they were not presented in a useful form, since only aggregates were included under the Ministry of Agriculture. 3.17 It is a general and striking feature of the project that the standards of quantitative documentation were, up to the beginning of phase II, very poor. It is difficult to find in all the reports and accounts of the project, of the Malawi Government and the Bank, any two sets of data on the same item that coincide. C. Development Work 3.18 The project was terminated according to schedule on March 31, 1973. With the exception of the settlement plots the physical targets of the project were practically all met (boreholes, roads, staff houses, headquarter buildings, market places; for details see annex 4). All physical construc- tion appears to meet the technical standards, and all physical construction, with the exception of the health facilities, is considered to have been necessary to achieve the impact on production and economic development in the phase I area. 3.19 All extension services in the area came under the Project Manager's direct control (except for veterinary and disease control services). Before the project there were 10 extension officers for the 16,500 farmers. The extension density of 1:1,650 was thus even below the national average of 1:1,500 (now about 1:1,000). Within the project some 50 primary school leavers were trained in the first year as demonstrators. Their number increased to 135 in year 5. An extension density of 1:150 was thus achieved -9- which is quite high (the Salima project had a ratio which was closer to 1:500) .!/ 3.20 The actual extension work took many different forms, such as individual -visits, visits to primary schools, demonstrations for groups of 2 to 50 farmers, public meetings, radio listening groups and mobile demonstra- tion units. Due to the lack of any suitable innovation in maize production all extension work concentrated on cotton. The purpose of agricultural exten- sion for cotton production was essentially twofold: (1) to induce the farmers to buy a sprayer and insecticides from ADMARC on credit and to practice pest control in cotton,2. and (2) to induce the farmers to adopt general agronomic improvements like timely and careful preparation of the land, row planting, timely planting and weeding. Taking the number of farmers who bought sprayers as a criterion, the success of extension can be illustrated as follows: Sprayer-owning Farmers 67/68 68/69 68/70 70/71 71/72 72/73 Number 200 853 1,819 2,994 5,364 4,800/ % of all farmers- 1.2% 5.2% 10.7% 14.4% 25.2% 21.8% a! Preliminary b/ The total number of farming families in 1967/68 was 16,500, increasing at an estimated 3% p.a.; in 1970 the number increased an additional 15% due to expansion of the area; in 1972/73 the number of farmer families was 22,000. 3.21 In addition to the sprayer-owning farmers there are a number of farmers who borrow sprayers to spray their cotton. It is estimated there were half as many borrowers as owners, or a total of 2,400 borrowers in 1972/73. Borrowers are estimated to have sprayed an average of 2 acres each; owners sprayed about 3.5 acres each. 1/ The project sometimes quoted ratios of 1:10 or 1:20; these ratios reflect the number of demonstrators per sprayer-owning farmer. Both sets of ratios are somewhat misleading: while the work of the demonstrators does concentrate on the sprayer-owning farmers it is not restricted to them; through radio transmissions, public meetings, e.g., it is attempted to address all farmers. 2/ Attempts to introduce aerial spraying were given up in 1972 as it was found that it was not suited for smallholder agriculture. - 10 - 3.22 Unfortunately there is no information about the rate of abandon- ment and thus the turnover of spraying farmers, a most important criterion of success of an extension activity. The project management believes that drop-outs are mainly a result of poor weather conditions and that the total number of permanent drop-outs is negligible. In comparison with the cotton project in Salima the record of adopting spraying in the Shire Valley is quite good. It must be pointed out, however, that in the Shire Valley only one line of agricultural development- was offered to the farmers, whereas in Salima development possibilities had been established for rice, groundnuts and maize as well. 3.23 Credit procedures were, after an initial phase with excessive mark- ups, satisfactory throughout the project period. The farmers were offered a spraye costing K 54 c.i.f. Limbe for K 24, i.e. with a subsidy amounting to K 30.1 Repayment would be effected over 3 years with K 6 being due in the first year, K 12 in each of the two subsequent years. The mark-up would thus constitute K 6 or 25% of the subsidized price. Insecticides costing K 10.50 per acre-pack were charged at an unsubsidized price of K 11 (i.e. with a mark-up of 5%) and were to be repaid in one installment in the same year. All credit was given through ADMARC in kind, i.e. the inputs were handed out by ADMARC, and repayment was ensured through an efficient system of recording and by confronting the farmer with his due statements seconds after he was paid for his cotton. 3.24 Overall costs of the credit program are believed to be in the order of 15% of the landed costs of the inputs, which is high but not excessive for a smallholder credit scheme. Repayment rates in the various years were 98%, 98%, 97%, 88% and 99%. Unless a crop failure can be proven credit defaulters are taken to court. 3.25 Before October 1972 farmers' repayments were deposited with the Treasury at no interest and new cash was drawn against the appropriate IDA category for the next season's inputs. Subsequent to the supervision mission of July 1970, and further discussions with the following missions, it was agreed that a revolving fund would be set up as of October 1, 1972 with K 351,083. 3.26 With respect to the maize activity foreseen by the Appraisal Report project achievements were nil. The Appraisal Report had recommended the introduction of improved maize varieties, but there existed no varieties of proven suitability for the area. The trial program which was started did not produce any tangible results before the start of phase II and even then the variety found suitable in the trials was not accepted by the population for reasons of palatability. Furthermore the Shire Valley is probably the most difficult area for maize growing in Malawi for climatic reasons and also because soils do not significantly respond to fertilizers. In this respect the valley 1/ The ARI recommended abolishment of subsidy after year 5. It is now proposed that the subsidy be progressively removed during phase II. - 11 - differs markedly from, e.g., the Lilongwe area on the plateau which is among the most productive maize areas in Africa. The valley is at a dis- advantage with Salima also because the lake shore land of Salima is suited to a large number of cash crops and shows a marked fertilizer response. 3.27 Given that the planned settlement activity was tacitly abandoned by the project, the only other activity directly connected with agricultural development was research. Ngabu resparch station became part of the project and was supplemented halfway through phase I by a station at Tomali for replication purposes. The major work concerned the development of an ultra- low-volume (ULV) sprayer for individual farms and the search for a second crop that could be promoted in the Shire Valley. The hopes which had been attached at one stage to ULV spraying (cheaper, lighter, no water require- ments, easier to handle) appear to have been too high, according to a recent paper produced by the project management -1 2/ The efforts to find a second crop had, by the end of phase I, and even today, unfortunately, not had any unambiguously positive result. 1/ The reference is T. Russel: Evaluation Working Paper 75/1. The finding that the economic advantages of ULV spraying are negligible is surprising and in contrast with previous work. More detailed analyses may be warranted. 2/ Aerial spraying was also tried but had to be abandoned because it proved too difficult to find sufficiently large areas in which all farmers were prepared to accept aerial spraying and to share the costs. - 12 - IV. IMPACT OF THE PROJECT A. General 4.01 Before going into a detailed quantitative analysis of the project's impact it should be acknowledged that the project has had a general develop- mental impact which is impossible to quantify but is nevertheless important and considered by some project officers to offer the major justification for phase I. This general impact refers to the mobilization of a populace, who had been known for its lethargy, to the improvement of health standards and the awareness of health problems, to the political reintegration of the area through the demonstration of central government efforts, to the benefits from the provision of infrastructure and central services over and above production benefits, to the reorganization of the Malawian administration in the direction of development and efficiency, and to the training of senior Malawian staff and hundreds of lower level staff. Phase I of the SVADP may be said to have set up the grass roots participatory system fo,r rural development in the area. It is likely that the high reputation the project has among the farmers, the public and the Government is as well the result of these effects as of the more readily quantifiable increases of production and income. 1/ B. Production- 4.02 The appraisal estimates of incremental cotton production are based on some assumptions which now have to be revised. 4.03 The Appraisal Report projected that farmers - by adopting spraying - would increase their yields to 900 lbs and eventually reach 1,000 lbs/acre. Actual yields of sprayer-owning farmers averaged less than 900 lbs over the 5-year period and showed a marked trend downwards to reach a low of 450 lbs per acre in the last year. To a large extent this is attributable to bad weather, as is shown by the development of the yields of non-sprayed cotton over that period (see annex 5). It is accepted by the project now that over the years an average yield of 800 lbs/acre can, with the presently applied techniques, hardly be exceeded. Compared with average yields forecast for non-spraying farmers of 350 lbs/acre, this represents nevertheless a substan- tial improvement. 4.04 About 20% more farmers than anticipated have so far been induced to buy sprayers. In addition there are a good number of farmers who borrow sprayers. 1/ For details see annex 5. - 13 - 4.05 The average cotton acreage sprayed by each sprayer owner was 3.5 acres, and if one adds 1.0 acre equivalent to account for spraying by borrowers with each owned instrument, a total of 4.5 acres were treated per sprayer, slightly less than the 5 acres predicted. The last figure allowed for an increase in cotton acreage on sprayer-owner farms, an increase that has not occurred. In subsequent estimates for the audit, total cotton acreage is expected to expand no faster than the rate of growth of population. 4.06 These three effects - with respect to changes in yields, number of participants and area - to some extent balanced each other out so that actual gross production figures closely followed the estimated production pattern over the first four years. Only in year 5, when unusually bad weather condi- tions were experienced, did a significant fall-off occur. Cotton Production (short tons) 68/69 69/70 70/71 71/72 72/73 Estimated 827 1,964 3,165 4,548 6,331 Actual 956 2,007 3,311 4,106 2,227 4.07 The major discrepancy between actual development and appraisal pro- jections concerns the period after the first five years. The Appraisal Report predicted a continuous .increase of production for another 15 years to come. Project management believes, on the basis of almost seven years of experience with cotton development in the area, that all that can be achieved with a continuation of current efforts is either the maintenance of production on the attained level or at best some modest expansion. This view was im- plicitly shared by the team appraising phase II of SVADP. Constant production per annum has therefore been assumed from year 5. On the basis of the 5-year experience with yields and acreage this production has been estimated at 4,350 short tons. At present this would roughly represent one-fourth of the total production of the Lower Shire Valley. Looking at the-phase I area only, the 4,350 short tons represent an increase of some 40% over the production that would have been achieved without the project. This increase in production is achieved by some 5,000 farmers who,have bought sprayers and apply the spraying technique and by another group of about 2,500 farmers who borrow sprayers and apply the technique on additional acreage. 1/ C. Farm Budgets- 4.08 - Before 'the project (in about 1967/68) the average 5-acre farmer in the area would grow 3 acres of maize.foi subsistence and 2 acres of unsprayed cotton. Valuing his subsistence at selling prices his,gross margin (which is 1/ For details see annex 8. - 14 - roughly equivalent to his family income) would be in the order of K 50. His return to labor would be around 21 tambala/manday (25 US cents). 4.09 The lot of this non-spraying farmer improved over the project life due to factors quite independent of the project, most notably cotton price increases. Thus, after the project (say, 1973/74) the same farmer's gross margin surpassed K 60 and his return to labor increased to 26 tambala/ manday. 4.10 With the project this farmer was given the opportunity of buying a sprayer and increasing his cotton production. Without any increase in acreage his gross margin would rise by 50% to K 90. A 7.5 acre farmer with 4.5 acres of cotton would increase his gross margin on account of the project from K 120 to K 180. A 50% increase in gross income attributable to the project may therefore be considered typical. The return to labor would be between 25 and 27 tambala/manday on the larger farm as well as the smaller farm.l/ 2/ 4.11 These figures show that the project has provided a way for the innovative farmer to substantially increase his income. At the same time it must be noted that the return to labor is hardly affected by the project. This means that by adopting the innovation the farmer can earn more, but he has to work more as well. In other words, the project shows a way of employing more labor (not necessarily more people) productively in agriculture. The economic value of this effect hinges on the existence of alternative sources of income and on the opportunity cost of labor. 4.12 Alternative employment of labor should first be sought on the farm. As a smallholder cash crop only cotton is presently of any importance in the Shire Valley. The major subsistence crop is maize. If one attempts to assess the relative value of a maize production activity within the farm, and if the level of this activity is kept within the limits of the subsistence require- ments of the farm family, the appropriate price with which to value maize is the buying price. Maize is presently sold by ADMARC at a price of 2.5 tambala/lb. The return to labor in maize production then works out at 36 tambala/manday, at a yield level of 800 lbs/acre at 52 tambala. This shows that quite apart from any risk considerations it will always be profitable for the farmer to grow his maize and that improved cotton production is by no means attractive enough to induce the farmer to divert to it labor and land from subsistence production. Once subsistence requirements are met, however, the return to 1/ The Evaluation Working Paper 75/1 (T. Russel/SVADP) shows that neither in terms of income nor in terms of return to labor is anything to be gained by switching from knapsack spraying to ULV spraying or even aerial spraying. 2/ A small sum should be added to cover income from renting equipment. But there are no set fees and the adjustment would be inconsequential. - 15 - labor in maize growing drops below that in cotton growing because of the significant difference between ADMARC's buying and selling prices (ignoring minor and unpredictable transactions between neighbors). It appears that there are then no real alternatives to cotton in terms of return to labor. 4.13 Little is known about the economic opportunities of the agricul- turalists and their families outside their farm. To some, fishing may be a secondary source of income. The area does not appear to be an important source of migrant labor for South Africa. There has been an inflow of people from Mozambique and from the Nsanje area in the valley. The minimum wage has for about a decade stood at 24 tambala/day, an amazingly low level compared to the neighboring countries. All these factors tend to suggest, at least for the past, that there has been a surplus of labor in the area. 4.14 That situation has begun to change. For one thing the demand for labor in cotton production shows peaks, particularly during harvesting, which have already become difficult to reach. Furthermore, the 10,000 acre Sucoma sugar estate is offering high wages (between K 0.5 and K 1.0 per day) throughout the year to attract labor away from small farms into wage employment. The planned expansion of this estate by some 18,000 acres, which is roughly equivalent to the area of sprayed cotton under phase I (Sucoma lies inside the phase II boundary), will create employment for several thousand men in the valley and thus profoundly affect the labor market. The relative attractiveness of cotton growing to the farmers may therefore be expected to decrease over time. 4.15 We have also attempted to analyze the actual development of farm income and return to labor over the years, i.e., taking into account the fact that yield levels and the input requirements are different in every year. For sprayers as well as non-sprayers, the gross margin per acre and the return to labor fluctuate considerably, and the cotton technology does not appear to result in a greater degree of certainty of yields and returns. Annex 8e shows these effects. 4.16 In summary, the project does provide the opportunity of increasing income. Returns to labor are, however, low and will only be accepted as long as there are few alternatives for employing labor, alternatives such as those offered by the expanded Sucoma estate. The differential in the return to labor between employment in an irrigated sugar cane plantation and in rainfed cotton production raises fundamental doubts about the long-term competitive prospects of rainfed. agriculture in,the Shire Valley, given the real possibility of further, substantial expansion of the irrigated area. 4.17 An interesting aspect of the project is the high number of sprayer- borrowing farmers. It must be assumed that this is an economically attractive way of cotton production for a good proportion of the smallest farmers. In addition, the credit risk to the borrower is low, if it exists at all. Unfortunately there is no quantitative information about the economic situa- tion of the sprayer-borrowing farmer. - 16 - D. Equity 4.18 The equity issue here refers to the distribution of wealth and income. It'has at least two dimensions: an interregional one and an intraregional/interpersonal one. 4.19 With respect to the interregional dimension, there is a value added of some K 800,000 p.a. attributable to the project. If the average family income in the valley without the project, but allowing for price increases, is correctly estimated at K 60 p.a. (see page 14) it would be on the order of K 90 to 100 with the project. The project can be credited with a significant first step towards interregional equity, even though the new level is still considerably below the national average and even below the national rural average-. 4.20 Interpersonal equity within the project area is a more difficult issue. There is no precise information on the ex-ante situation, although, given the absence of any large landowners in the valley, one may assume that there had not been significant and permanent differences in income and wealth. 4.21 There has for some time been a tendency to think that the intro- duction of any form of technical progress into traditional agriculture would by inherent mechanisms favor the larger agriculturalists, thus leading to a polarization of classes. The opposite view, that technical progress is basically neutral with respect to the distribution of income and wealth, has recently regained scientific standing. It essentially claims (1) that reality is complex and there is no one view doing justice to all innovations in agriculture, and (2) that instead of pola-ization of classes we find a differentiation according to effort and performance which cuts across the classes. With respect to the project under consideration one might in the absence of any form of quantitative information lean towards the second point of view. The adoption of spraying does not presuppose any minimum acreage; no securities are needed to obtain credit. The only criteria for participa- tion are (1) ridging of 2 acres and (2) acceptance by the Farmers' Committee. Procurement of the inputs does not require social status or wealth and the end product can be easily sold by everybody without discrimination. Further- more, since one cannot talk about an extremely uneven distribution of income and wealth in the pre-project period the possibility that large landowners benefited disproportionately from the services provided by the project does not arise. 4.22 This does not, mean that all farmers in the area can with time be expected to reach the income levels of those 4,000 farmers who have success- fully adopted spraying, or even that all of these 4,000 will establish them- selves as viable cotton farmers. As has been clearly shown by the five years 1/ In 1972 the per capita income in Malawi was estimated at K 80; assuming 5 family members this represents a family income of K 400. The average rural income was believed to be about half that amount. - 17 - of phase I, weather is extremely variable and there is a pronounced risk involved in rainfed agriculture; accordingly a profound differentiation is to be expected to develop in the future. The project management describes the situation as follows: "For the good farmer four out of five years are above average, the poor farmer experiences four bad years out of five. In the long run we cannot expect that more than 10 to 20% of the population will reach adequate living standards through the establishment of proper rainfed farming systems." The quote does not suggest that the use of sprayers will eventually be confined to 10 to 20% of the farms; rather that whatever the number of farms spraying, and the number may be large, the percentage of all households for which the cotton culture can provide a viable income level in all years will be very small. E. National Economy 4.23 A cost-benefit analysis, whose chief feature lies in the valuation of all goods and services at border prices, is presented in detail in annex 9. Not taking into account the price increases on the world market, assuming constant benefits from year 6 on, as explained above, allowing a decline of costs after year 6, assuming the shadow wage rate is zero, and excluding as project costs items unrelated to production (health services), the internal rate of return turns out to be 15%.!/ This shows that the project constituted a good combination of resources in the light of the choices of production and trading that were open to the Malawian economy in the late 1960s. The justi- fication neither depended on hazardous assumptions about "costless demonstra- tion effects" nor on overoptimistic cost reductions for the future. It also did not depend on any price increase of cotton on the world market. 4.24 With the recent price increases of cotton, and on the basis of revised price projections for the future, the rate of return of the project is boosted to 27%. It is shown in annex 9 that substantial increases of farm input co ts, e.g. the 60% increase feared by the project management for this season,- as well as a doubling of marketing costs, could be borne from 1974/75 on with the rate of return of the project still staying well above 15%. 4.25 Annex 9 also shows that the opportunity cost of the additional farm labor necessary for spraying could rise to K 0.4/day before the rate of return 1/ Indirect benefits, difficult to quantify, exist, such as on-the-job training of extension workers and improved infrastructure. Thus the 15% calculated may only be a floor estimate. 2/ The project originally bought large stocks of cotton equipment at low prices, and has thus been able to defer the impact of the rapid rise in those prices. Fertilizers were not stocked, but the effects of recent increases in fertilizer prices is a phase II issue. - 18 - drops below 10%. If we assume that any project returning at least 10% is acceptable for Bank finance the calculations of annex 9 mean that from the point of view of the national economy, work in the sprayed cotton culture is economical and more valuable than indicated by the return to labor calculated from the point of view of the farmer (see preceding section D). Should irri- gated estate agriculture continue to expand and force the opportunity cost of labor above the K 0.4/day level, it would be desirable from the point of view of the national economy to restrict cotton growing to the non-irrigable parts and to promote irrigated agriculture in the remainder of the valley.l/ 4.26 The type of social cost-benefit analysis applied shows, in the absence of those employment alternatives, not only the desirability of phase I in terms of the national income objectiveV' but at the same time its efficiency in earning foreign exchange.i3 In terms of absolute amounts the project has an estimated annual net contribution of foreig. exchange of over K 600,000 by year 10 (over K 300,000 at unchanged prices of inputs and outputs). F. Government Budgeti' 4.27 For developing countries government revenues often constitute the most serious bottleneck in all development. The effect of a project on the government budget may well be considered more important than the effect on equity or on national income. 4.28 The major way in which the Malawi Government received revenues from the project is through the de facto export tax on cotton. The gross proceeds from cotton sales at Malawi's borders are regularly higher, after being converted into Kwacha, than the payments to the farmers plus costs of trans- port, handling and processing, resulting in a cotton margin accruing to ADMARC which for the purposes here can be taken to be synonymous with the Government. 4.29 A juxtaposition of direct government revenues and expenditures on account of the project shows that the net fiscal flow is negative. The project does not generate sufficient government revenues to ever allow a repayment of the credit with which it was financed, and this in spite of the fact that cotton price increases have been taken into account. The Appraisal Report argued in 1968: "while the latest Bank economic report on Malawi recommends that external assistance cover 85% of project costs, the proposed IDA Credit will cover a somewhat smaller percentage since a large proportion 1/ This would imply offering year-round employment to a high propori ion of the active population of the valley at sustainable wages beLwcen K 0.5 and 1.0, as is expected now. 2/ Under the assumption that an optimal trading policy is followed. 3/ The contribution to nat ional income is rcv.ihled in foreign exchange units. 4/ DetaiLs given in annex 10. - 19 - of the local contribution will be recovered by the government from farmers during the project period." However, government recoveries are likely to be much less than anticipated, and without increased taxes or cesses will probably be insufficient to meet repayment obligations. 4.30 One might point out other means by which the Government could partake in the income generated in the project area, like raising an income tax, but this would politically not be feasible.l/ One might also point to indirect effects and multipliers from the spending of the increased income. While one can hope that such effects will occur, it should not be overlooked that these are unspecific effects.!/ Resorting to them for the justification of a project conceals the fact that the type of project under consideration (smallholders;.large social investments) always implies a heavy burden on the government budget, the more so the more a project is conceived as a "rural development" project. This is not surprising, since in rural development projects by definition a large proportion of the gene- rated benefits stay with the rural population to the detriment of the fisc. If one is dealing with only a small project of that type in a large country, one may simply consider this as a desirable form of redistribution. If, however, a small and poor country like Malawi establishes several externally- financed extensive smallholder development projects, one wonders whether this will not seriously affect the overall repayment capacity of the country irrespective of the economic rates of return of these projects. This is an illustration of the importance of fully considering projects' cumul.iiiv. fiscal impact as well as their economic rate of return. V. THE BANK'S ROLE 5.01 The Bank accepted the ARI in spite of some overly optimistic elements such as the maize activity and the costless demonstration effect, which made the proposed project look more profitable than it actually was. Nevertheless, the project was rightly accepted because the fundamental cotton technology made sense and the proposed scale of the operations was realistic. The extension services were necessary to diffusion of the technology, and project credit was apparently used by all sprayer owners. 5.02 The Bank can be criticized for having committed.funds substantially in excess of those actually needed, partly because of simple oversight. Thus the turnover of farmer credit each year made K 500,000 of Bank funds superfluous. 1/ According to the ARI, an IMF fiscal mission had proposed the introduction of a cess of 5% on crops marketed by ADMARC. The Malawi Government did not accept this- proposal; from the point of view of farmers' incentive this may have been a good decision. 2/ In the sense that they are not dependent on the project type but would result from any income-generating activity of a similar order of magnitude. - 20 - These were handed over to the Malawi Government at the end of phase I and used as counterpart funds for phase II. Also, K 100,000 of other phase I funds were put forward to phase II with the consent of the Bank, and stocks worth well over K 100,000 were carried forward. This means that the economic impact as calculated was achieved at a cost some US$850,000 below the amount that had been allocated to the project. These funds were not wasted, of course, but their premature commitment nonetheless carried some costs. 5.03 During the execution of the project the supervision missions were found very useful by the project management and by the Malawian authorities. They also consider it important that Nairobi took over the supervision duties, since this allowed a quicker exchange of ideas and a more personal relationship. Concerning the frequency of the supervision missions, the project management feels that six-monthly intervals would be the optimniim. But the management also feels that in more recent times, including the phase II period, the.supervision missions have not comprised enough tech- nical disciplines-, as a- result of which many of the most burning problems of practical agriculture development could not be discussed. 5.04 On the whole, the Bank's performance in preparing and supervising the project can be considered satisfactory. It was less satisfactory, however, with respect to the ex-post assessment of phase I and the derivi- tion of the phase- II concept out of the phase I project. 5.05 In June 1972, a mission was sent to Malawi to appraise plans for phase II of the SVADP.. The Appraisal Report for phase II (ARII) was offi- cially released in Fe5ruy 1934. Phase I was terminated in March 1973;1/ phase II followed on without a break. In May/June 1974, a mission went to the Shire Valley to cdlect data and information on phase I for a Project Completion Report (fCR) which was presented in September 1974. Probably because of a lack of systematic monitoring of phase I, and because of time pressure in preparing phas& II, the reports produced by these missions were ambiguous and inconsistent in a number of aspects. 5.06 Thus the separation of phase I benefits from phase II benefits - necessary in order t6 avoid double counting - is not clear.2/ The ART predicted that the number of farmers adopting spraying would continue to grow after the first five years for another twenty years. The PCR accepts and reconfirms this assup Iptidn, that annually growing benefits can be attributed to phase I.-' ARII on tiie othier hand assuimes tha, wiLu phasc II the number of spraying farmers would remain constant and that it is due to the phase II efforts that their numbers, and thus cotton production, increase, 1/ Though the credit was not closed until March 1974. 2/ For details see annex 11. 3/ The ARI thus calculated a rate of return of 21%. The PCR on the basis of revised data estimates a return of "at least 15%." Information con- tained in the PCR is insufficient to repeat and verify that calculation. - 21 - The phase II calculation, which gives an economic rate of return of 22%, thus includes the post phase I benefits which had previously been attributed to the first project. If one reconciles the two projections by keeping phase I benefits constant from year 5 on, and if one adds the cost-benefit stream of phase II, the internal rate of return to both phases strung together drops below 19%. At first sight this may seem a negligible change, but this is only so because the effect of the double-counting of cotton production is cushioned by the significant proportion of benefits that are assumed to accrue from other crops. In phase I they did not materialize. The question is whether they were justifiably expected from phase II. In early 1975, there was still every indication that these benefits would not materialize, just as in phase IL/ If that were to prove the case, and if there were no other offsetting benefits (such as occurred with the increased cotton price under phase I), the economic rate of return in phase II would drop from 22% to well below 10%. Taking the two phases together and elimi- nating the effect of double-counting, the economic rate of return of the Shire Valley Agricultural Development Project phase I and phase II strung together is only around 12%, ten percentage points less than the figures given in ARI and ARII. 5.07 In addition, and again without reconciliation with the assumption in the ARI (although later refuted in the PCR), the ARII assumes an average cotton acreage of 1.8 acres for non-spraying farmers and 3.5 acres for spraying farmers. This implies that by adopting spraying a farmer automa- tically increases his cotton acreage from 1.8 to 3.5 acres. All the surveys of the project management, in particular of the Evaluation Unit, point to an aveyage cotton acreage of the unimproved farmer of about 3 acres. Thus one-half of the cotton benefits from phase II are based on a shaky assumption about area expansion which is not substantiated by any firm evidence. 5.08 The above inadequacies can be summarized as follows: - the Bank did not take the time to look back and learn from past experience2! In particular this refers to the inclu- sion of secondary crops on the basis of flimsy evidence as to their suitability. - the Bank did not take the time to achieve consistency among the major project documents of phase I and phase II. - the measure of optimism underlying the cost-benefit projec- tions and the rate of return calcula'tiuns appears to be unacceptable in some instances. 1/ The PCR states as the second of the two major lessons to be learned from the phase I project: "The failure of the new maize variety shows that appraisal teams should only include benefits based upon known research facts." This criticism was directed against ARI but holds for ARI as well. 2/ During negotiations for phase II the Bank turned down the project manage- ment's request for a year zero between phase I and phase II to allow more adequate preparation. - 22 - 5.09 The most recent supervision report, prepared subsequent to the audit field visit,i/ supports the reservations expressed in this audit with respect to project benefits.2/ The phase II appraisal crop projec- tions are now considered unrealistic and a complete review is planned by the Government. Average sprayed acreage per spraying farmer has dropped, and the number of non-spraying cotton growers has begun to decline, possibly as a result of the new employment opportunities. This state of affairs may reflect the limitations of a development strategy based on raiifed cotton, and subsidiary crops presently available, for the valley as a whole. 1/ The supervision report was issued in May 1975. The supervision mission coincided with the audit mission. 2/ It should be stressed that the above critical remarks about phase II of the SVADP refer to the cropping component only, and that other components like fisheries and livestock development look very promising. - 23 - VI. CONCLUDING REMARKS 6.1 Irrespective of some critical points in the Appraisal Report and of some problems in the transition to phase II, the SVADPI can be called a good project, well-conceived and well-implemented. The fact that phase I, one of the Bank's first smallholder projects and sited in a desperate area, worked as well as it did, should encourage those who believe that no rural condition is hopeless. 6.2 The project's role as a pilot scheme should not be misinterpreted however. The results suggest that farmers elsewhere in the valley can also make profitable use of the rainfed cotton spraying package and expand their total cash income. They do not suggest that the package promises to attract the great majority of valley farmers, that it provides a basis for a viable self-supporting small farm economy for any other than the more capable farmer, or that it can be used as a leading edge of a multipurpose rural development program many of whose other components are non-remunerative. 6.3 The tentative conclusion that the long-term development potential of rainfed agriculture in the valley is limited may lead planners in the future to shift emphasis towards irrigation strategies, as had been under study at the time of project conception. - 24 - ANNEXES Sensitivity Analysis of ARI Projections Annex 1 PPAR/ SVADP Credit 114-MAI 1. Cost-Benefit Flow as Projected 2. Cost-Benefit Flow as Projected but Excluding Benefits from Maize YEAR COSTS BENEFITS YEAR COSTS BENEFITS 1 1052000. 30000, 1 1052000. 24000. 2 624000, 14.0u0 2 624000. 130000. 3 592000. 26e000, 3 592000, 24.000. 4 342000. 584000. 4 342000, 348000, 5 326000. 582000. 5 326000, 516000. 6 230000. 734000, 6 230000, 642000, 7 23000. 810000. 7 231000. o90000, a 8 231000. 894000. 8 230000. 746000, 9 210000. 1000000. 9 210000. 822000, 10 190000. 1036000. 10 190000. 824000. 11 170300. 1044000. 11 170000. 818000. 12 171000. 1036000. 12 * 170000. 814000. 13 170000. 1U28000. 13 170000. 812000. 14 170000. 1026000. 14 170000. 818000. 15 170000. 1006000. 15 170000. 802000. 16 170000. 99ti00. 16 170000. 798000. 17 171000. 9u000. 17 171100. 186000, 18 170000. 960000, 18 170000. 770000. 19 171000. 862000. 19 170000. 676000, 20 170000. t88000. 20 170000, 600000. 21 170000. l0000. 21 170000. 526000, 22 170000. 626000. 22 170000. 454000. 23 170000. 552000, 23 170000. 586000. 24 174000. 478000. 24 170000. 31o000. 25 170000. 412000. 25 17o0o. 254000, Internal Rate of Return: 21.144% Internal Rate of Return: 17.077% 3. Cost-Benefit Flow as in 2. but 4. Cost-Benefit Flow as in 3. but Assuming Constant Benefits from Assuming Constant Costs from Year 5 on Year 6 on at K 230,000 YEAR CoSTS BENEFITS YEAR COSTS BENEFITS 1 1052000. 24000. 1 1052000, 240U0-, 2 624000. 13(000. 2 624000. 13000. 3 592000. 24k000. 3 592300, e4 000. 4 342000. 34no00, 4 342000* 348009. 5 326000. 51 6000, 5 326000, 1 6 230000. 516000. 6 230000, 516000, 7 231000. 51f00.0 7 230000. 516000, 8 230000. 516000. 8 230000, 510000. 9 210000, 516000, 9 230000, 516000. 10 190000. 5160000 10 230000, :10000, 11 170000, 516000. 11 230000, 516000, 12 170000, 516000. 12 230000, 516000. 13 170000. 516000. 13 230000, 516000, 14 170000, 516000 14 230000, 516000, 15 1710000, 51oo000 15 230000, 51 (0004, 16 170000. 516000. 16 230000, 516000. 17 170000. 51000o. 17 230000, 516000. 18 170000. 510000 18 230000. 516000. 19 170000. 516000. 19 230000. 516000. 20 170000, 516000 20 230000. 516000. 21 17o0 0, 516000, 21 2300Q0, 516000, 22 170000. 516000 22 230000, 516000, 23 170000. 516000 23 230000, 51600). 24 170000. 516000. 24 230000. 516000. 25 17,l00, 516000, 25 230000, 5160jO. Internal Rate of Return: 11.391% Internal Rate of Return: 10.185% 5. Cost-Benefit Flow as in 3. but Assuming Constant Costs from Year 6 on at K 326,000 YEAR COSTS SENEFITS 1 1052000, 2 624000, 13l.:) 3 592000, Z42000, 4 342000, 343000, 5 326003, 51o0il. 6 326000, 7 326300, Si 000. 8 326003. 516000. 9 326000, 510003. 10 326000, 51600). 11 326000, 12 326000, 13 326003, 14 326000, 51O00. 15 326000, 5.1 o '. 16 326000, 51000. 17 326000. 51t'0i. 18 326000, 51C)00;1. 19 326000, 51h000. 20 326000, 51.5000. 21 326000, 51600). 22 326000, 510000. 23 326000, 51no0). 24 326000, 31600). 25 326000, 1630'W. Internal Rate of Return: 6.391% ORGANIZATION CHART OF THE SVADP - YEARS 2-3 1)Annex 2 SECRETARY FOR AGRICULTURE & NATURAL RESOURCES Cr PROJECT MANAGER ji ~~EXTENSION AMNSRT F2O ON CREDIT -TRAI1NING DEVELOPMENT_ Statistcs and Agricultural Extension. Project accounting Credit accounting Farmer and staff Settlement programme Data analysis. Agro-economic survey. Inter cash and Credit training. Training (Planning and imple- Agronomy and Departmental liaison, port and workshop farmer input Centre management mentation) Borehole Entomological Animal Husbandry. F.M.B. Establishment and sales.Covern,ent District Adminis- and road construction research. Crop liaison, crops, and market- salaries. Housing. Loans Board tration and Politica liaison and planning. pests and diseases ing. Discipline Accounting, liaison "Achikumbel Land Use planning. Work oxen programme. ___________ ________________training team, AGRICULTURAL PRINCIPAL FIELD CHIEF EXECUTIVE SENIOR FIELD SENIOR 'FIELD SENIOR FIELD OFFICER (1) OFFICER (1) OFFICER (1) OFFICER (1) OFFICER (1) OFFICER (1) c Field OfficerM1 Field Officer (4) Executive OfficerM1 Field : fficer(l) Field Officer(1) Charge Hand Mechanic() Technical Assistants Technical Assistants Clerical Officers Technical Assistant Technical Assistants Technical Assistants (6)Seeconded from A.R.C. (15) (8) (M (5) (2) Development Assistants Development Assistants Development Assistants Credit Assistants De velo pme nt Assistant Devel1opme nt Assistants (6) (110) (3) (12) (M (6) Typiats Developent Assistants Homecraft Worker (M (10) (1)> Subordinate Staff Subordinate Staff (55) (2) 1) i.e. before the reorganization for phase II. - 29 - Project Finances Annex 3 PPAR/SVADP Credit 114-MAI Table 1 Estimated and Actual Disbursements (US$'000) Credit 114-MAT Disbursements by end of Estimatedl/ Actual!' 1968 - 41 1969 1090 433 1970 1805 1807 1971 2539 2355 1972 3098 3125 1973 3700 3700 1/ Computed from data in the appraisal report; an explicit schedule of estimated disbursements is not given. 2/ From audited project accounts in Malawi; converted into US$ on the basis of 1 MK = 1.2 US$. The fact that the Malawi. financial year does not coincide with the calendar year has been neglected. - 30 - Project Finances Annex 3 PPAR/SVADP Credit 114-MAI Table 2 Actual Allocation of Loan in US$ Equivalentl/ Credit 114-MAI Buildings and Office Equipment 9,039.73 Vehicles 111,981.61 Well Casings, Fittings, Pumps 81,983.14 Knapsack Sprayers and Insecticides 1,016,900.26 Other Equipment 77,078.39 Staff Salaries (F.E. cost) 77,913.28 Pest Control, Extension, Credit 770,619.23 Housing and Marketing Facilities 627,478.40 Public Works a Equipment 647,517.13 Loan to ADMARC,1 182,216.04 Health Facilities 97,272.79 Grand Total 3,700,000.00 1/ According to Controller's statement of March 23, 1974. An estimate of loan allocation is not con- tained in the appraisal report. 2/ Onlending conditions: 6% interest rate; repayment in 20 equal half-yearly installments beginning October 1, 1973. Note: The above statement does not distinquish between phase I and phase II items. The IDA loan element allocable to phase I amounted to US $ 3,577,200 (K 2,981,000) only according to table 3 of this annex. The balance was put forward to phase II. - 31 - Project Finances Annex 3 PPAR/SVADP Credit 114-MAI Table 3 Total Project Costs (Revised) (Shown in Malawi Kwacha '000) Category Item Description Years Total I II III IV V Project '68 '69- '70- '71- '72- Cost '70 '71 '72 '73 5/ .5/ 1/ 2/ 3/ 3/ 4/ I 1 Buildings & office 4 2 5 - 1 12 equipment 2 Vehicles 15 26 9 32 31 113 3 Well casings, - 49 15 - 5 69 fittings & pumps 4 Knapsack sprayers - 162 199 205 256 822 and insecticides 5 Other equipment 14 44 - 4 18 80 6 Staff salaries (Foreign exchange 2 14 15 18 20 69 cost of expatriates) II 7 Pest control, 20 94 153 230 280 777 extension and credit 8 Housing and marketing 12 450 113 69 53 698 facilities 9 Public works and 21 190 323 78 110 722 equipment II 10 Loans to ADMARC - 21 60 118 - 198+ III 11 Health facilities - - - 5 93 98 Total 88 1052 892 759 867 3658 Local Costs 18 270 51 160 178 677 I.D.A. Loan Element 70 782 841 599 689 2981 1/ 12 month fiscal calendar year, expenditure commenced 1st April, 1968, actual expenditure shown. 2/ 15 month fiscal year from 1st January, 1969 to 31st March, 1970, actual expenditure shown. 3/ 12 month fiscal year from 1st April to 31st March, actual expenditure shown. 4/ 12 month fiscal year from 1st April to 31st March, revised estimates of expenditure shown, prepared at 30th September, 1972. 5/ Categories fixed by I.D.A., Items and descriptions agreed between I.D.A. and the Malawi Government. + Total incompatible with parts due to rounding errors. - 32 - ANNEX 4 PPAR/SVADP Credit 114-MAI Physical Execution of the Project 1) Boreholes: Calendar: 1968 1969 1970 1971 1972 Total Total drilled 12 64 49 66 16 207 Abandoned 2 8 7 5 2 24 Productive 10 56 42 61 14 183 Ratio of failure 1:6 1:8 1:7 1:13 1:8 2) Roads Appraisal estimated that 160 miles of crop extraction roads would be built and 24 miles of roads improved. The programme was completed except for a section of eight miles which ran through the extended Lengwe game resefV, whete in any case soil is unsuitable for farming. 3) Buildings The construction of buildings refers to the following units: 18 senior staff houses 32 intermediate staff houses 16 junior staff houses 132 low cost staff houges 1 two-storey offic6 block 2 single stor6y (foui room) blocks at present housing Land Use, Research and E*aluation Sections 1 three-bay workshop and 2 stores In addition a health cehtie was contructed. 4) ADMARC completed its new regional offices in Ngabu and moved in its staff. Additional stores for cotton were built (Dutch barns) in Bangula and Mitote (Chikwawa). At appraisal there were 8 permanent and 3 temporary ADMARC markets in the project area. In year 2 the project was extended to include two more permanent markets. ADMARC afterwards constructed one new permanent market and converted two temporary markets into permanent markets. In addition ADMARC created three new temporary markets. The market infrasttucture attained was considered satisfactory. The appraisal anticipated construction of four new permanent markets and upgrading of three tempotary markets. 5) Settlement At appraisal 4,000 farm families were to be resettled on 20,000 acres of arable land. Up to 1970 only 338 settlers had taken up individual plots of about 12 ackes each. The table below shows progress up to that date. - 33 - ANNEX 4 Total No. of plots available Av. size No. of Settlement Acreage End End of plots Settlers Area Allotted 1970 1972 1973 (Gross) Mangulenji 3,700 306 - - 12 210 Ndakwara 9,100 240 740 - 12 128 Phadze 7,900 - - 580 12 nil + projected In the course of the supervision mission in July 1970 the target was reduced to 1,300 families. After that date the settlement component was apparently no longer considered a project activity. The audit did not investigate the uses of land already cleared but not settled by 1970, nor of the remaining area removed from project control. Projected and Actual Cotton Production in Phase I Project AreaA Ain/e 5 P P AR/SVAD? Phase I Period Phase II Periwd Credit I ACTUAL FIGURES 67/68 68/69 69/70 70/71 71/72 72/73 73/74 74/75 75/76 76/77 77/78 82/.3 87/88 Production of SOF and SBF witnproj ect2/ Nro. of sprayer-owning farmers (SOF) 200 853 1819 2994 5364 4800 4800 4800 4800 8 C 4 ?0O 4800 4800 Average cotton acreage of SOF 3.5 3.2 3.2 3.5 3.9 3.5 3.5 3.5 3.5 3.5 3.5 3.5 3.5 Total cotton acreage of SOF (rounded) 2700. 5900 10500 21000 16800 16800 16800 16800 16800 16800 16800 16800 Average. vield, n SOF lbs/acre . 13.09 1059 97.7 65-3 450 800 .800 800 800 800 200 800 Cotton production of SOF(sht) 1767 3,124. 5129 6857 3780 6720 6720 6720 6720 6720 C720 6720 No. of sprayer-Iorrowing f-a-rners. (SBF) 400 1000 L500 2500- 3?100 2400 2400 2400 2400 2400 2.00 2400 Aver-age cotton. acreage- of SBI 2.0. 21..0 2'.0 2:.O. 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0 Total cotton acreaqe of. SBF 800 2000 3000 5000 6200 4800 4800 4,;00 4 :30 4800 400 4300 Averaae yield of SiýF lbs/acre 1003 811 748 500 300 6/2 672 172 672 o12 672 672 Cotton production of SBF (sht) 401 811, 1122 1250 930 1613 1613 1613 11113 1613 1613 _;L3 Total Prod. of SOF and SBF 2168 3935 6251 8107 4710 8333 8333 8333 8333 8333 8333 6333 with (sh t) Cotton yields without spraying (lbs/acre) 350 500 400 400 324 200 350 350 350 350 350 350 350 Production of SOF and SBF without 3/ - n,ravers- No. 200 206 212 219 225 232 232 232 232 232 232 232 232 preducrion (sh t) 427 364 375 288 183 325 325 325 325 225 :25 325 - non-sprayers that with project. become sprayer No. 647 1607 2775 5139 4568 4568 4568 4568 468 45(8 4568 4538 procuct ion- (sh t) 485 964 1665 2498 1370 2398 2398 2398 23Pl 2398 2298 2398 - non-sprivers th-at with project become sr)o av,,er borro5rs N / 5 400 1000 15-00 2000 3100 2400 2400 2400 24<0 2400 2400 2400 iro uct on 300 600 900 1215 930 12 h, 1260 12_60 12c<) 1260 :c C>;) BF(sh t) 1.212 1928 2940 4001 24 3 3983 3983 3983 3983 3983 353 39% TA -PR <s t 956 2007 331.1 4106 2227 43 9 35043O T,Q 50 43 I Predicted Figures INCREMENTAL PROD. (sl t) 827 1964 3165 4548 6331 7131 7791 8488 9044 9238 10879 10982 - 35 - ANNEX 5 Footnotes to the Table in Annex 5 1/ A modification of boundaries in 1970 increased the project area from 2,100 sq. km by 15% to 2,415 sq. km. The total population in the project area is increased correspondingly. 2/ SOF's have a greater cotton acreage and higher yields due to the better availability of the sprayers and due to the greater amount of assistance they receive from extension staff. 3/ There were 200 spraying farmers before the project. Their pro- duction as would have occurred "without" is based on the same assumptions as for sprayer-owning farmers "with" the project; their number is assumed to increase at a rate similar to the population growth rate for five years. Afterwards, for simplicity's sake assumptions about growth rates are dropped for production "with" as well as for production "without". 4/ Three acres per farmer, average yields as measured for non-sprayers over project period. 5/ Assuming that there would have been no borrowers without the project. 6/ Three acres per farmer, average yields as in 4/. This creates an apparent problem of consistency since for SBF the non-sprayed production from three acres is considered as production "without", while production "with" is assumed to come from two acres only. Since, however, the three acres of production area "without" represent the mean for all farmers, it appears justifiable to retain the apparent inconsistency implying that production "without" of the SOF would have come from more than three acres per farmer on the average. In any case, the order of magnitude that is under discussion does not warrant further elaboration. - 36 - Cotton Prices Annex 6 PPAR/SVADP Credit 114-MAI Scheme Gate Pricesy Border Price Net ValuesY tembala/1b Kwacha/sh t 2/ ' 3/5/6 actual estimated- actual- / estimated6' 67/68 4.4 5.6 113 113 68/69 4.5 5.6 134 134 69/70 4.8 5.6 134 134 70/71 4.9 5.6 134 134 71/72 4.9 5.6 134 134 72/73 5.2 5.6 134 134 73/74 7.0 5.6 238 134 74/75 7.0 5.6 238 134 thereafter 7.0 5.6 178 134 1/ Weighted accordinq to average grade composition. 2/ Assuming constant prices from 74/75 on. 3/ ARI; the assumption about a slight price increase over time due to quality improvement has been dropped for simplicity's sake. 4/ Based on the free-on-rail border prices of lint, cotton- seed oil, linters, cake, soapstock and husks and weighted according to the proportion of these com- ponents. The fact that in the first years cotton seed was not utilized is neglected. 5/ For the years 68/69 to 72/73 the lint price showed variations of up to 20% only. The average price was taken which was very close to the price ARII pro- jected for 1974. For the further calculation of the economic net value the procedure of ARII (Annex 15, Table 1) is adhered to (the allowance for handling, processing and marketing is K 44 per ton of seed- cotton). For the years 73/74 and 74/75 the net value is increased in proportion with the increase of the lint price by some 85%. For the subsequent years the projections of-the German team that evaluated the Salima cotton project in 1973/74 are adopted. ANNEX 6 - 37 - 6/ By ARI; the assumptions about a slight price decline and a gradual quality improvement are again neglected. The ARI assumes a border price value of K 155 per ton and an amount for handling, transport and marketing costs' which varies but is in the order of K 20 Per ton. For the sake of simplicity an average constant price rounded to be identical with the actual prices over the first five nroieCt years was used here. Value of Incremental Cotton Productionl-/ (Border Price Net Values) Annex 7 PPAR/SVADP Credit 114-MAI 67/68 68/69 69/70 70/71 71/72 72/73 73/74 74/75 75/76 76/77 77/78 82/83 87/88 Actual2/ production (sh t) 0 956 2,007 3,311 4,106 2,227 4,350 4,350 4,350 4,350 4,350 4,350 4,350 value (000 Kwacha) 0 128 269 444 550 298 1,035 1,035 774 774 774 774 774 Estimated- production (sh t) 0 827 1,964 3,165 4,548 6,331 7,131. 7,791 8,488 9,044 9,238 10,879 10,982 value (000 Kwacha) 0 1l1 263 424 609 848 956 1,044 1,137 1,212 1,238 1,458 1,472 1/ For a complete comparison of estimated and actual benefits it would have to be remembered that the appraisal report envisaged incremental maize production in addition which, however, did not occur. 2/ For underlying data and assumrtions see Annexes 5 and 6. 3/ ARI. - 39 - Farm Budgets Annex 8 PPAR/SVADP Credit 114-MAI a) Before project (1967/68)!/ - model calculation 5 acre farm 7.5. acre farm 10 acre farm 2/ Gross Revenue - maize- 13.8 13.8 13.8 3/ (Kwacha) - cotton- 39.2 88.2 137.2 - total 53.0 102.0 151.0 Expenditures - hired labour - 2.0 3.8 (Kwacha) - transport and other4/ 3.0 6.75 10.5 - total 3.0 8.75 14.3 Gross Margin-5/ (Kwacha) rounded 50.0 93.0 137.0 Labour Requirements-y(MD) 235.0 397.5 560.0 Return to Labour-/ (tembala/MD) 21.3 23.9 25.1 1/ Based on the estimates in the appraisal report but then re- conciled with the findings of the German team evaluating Salima. 2/ All farm sizes assumed to grow 3 acres of maize, average yield 400 lb/acre, price 1.15 t/lb. 3/ Remainder of farm devoted to cotton; average yield 350 lb/acre, price 5.6 t/lb (using rounded figures as opposed to the appraisal report). 4/ On the basis of K 1.5 per acre of cotton if yields are un- improved (Salima Evaluation). 5/ From which fixed costs have to be deducted yet to give family income. 6/ 65 mandays per acre of cotton and 35 mandays per acre of maize (the Manchester University survey team and the German evaluation team came up with practically identical figures). 7/ Family income ulus exnenditure for hired labour over mandays. ANNEX 8 - 40 - b) After project (1973/74) - assuminq there had been no project - model calculation 5 acre farn 7.5 acre farm 10 acre farm 1/ Gross Revenue - maize-/ 15.0 15.0 15.0 - cotton- 49.0 110.25 171.5 - total 64.0 125.25 186.5 Expenditures - hired 3/ - 2.00 3.8 labour- - transport and other _ 3.0 6.75 I0.5 - total 3.0 8.75 14.3 Gross Margin 61.0 117.0 172.0 (Kwacha) rounded Labour Requirement (MD) 235.0 397.5 560.0 Return to Labour (t/MD) 26.0 29.9 31.4 1/ Price 1.25 t/1b, yield unchanged at 400 lb/acre. 2/ Price 7.0 t/1b, yield unchanged at 350 lb/acre. 3/ Wages have practically remained constant (this may, however, change with the growing competition for labour from Sucoma sugar estates). -41 - ANNEX 8 c) After project (1973/74) for spraying farmers as projected by the appraisal reporti- - model calculation 5 acre farm 7.5 acre farm 10 acre farm GrossReveue - 2/ Gross Revenue - maize- 24.15 24.15 24.15 (Kwacha) - cotton- 112.00. 252.00 392.00 - total 136.15 276.15 416.15 Expenditures (Kwacha) - sprayer (K60 over 3 years) plus credit cost (13% p.a.)22.6 22.6 22.6 - insecticide a 11 K/acre 22.0 49.5 77 0 - hired labour - 2.5 20.9- - transport and other 6.0 13.5 21.0 - total 50.60 88.1 141.5 Gross Margin 86.0 188.0 275.0 (Kwacha) rounded Labour Requirements7/(MD) 380.0 705.0 1030.0 Return to Labour (t/MD) 22.6 27.0 28.7 1/ Again minor modifications for reasons of reconciliation and simplicity have been made. 2/ Yield of 700 lb/ac; price as before the project: 1.15 t/lb. 3/ Yield of 1000 lb/ac; price as before the project: 5.6 t/lb. (neglecting assumptions in the appraisal report which would lead to an unsignificantly higher price). 4/ 5 days for land preparation, 12 days for spraying. 5/ 15 days for land preparation, 32 days for sprayinq, 120 days for harvesting. 6/ K 3/acre at the assumed yield level. 7/ 40 MD/acre of maize (at increased yield level); 130 MD/acre of cotton (since the appraisal report does not provide the infor- mation the figures from the Manchester University Report, which are comparable to those of the Salima Report have been used. ANNEX 8 - 42 - d) After project (1973/74) for sprayinq farmers - actual coefficients - model calculation 5 acre farm 7.5 acre farm 10 acre farm Gross Revenue - m 1/ 15.0 15.0 15.0 (Kwacha) - cotton- 112.0 252.0 392.0 - total 127.0 267.0 407.0 Expenditures (Kwacha) - sprayer depreciation- 2.8 6.3 9.8 - insecticide4/ 28.0 63.0 98.0 - hired labour/ - 2.5 20.9 - transport and other 6.0 13.5 21.0 - total 36.8 85.3 149.7 Gross Margin (KwacMarond 90.0 182.0 257.0 (Kwacha) round(dd 6/ Labour Requirements (MD) 365.0 690.0 1015.0 Return to Labour (t/MD) 24.7 26.7 27.4 1/ Price 1.25 t/ib, yield unchanged at 400 lb/acre. 2/ Price 7.0 t/lbi yield 800 lb/acre (by coincidence the gross revenue is the same as for a situation with 1000 lb/acre and a price of 5.6 t/lb As assumed by the appraisal report). 3/ Knapsack sprayer; project credit price K 35; life of 25 acre seasons (K 1.4 Der acre). 4/ Assuming that recommended rates are given: K 14/ac. 5/ Wage labour r6quiremeht as in a). 6/ 35 MD/acre of maize; 130 MD/acre of cotton. - 43- ANNEX 8 e) Estimated Actual Enterprise Budgets for Cottonl/ 68/69 69/70 70/71 71/72 72/73 73/74 Knapsack Spraying Gross Margin/acre (K) 47.65 40.24 39.42 24.05 19.02 46.75 Gross Margin/MD (t) 29.8 29.2 29.2 22.3 20.5 38.3 Aerial Spraying Gross Margin/acre (K) - 34.3 35.90 19.55 - - Gross Margin/MD (t) - 23.5 24.6 16.9 - - SULV Spraying Gross Margin/acre (K) - - - - 17.68 46.66 Gross Margin/MD (t) - - - - 19.6 39.2 2/ No Spraying- Gross Margin/acre (K) 22.00 18.00 18.80 14.95 10.35 20.25 Gross Margin/MD (t) 28.9 26.1 27.3 23.4 18.2 32.7 1/ From T. Russel: Evaluation Working Paper 75/1. As opposed to some or all the model calculations before the estimated actual budgets are based on - actual prices over the years - actual yields over the years - actual inputs (e.g. in a year of lower yields less labour for harvesting is needed and transport costs are lower; actual insecticide application is usually below the recommended level. Otherwise the assumptions are by and large reconcilable with those of the model calculations. 2/ Assuming zero expenditures. Social Cost Benefit Analysis Credit 114-MAI Run 1: Actual production up to 1972/73, coq7tant at 4,350 sh t from thereon;" prices accordin to ARI projection;- actual costs up to 1972/73 at world market prices,3 from then on constant as in projections except for declining costs at project level. 67 68 1 2 3 4 5 6 7 8 9 10 11-25 Incremental Cotton Production (sh t) 0 956 2007 3311 4106 2227 4350 4350 4350 4350 4350 Benefits from " " (000 K) 0 148 311 513 636 345 674 674 674 674 674 Costs (Project Level)3/ (000 K) 70 712 554 443 489 184 184 184 168 152 136 Costs (Far level excluding labor)- (000 K) 0 22 47 84 168 134 134 134 134 134 134 Costs (ADMARC Level)5/ (000 K) - 16 34 56 69 37 73 73 73 73 73 Net Benefits (000 K) -70 -602 - 324 - 70 - 90 - 10 +283 +283 +299 +315 +331 Internal Rate of Return/Run 1: 15% 6/ Run 2: As Run 1 except that cotton production is valued at actual world market prices- Benefits from Cotton Production 0 148 311 513 636 345 1227 1227 966 966 966 Costs (Project, Farm and ADMARC Level) 70 750 635 583 726 355 391 391 375 359 343 Net Benefits -70 -602 - 324 - 70 - 90 - 10 +836 +836 591 607 623 Internal Rate of Return/Run 2: 27% ANNEX 9 - 45 - Footnotes to the Table in Annex 9 1/ Compare annex 5. 2/ Prices projected by ARI are up to 1972/73 identical with actual prices (see annex 6). 3/ Costs at project level up to 72/73 from annex 3, table 3; from the row of total costs in that table the cost of knapsack sprayers and insecti- cides has been subtracted since it is accounted for in terms of costs at the farm level. The resulting figures were accepted as a first approximation of the foreign exchange value of costs at the project level. This has the following considerations: The loan element represents 81.5% of total project costs and therein 100% of direct foreign exchange outlays (i.e. inputs valued at border prices) and 65% of local costs. This latter item may be under- stating the foreign exchange value of locally purchased items. On the other nand the figures overstate actual costs somewhat since stocks worth K 213,000 (although most of it is for sprayers and insecticides whose costs have been subtracted) were still held in year 5 and further- more, with a view to phase II additional staff costing some K 10,000 p.a. had been recruited during the second halt or phase I. Furthermore some building and construction costs could have been saved had the intention only been to maintain the level of production of year 5. From 1972/73 on a decline is assumed at market values to K 230,000 in year 6, K 210,000 in year 9, K 190,000 in year 10 and to K 170,000 in year 11, staying constant at that level. Again 80% of the figures are taken to represent the foreign exchange value of these costs. 4/ At 10 K/acre of cotton according to the actual farm budgets estimated by the Project Management of which again 80% is taken to reflect foreign exchange costs. These estimates are based on the known total quantities and costs of insecticides and sprayers. Since these are sold to sprayer owners only, these costs can only be applied to the acreage of sprayer owners (16,800 acres). It is known that they apply insecticides at a sub- optimal level there being a transfer to sprayer borrowers (and probably a fee payment back in kind or money). From the point of view of the cost-benefit analysis these transfers can be neglected. 5/ 80% of the difference between the gross border price value (K 155/sh t) and the net value (K 134/sh t). 6/ i.e. up to year 6 identical with run 1, then the net value price of K 238/ sh t increased by the ADMARC costs (K 44/sh t) i.e. a price of K 282 is used for two years. The net value price is then assumed to drop back to K 178. Adding K 44 again the price used is K 222 (compare annex 6). In annex 6 higher marketing costs are assumed than used in run 2, since here the effect of increased cotton prices alone is to be tested. - 46 - ANNEX 9 It can now be calculated to what extent the project can support cost escala- tions from year 7 on (i.e. from the year of increased cotton prices on) before the rate of return drops below a given cut off rate, say 15%. The discounted cash flow of run 2 at 15% looks as follows: 1 2 3 4 5 6 7 8 9 10 11-25 -61 -455 -213 -40 -45 -4 +314 +273 +168 +150 +900 -818 +1805 Net Present Value at 15% = +987,000 Present Worth of an Annuity Factor from year 7 to year 25 at 15% is 2.679666 The annuity that would have to be subtracted from the cash flow from year 7 on to render the net present value zero is 987,000 t 2.68 = 368,329. This means that increased annual costs of K 368,000 could be borne by the project while the economic rate of return would still be over or at 15%. The escalation of project costs by 60% as feared by the Project Management to only 22% of this margin and increased marketing costs as indicated in annex 6 would account for about the same. These cost increases would there- fore in no way threaten the economic viability of the project. An interesting question is that of the maximum opportunity costs of labor that can be borne by the project. Assuming that sprayed cotton requires 65 mandays of labor more per acre than unsprayed cotton and that 16,800 acres of cotton held by sprayer owners (neglecting the borrowers whose real labor up to opportunity costs of labor of 33 tembala per manday it is from the point of view of the national economy desirable for a farmer to take up cotton spraying instead of taking up other employment. Looking at the wages paid by Sucoma Estate going up to a Kwacha per day it appears that this figure may soon be surpassed. The above calculation carried out at a rate of return level of 10% looks as follows: Cabh flow discounted at 10% 1 2 3 4 5 6 7 8 9 10 11-25 -64 -497 -243 -48 -56 -6 429 390 251 234 1827 914 3131 Net Present Value at 10% = K 2,217,000 Present Worth of an Annuity Factor from year 7 to 25 at 10% is 4.721779 The possible cost increase K 2,217,000 t 4.72 = K 469.703 With respect to the above calculation the opportunity cost of labor could go up to K 0.42 per manday. - 47 - ANNEX 10 PPAR/SVADP Credit 114-MAI Fiscal Analysis a) Government Expenditures A rough estimate of total government expenditures in connection with the project is obtained in the following way: - Take actual project costs from annex 3, table 3 and subtract the cost of inputs financed by the farmers. For simplicity's sake the total expenditures for pest control, extension and credit are subtracted. There results an estimate of government expenditures for the project over the first five years in 000 K 67/68 68/69 69/70 70/71 71/72 Total Project costs 88 1,052 892 759 867 3,658 Gov't exp. 68 958 739 529 587 2,881 These figures are largely net of indirect taxes and duties; the salary component is net of income tax. - Estimate the necessary annual government expenditures to maintain the level of production (the estimate of the appraisal report of K 230,000 p.a. is accepted as reasonable: a smaller and slower decline than allowed in the appraisal report is assumed, however; 1/ no other deductions are made since this figure already constitutes a debatably low estimate). - Account for repayment of the IDA-credit. For simplicity again a moratorium of 10 years (including a moratorium on interest payments) followed by a period of 40 years of annuity repay- ment at 1% is assumed. b) Government Revenues - The disbursement of the IDA loan constitutes government revenue over the first years of the project. 1/ No assumption about continuing costs is clearly correct. Presumably even to maintain phase I sprayers at peak activity some government services are necessary, but if services provided in year five are held constant there- after, it is reasonable to wonder whether some spread to new farmers is attributable to these services. If so, the analysis gets tangled through double counting with the cost benefit study of phase II. The project management recommended that costs as well as the number of spraying farmers be held constant after year 5 in the phase I analysis. It the tables on pages 43 and A?, costs are reduced, but only gradually. ANNEX 10 - The main source of government revenue from the project are ADMARC profits derived from the marketing (including processing) of the additional cotton production attribu- table to the project. The border price value attached to cotton has been interpreted as a value net of marketing and processing cost. ADMARC profits are thus roughly represented by the difference between what has been called the border price value of cotton and its scheme gate value. 68/69 69/70 70/71 71/72 72/73 73/74 74/75 thereafter Border Price Net Value K/sh t 134.0 134.0 134.0 134.0 134.0 238.0 238.0 178.0 Scheme Gate Value K/sh t 90.0 96.0 98.0 98.0 104.0 140.0 140.0 140.0 Government Revenue K/sh t 44.0 38.0 36.0 36.0 30.0 98.0 98.0 38.0 Incremental Production sh t 956 2007 3311 4106 2227 4350 4350 4350 Incremental Government Revenue 000 K 41.1 76.3 119.2 147.8 66.8 426.3 426.3 165.3 It should be noted that the above figures represent high estimates for the following reasons: - costs of transporting, processing and marketing have in various years been higher than the assumed 10% of gross value (documented by ADMARC accounts of the various years). - for purposes of the national income calculation and of enterprises budgets it appears justified to calculate with constant prices for lack of evidence of change in relative prices and because inflationary price increases on the input and output side tend to cancel each other out: this is not the case with government revenue: the government will continue to be forced to increase produce prices; it is questionable whether this socio-political internal factor will be counter- balanced by a correspondinF increase in world market prices for cotton. c) The resulting fisc.1 cash *low is shown in the following table. c) Fiscal Cash Flow Attributable to SVADP Phase I (000 K) Y e a r 68 69 70 71 72 73 74 75 76 77 78 79 1980- 2018 Inflow IDA Loan 34 327 1145 457 642 479 - - - - - - Cotton Margin - 42 76 119 143 67 426 426 165 165 165 165 165 Total 34 369 1221 576 785 546 426 426 165 165 165 165 165 Outflow Project Expenditure 68 958 739 529 587 230 230 230 210 190 170 170 170 IDA Loan Repayment - - - - - - - - - - - 93 93 Total 68 958 739 529 587 230 230 230 - 190 170 63 263 Net Cash Flow -34 -589 +482 +47 +198 +316 +196 +196 -45 -25 - 5 - 98 - 98 1978 2018 = +737 = -3920 1968 1979 z 0 - 50 - ANNEX 11 If now the calculations in the ARII are correct the figures of incremental production attributable to phase II would have to be added to the figures of "production without" in order to arrive at a figure of total production from the Shire Valley (phase II area) as shown in the tabld below. 1972/73 1973/74 1982/83 1987/88 Production without phase II 18,300 27,000 31,800 37,600 Incremental production from phase II 0 18,400 18,400 18,400 Total production from Shire Valley 18,300 45,400 50,200 56,000 This is, however, contradicting the calculations in annex 2, table 3 where the production without the project phase II, at least as far as sprayed production is concerned, is assumed constant. That calculation no longer considers the fact that ARI had assumed a growing number of farmers adopting spraying on account of phase I. The following graph illustrates the discrepancies in the assumptions. - 51 - ANNEX 11 PPAR/SVADP Credit 114-MAI Development of Cotton Production in the Shire Valley According to Different Bank Reports Assuming there had only been a phase I of the SVADP the different projections can be characterized as follows: No. of spraying farmers /1 Sprayed Acreage 1967/68 1972/73 1987/88 1967/68 1972/73 1987/88 Acc. to ARI 200 4000 11000 /3 1000 19800 55000 Acc. to PCRI 300 4000 11000 /3 1500 /3 20000 /3 55000 Acc. to own estimate/2 200 4800 4961 /4 700 16800 17364 /4 /1 Sprayer owning farmer. /2 In accordance with the Project Management. /3 Not stated explicitly in the report; on the basis of 5 acres per spraying cotton farmer the figures have been extrapolated. /4 a 3% growth rate is implied of the number of spraying farmers without project is implied. It is obvious from the above taable that the PCRI confirms the projections of the ARI. Thus, 11,000 farmers spraying cotton on 55,000 acres in year 1987/88 would, according to the ARI and the PCRI, be the result of autonomous growth and the phase I project. Together with the nonspraying farmers they would pro- produce 29,300 sh t of cotton in 1987/88. With respect to phase II of the SVADP this has to be considered as "production without the project in the phase I area" (quantity one). The area of phase Ii covers the entire area of phase I plus additional acreage of the Shire Valley (see Map). In this additional area there has always been cotton production (unaffected by phase I efforts). In 1972/73 the Project Management estimated this production at 4,000 sh t and considered possible a 5% autonomous growth due to natural population growth and due to some immigration into this area. Thus by 1987/88 there would be a production of 8,300 sh t (quantity two) in the additional area without there being a phase II of the project. For the whole of the phase II area the sum of quantity one and quantity two yields the "total production with phase II of the project". This would mean 18,300 sh t of cotton in 1972/73 rising to 37.6 sh t of cotton in 1987/88. Projection of Cotton Production in the Lower Shire Valley According to Different Bank Reports COTTON PRODUCIION 000 sht b 0.0 50.0 46.8 4 2.1 4 2.3 40.0 30.0- 20.0 22 18.3. 165197b18 12. 2.9 13.4 18 4000 Sht 13.2 10. 20 10.4 ) . 9.07 .0 9.3 9.6 9.8a YEAR 097/68 5=1972173 7 10 15 20 25=1992/ - 53 - ANNEX 11 Notes to the Graph in Annex 11 a. production without project in phase I area according to ARI (a rate of increase of 3% p.a. is assumed). b. production without project in additional area covered by phase II according to project amnagement (the additional production in 1972/73 was estimated at 4,000 sh t growing - without project - at 5% p.a. since there is a light immigration into that area). c. incremental production from phase I area attributable to the phase I project according to ARI and accepted by PCRI. d. incremental production from total phase II area attributable to the phase II project according to ARII. e. production which could alone be attributed to the phase II project if production from the phase I project attained the levels projected by ARI and accepted by PCRI. f. incremental production from phase I area attributable to the phase I project according to measurements 1967/68 to 1972/73; from there on revised estimate. g. incremental production from total phase II area which should, according to the project management, be attributed to the phase II project and not to the phase I project. ECONOMIC RATE OF RETURN CALCULATION - SVADP PHASE 1 1 2 3 4 56 7 8 9 10 11 12 13 14-19 20 Year of Project- - - - - - ----- - - - ------ - - - - - -Kwacha- - ---- --- - - ------ 1. According to Appraisal Report, Annex 17, Table 1 A. Costsl 2,155.4 2,591.2 2,474.8 2,230.3 2,368.3 1,749.0 1,749.0 1,749,0 1,749.0 1,749.0 1,749.0 1,749.0 1,749.0 1,749.0 1,749.0 B. Benefits 1. Seed cotton - 550.8 q6l.8 1,451.5 1,815.0 2,244.a 2,244.8 2,042.4 2,042.4 2,042.4 2,042.4 2,042.4 2,042.4 2,042.4 2,042.4 2. Melee - 54.6 191.0 362.9 534.8 706.7 706.7 706.7 706.7 706.7 706.7 706.7 706.7 706.7 706.7 3. Rice -28.4 91.4 154.4 223.4 294.3 294.3 294.3 301.9 301.9 301.9 301.9 301.9 301.9 301.9 4. Groundnuts - 35.8 78.7 121.6 157.2 196.5 196.5 180.0 180.0 180.0 180.0 180.0 180.0 180.0 180.0 S. Cocoa - - - 0.8 4.6 12.2 26.6 37.1 44.5 44.5 44.5 44.5 44.5 44.5 44.5 6. Sorghum2 - (2.1) (6.4) (10.7) (15.0) (20.0) (20,0) (20.0) (20.0) (20.0) (20.0) (20.0) (20.0) (20.0) (20.0) Crops subtotal 667.5 1,316.5 2,080.5 2,719.7 3,434.5 3,448.9 3,240.5 3,255.5 3,255.5 3,255.5 3,255,5 3,255.5 3,255.5 3,255.5 7. Beef rattle - - 10.1 21.9 35.6 27,1 12.0 23.5 35.3 45.5 52.4 59.7 67.2 140.1 140.1 8. Fish - 14.4 69.8 160.6 329.7 478.8 478.8 478.8 478.8 478.8 478.8 478.8 478.8 478.8 478.8 Total Benefits - 681.9 1239.40 3 3, 3 3,874.4 3,874.4 NET HENEFiTS 2,00) 3.) (951.7) 31,2 827.6 2 1 1 2 2,014.3 2 91 2,102.0 2 Internal Economic Rate of Return- 227 II. Under Assumption That No Benefits Accrue from Crops Other Than Cotton MODIFIED NET BENEFITS21 (1,935.5) (2,120.5) (1,306.4) (597.8) (77.1) 978.3 903.2 772.3 784.1 794.3 801.2 808.5 816.0 888.9 888.9 Internal Economic Rate of Return - 7.4% 1/ 5% physical contingencies on all capital inputs included. 2/ Due to overall reduction in sorghum acreage, production would decrease. However, the acreage released would be planted to other crop. 3/ Benefits from crops other than seed cotton (82-86) excluded; to be accurate some on-farm costs of these other crops should also be subtracted but this is a small order of magnitude difficult to determine and has been neglected here. IBRD 10115 REPUF LC TANZANIA vronga rZAIR Proýq - Chi t ANGOLA r < AA Komnbwe A8A~. KARONGA -SyHRýC NAMIBIA i B- CHILUMBA REPUBLIC ' WAZILAND SOUTH Chiweto ARC R7pi RMmp * NKHATA .. ....... Mz mbo I \ L'i AKE ......... MAZ AW/ Nkhoto'Kora Ka3ungu Chipoto Dowa Senga Color To NACAL.A MchCiniwl :,:j. B ZOMB MAL W-L R/ BLANTYRE U.DMBE RO~DSGI~ PROJECT chLTkRwawa MJEni •- •--IINTERNATIONAL BOUNDAfRY clrom To sëSnAugust 9g72