Cotton development in Tanzania : a review of cotton program in Sukumaland

CONFIDENTIAL AFRICA RURAL DEVELOPMEN STUDY BACKGROUND PAPER COTTON DEVELOPMENT IN TANZANIA: A REVIEW OF COTTON PROGRAM IN SUKUMAIAND by M. P. Collinson Series : Studies in Employment and Rural Development No. 13 Division : Employment and Rural Development Jivision Department: Development Economics Development Policy Staff International Bank for Reconstruction and Development The author prepared this background paper in his capacity as consultant to the Bank. The views expressed represent those of the author and not necessarily those of the Bank. This report may not be published nor may it be quoted as representing the views of the Bank and its affiliated organizations. Washington, D. C., September 1974 6OTTON DEVELOPMENT IN TANZANIA: A REVINT OF COTTON PROGRAM IN SUKUMALAND TABLE OF CONTENTS Page No. PREFACE * * * * * * * * - ii SUMMARY****** ** * * * * o e 0 a * - vi I. INTRODUCTION: THE DEVELOPMENT OF COTTON PRODUCTION AMONG THE SUKMA AND NEIGHBORING PEOPLES OF WESTERN TANZA9IA . . . . . . . . . . . . . . . . . . . 1 A. The Area and People of Sukumaland. 0 0 0 0 0 1 II. THE EXPANSION OF COTTON PRODUCTION IN THE WESTERN COTTON GROWING AREAS OF TANZANIA . . . . . . . . . . . . . . . A. The Rate and Timing of the Increase in Cotton Production . . . . . . . . . . 0 0 0 0 . . a 0 0 0 0 0 5 B. Sources of the Increase in Cotton Production . . . . . 9 0. Better Husbandry and the Use of Inputs . . . . . . . . 11 D. Improved Varieties of Seed 00000000000000 11 E. Movement of Population into New Growing Areas 0 . . . . 14 F. Increased Cotton Acreage per Farm 0 0 .. . . . . . . 14 G. Intensification in Subsistence Production 0 . . . 18 III. THE REASONS FOR THE EXPANSION OF COTTON ACREAGE FROM 19o 20 A. Increases in Price and in Growers' Returns . . . . . 0 21 B. The Spreading out of the Sukama Population . . . . . . 23 C. The Development of a Cotton-Marketing Infrastructure . 28 D. The Financing of Assets for the Marketing Cooperatives. 32 E. Control of Marketing Margins and Stabilization of Grower Prices .0 . . . 0 0 a . . . . . . 0 0 0 a 35 F. Distribution of Seed . 0 . 0 0 0 a . . 0 . 0. 39 G. Board Sales Policy 0 ,0 0 0 * 0 0 0 0 0 0 0 0 0 40 H. Summary of Reasons for the Expansion in Cotton Production 0 * 0 * 0 * 0 0 0 0 0 . . 0 0 . 0 42 IV. GOVERNMENT INITIATIVES TO PROMOTE COTTON PRODUCTION IN THE WESTERN GROWING AREAS . . . . . . . . . 0. 42 A. Introduction 0 * 0 * 0 * . . . . . . 42 B. Government Efforts to Improve Cotton Yields per Acre as a Means of Increased Production 4 5 . . 0 .. C. Tie Ridging. * 0 . 0 0 0 . 0 . . * * 46 -2- D. Insecticides and Fertilizers . . . . . . . . a a . a a 47 E. Extension Organization. . . a . . c a a . a a . 0 *49 F. Credit Programs . . ..5..e** ** * 1 G. Government Efforts to Transform Production, Organization in Agriculture . . . . . . . . . . . . * 54 H. The Village Settlement Scheme: Rwamkoma . . . . . . . 7 I. Block Farms . . . . . . . . . . . 0 . a a a * a * a * 8 V. BENEFITS FROM THE EXPANSION OF THE COTTON CROP . . . . . . 61 A. Farmers' Benefits . . . . . .a 0 0 a 0 0 . . 61 B. Area Benefits . . . . . . 0 a 0 0 * a * a 0 0 * * 0 * 65 0. National Benefits . . o a . a . t - * . * *a * * 68 VI. LESSONS FROM THE SUKUMA EXPERIENCE . . . . . . . . . . . 70 A. Reinforcing or Reversing System Trends a . o . . . . . 70 B. Short- and Long-Term Potential . a . . . a a a a 71 0. The Sukuma Migrations . . a 0.0.. . . 72 D. Lessons from Government Efforts to Encourage Cotton Production . . . . . . . . . 6 . .0. 0 * 0 * * 73 E. New Varieties: A "Captive" Innovation . . . . a a . . 74 F. Lessons from the Experience with the LSMB . . . . . . 7 G. Lessons from the Cooperative Movement in Sukumaland Cotton . a . . . a . . a * * * a* * a o . o . . . a . 76 H. Lessons for Smallholder Credit Administration . . . . 78 I. Secondary Benefits Arising from Cotton Expansion . a a 81 J. Two Prerequisites for Bank-Lending Policies . . 82 BIBLIOGRAPHY 0 0 0 0 0 0 a a 1-2 LIST OF TABLES Table No. Title 1 Percentage by District of Farmers Growing Different Crops . 0 0 a a 0 0 0 0 0 0 0 0 0 0 4 2 Baleage Produced, Lint and Seed Realizations, and Producer.Pricos, 1950-70 . 0 0 0 6 3 Rice Purchases by the NAPB, 1966-71 . . . 000 . 7 4 Growth of Cotton Production by Source, 1948, 1957s 1967 . . . . . . . 0 0 0 0 0 0 06 0 0 0 6 0 0 a 10 5 The Use of Purchased Inputs on Cotton, 1962-72 . . . . 12 6 Results of an Evaluation of Yield Increases from the Use of Inputs, 1968-69 0 0 0 0 0 0 0 0 0 0 0 . 12 -3- LIST OF TABLES Table No. Title Page o. 7 Staple Lengths, 1958-62 and 1963-67 0 0 0 0 13 8 Estimated Yield in Pounds of Seed Cotton per Acre by Ginnery Zone ao o o o a a 0 a 0 0 0 0 0 1 9 Data Comparing Sukuma Land Holdings with Cotton Acreages, 1950 and 1966 0 0 0 0 a 0 0 17 10 Estimates of Family Subsistence Production, 190 and the 1960's . . a 0 0 0 0 a 0 a . . 0 0 0 0 0 0 0 19 11 Prices and Production of Cotton in the Western Growing Areas, 1943-60 . 0 0 0 0 0 . 0 21 12 Increasing Returns from Cotton as Compared to Sorghum and millet, 1944-54 . . . . . . . . 0 . 22 13 Annual Rates of Growth and Population Densities by District, 1948-67 . . . . . . . . . . . . . . . 24 14 Population and Livestock Variations among Selected Districts between 1934 and 1947-48 0 - 0 . 0 0 a 26 15 Cotton Production and Population Expansion, 195-1970 27 16 A Comparison of Land Use with Different Population Densities in Parts of Greater Sukumaland, 1965 29 17 The Increasing Role of Cooperatives . . . . . . . . . 33 18 Processors Costs 1952 and 1958. 0 . a . 0 0 00 0 0 a 36 19 Prices Paid to Farmers, Levies, and Ginning Fees, 1961-70 . 0 o 0 0 . o . 0 0 a 0 a 0 * a o 0 o 0 37 20 An Example of the Sukuma Farming System: Crops, Plant- ing Times, Acreages, Food Supply, and Cash Returns 00 0 0 0 o 0 0 0 o * o o 0 0 0 0 0 43 21 An Example of the Sukuma Farming System Labor Use (Mandays) . . . . . . . . . . . . . o . . . . . 44 22 Cost Return Ratios for Research-Station and Farmers' Results from the Use of Inputs 0 0 0 0 0 48 23 Alternative Improvements Interpolated into the Traditional System by a Linear Programming Technique . . . . . . Q . 48 24 Concentration of Contact Staff and Degree of Contact by District, 1968-69. . . . 0 0 0 0 a0 0 0 0 a . . . 0 50 25 ACA Loans, 1962-64 . . . . . . . . . . . . . . . . . . o 2 26 Hired Labor Needs Resulting from the Use of Machinery in Cotton Cultivation . . . a 00a 0 0. 0 0 0 0 0 0 0 0 6 27 Results of the Block Schemes in Three Districts . o a . a o 9 59 28 Income and Expenditure for a Smallholder Farm in Sukumaland, May 1 - October 31, 1966 o . . . 0 0 a 0 0 0 a ' 0 .0 62 29 Estimated Annual Income and Expenditures for A Smallholder Farm in Sukualand in the Late 1960's . o . . . . a 0 0 63 30 Main Sources of Cash Income for A Smallholder Farm, 1950 and 1966 *, , . o e 0 . . * . 0 . 0 0 0 0 . 0 000 0 63 31 Statistics on the Manufacturing Industry for the Western Cotton Growing Areas and for the Nation . . . . . . . . 66 32 The Growth of Manufacturing Industry Associated with Cotton Production 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 .0 0 67 33 Local Use of Domestic Cotton as a Raw Material . . . . 0 . 68 List of Maps 1 Western Cotton Areas of Tanzania Divided into Adinistrative Districts 0 * * * * * * * . * . . . . . . . . . . . . . 2 2 Water Supplies in Sukumaland (taken from D. W. Malcolm, Sukumaland) . . * . . * . . . . . . . . . . . . . . . 25 3 Ginnery Zones, Ginneries and Oil Mills, and Routes to Railheads and Ports, Western Growing Areas, Tanzania . 31 4 A Schematic Representation of the Increase in Cooperative Primary Societies in the -'.stern Growing Areas of Tanzania, 1953-70 . . 0 & . . . . . .. 3h Figure 1 Reasons for the Increase in Cotton Production, 1948-67 0 . . 8 PREFACE this report vas prepared as a background paper for the Africa. Rural Development Study (ARDS) conducted by the Development Economics Department of the World Bank. The ARDS was prompted by the very substan- tial interest in the East and West Africa Regional Offices.of the Bank in finding ways of designing rural development projects which, despite the limited financial and manpower resources available for rural development in Africa, will effectively "reachrl large numbers of low income rural people. Thirteen rural development projects and programs from sub-Saharan Africa, representing diversity in location, design and implementation, were selected for analysis (see attached list). The reviews raised a consistent set of questions regarding design, implementation and performance of these projects and programs to provide the basis for a comparative evaluation. The main report on the Africa Rural Development Study by Uma Lele summarizes the lessons that can be learned from these project reviews. 2. Each of the background papers is based on existing information, -supplemented by field investigations. With rare exceptions, no additional quantitative data were collected. Because the availability of data varied widely between projects, there are substantial differences in the scope of the individual project reviews. 3. Since most of the data were collected by other agencies, the IBRD can not attest to the accuracy of the statistical information which appears in this report. The views expressed in this report are those of the author(s) and do not necessarily represent the views of the IBRD, - ii - French Techvical Asjblnc in CamcTron: A Rv eiw of the Zones dActir1 Priorit:iren Inågrås (THI) and tbc Sociåt du Developpsmant du KIzar (SObm ') ,, Ly G. Bellonele und D. Gentil, Studies in Employiset and Rural Devlr nt Nor. 2. Eurel Doveloparnt PrograEs in Ethinpir: A Review of tbe C:D11al Abricultural Dovelopbont Unik, The Wolven Agricultural Devoloprent Unit and !he Ninimum Package Piogrus, by T. Teele. Studies in Employment and luural Development No. 3. Rnage Daeoparat in kenya: Å Review of Conercinl, Company, Irdivådual ndj Grouj Ras ,s, by E. Jönake, E. Ruthenberg, I. Thism. Studies in Fgloyent and lural Development En. 4. Sna3 Farnrr Crcdit in Kuya: A Review of Major Credit Schemes, by . D. van Fischoe. Studivs in Employment and Rural Development No5. re ellolder Tea Project in Kenya: A Review of Kenya Tea Development Anther5ty, by D. lullivru. Studios in Employmont and Rural Development No. 6. Rural DeveloprnC in Kenya: A Review of Spec0l Rural Development Propin by P. Bedi. Studies in ipleyment and Rural DDvulpment 1O. 7. Lan9 Settlemnt in Henya: A Review of the Squvtter rroblen, by P. 1M1i and C. Bares. Studies in Employtwat and Rural Bevelopment No. 8. Rural Developrent in Malawi: A Review of the Lilongwe Land Davelopment Program, by D. E. Kinsey. Studies in Esployrent and Rural Development No. 9, French Technicnl Assistner in Mali: A Review of Groundniut Operation and Cotton Scheme, by G. Anderson. Studies in Employment and Rural Developtent No. 10. Centers, by H. Turner. Studies in Employment and Rural Development No. 11. Smallholder Tobacco Development in Truzania: A Review of Ur2mbo and Turbi Scheccs, by M. Agarwal and D. Linsenseyr. Studies in Frployment and Rural Deeloprent Ko. 12. Cotton Dovelonvert in Tonzania: A Reviaw of Cotton Program in Sukumaland, by M. Collinson. Studies, in Employmrent and rural Deveop ent No, 13. Rural Development in Tanzania: A Revi r of Ujemaa, by P. Abrian and F. Tobinsan. Studi,7es in iýployent ind Utral Dvelopent io. 14. - iii - COTTON DEVELOPMENT IN TANZANIA: A REVITi OF COTTON PROGRAM IN SUKUMALAND SUMMARY 1. Sukumaland is the area of northwest Tanzania occupied by the Sukuma people and lies roughly between 2 and h degrees south and 31 and 35 degrees east. Sukuma agriculture is based on arable farming and live- stock; the two are not integrated. The farming-system comprises subsis- tence crops and cotton ,,,nd is operated mainly by family labor. 2. Cotton production has increased tenfold between 1950 and 1970, from the cultivation of an increasing acreage of cotton by each family helped by an improvement in yields per acre mainly due to better varieties. 3. The extra acreage of cotton per farm has involved a dramatic change within the farming system, substituting relatively intensive food intercrops for extensively grown grains, thus releasing labor for the cultivation of increased cotton acreage. This internal change in the system has been supplemented by a migration of Sukuma into areas with plenty of land and by the use of ox ploughs and the hiring of tractors to cultivate more extensively. 4. The rapid expansion in production was stimulated by large increases in producer prices, reflecting a shortage of cotton on world markets in the 1940's and 1950's. Together with some improvement in yield, the price increases altered the relationship between the returns from cotton growing and those from alternatives in favor of cotton. 5. This improvement in returns to cotton growing is the basic cause of expansion in production, which was facilitated by two government initiatives: (a) the development of a cotton marketing infrastructure- the Lint and Seed Marketing Board and the cooperative movement; and (b) the encouragement of the Sukuma to move into areas with an abundance of land. 6. The expansion of production would have been inhibited without easy access to market for cotton producers and particularly without a change in the settlement pattern. Internal pressures on both arable and grazing land were encouraging the Sukuma to migrate. The colonial admin- istration removed the security reason for the traditional defensive settlement pattern, and government policies of tsetse clearance and water- supply development reinforced the incentives to move. 7. Two other government initiatives, aimed more directly at expanded cotton production, were the improvement of yields per acre by better husbandry and the transformation of production organization by villagization and block schemes. Neither has significantly increased the crop. - iv - 8. The improvement of yields involved the use of capital for intensification when, in most parts of Sukumaland, the returns to increased acreage were far greater and less comp2ex managerially. This failing in program content was exacerbated by weak organization of extension and,until the late 1960Ts, no effective credit facilities for smallbolders. 9. The transformation approach through villagization failed due to the imposing of a level of welfare services that cauld not be carried by the productive potential of the agricultural technology available. This was reinforced by badly organized implementation of the schemes. Concep- tually, there was scope for mechanical cultivation in Sukuma agriculture and for the use of large blocks to avoid the problems of cultivating small fields for individual farmers, but these involved skilled management that was not available, and cut across the traditional land tenure system. Where there were people, there was no land; where land, no people. 10. The benefits of the expansion are clearly widespread. At the national level the crop became the largest single source of foreign exchange in 1966, and, latterly, use of locally grown cotton in new textile mills has given substitutes for imports. The institution building within the Lint and Seed Marketing Board and the cooperative movement have been important nationally in the dissemination of experienced staff to several associated fields. Government revenues have benefited from the increased incomes generated in the economy. 11. The farmers have clearly been the major incme beneficiaries; cash incomes from cotton have risen from approximately Shs. 40 in 1950 to Shs. 700 per farm by the mid-1960's. Farmers have been drawn more deeply into the exchange economy. Other significant benefits have accrued within Sukumaland. In addition to the establishment of part of the textile industry in Mwanza, the entire seed-crushing industry is located locally, and the ginneries have brought new prominence to small rural centers of population. Additionally there has been a buildup in service industries associated with cotton, notably in transport. Finally, there has been the proliferation of small-scale industries and of trading, based on the increased purchasing power of the local population. A. Lessons from the Development of Cotton in Sukumaland 12. Although the intensification of cotton production is in the long-term interests of Sukumaland, with the increasing population pressure on a limited land area, efforts to promote it were made within short-term economic conditions. Currently, for most of Sukumaland, intensification represents a reversal of the trend in the present development of the farming system toward larger acreages. Whereas reinforcing system trends may be relatively easy, reversing them is inevitably complex, involving the arrang- ing of conditions in the economy. Both possibilities should be based on a diagnosis of the existing system of farming and its position vis-a-vis the economy of the area. In a similar way, the reorganization of production by villagization and block schemes sacrificed the diffusion potential of the changes introduced by altering the production conditions facing the ordi- nary farmer. T-ansformation thus gave very limited penetration of the rural population at a particularly high cost. 13. Given the foreseeable political repercussions from the cultural division between trading and farming classes in Sukumaland, the LSMB played a useful role in ensuring that the cotton industry was not sacrificed in the course of the clash that arose. However, this could not have been a consideration when the board was formed. The LSMB also insulated the cooperatives from trading activities, since commercial operations were normally their downfall. Once the political battle was won, cocperatives were given a monopoly in marketing. Experience here shows that both the board and the cooperatives were successful only because of very special circumstances; doubts must remain on the success of their replicability where there is commercial competition within the local community. It is important to emphasize the lack of local African competition in both trading and marketing. 14. Once established, marketing cooperatives, which by their function penetrate the rural community to the individual farmers, are a useful vehicle for the administration of smallholder credit, particularly when they hold a monopoly over produce marketing. In the western growing.areas the democratic element in primary society organization has been used to create pressures on borrowers to repay their loans. However, no mechanism will succeed unless there clearly is a benefit to growers from the practices being encouraged by credit. Political enthusiasm cannot be a long-term substitute for a genuine demand for farm requisites. 15. Three main points must be kept in mind for IBRD lending policies: Changes in practice that are appropriate to farmers' priorities and their economic circumstances, spread rapidly throughout the community. Diffusion multiplies the benefits of the programs that find and introduce such improve- ments. There should be more effort in evaluating the diffusion potential of available technology, and this requires knowledge of the economic rela- tionships governing resource allocations on existing farms and the producti- vity of these allocations. 16. The high infrastructural charges on village settlements are an example of the need for guidelines in allocating project overheads between direct beneficiaries and the government. In principle, if the government chooses to raise revenues for development expenditure, the programs selected should be the best way of employing the'public resources involved and should not be recoverable from the direct beneficiaries. In practice, the criterion of diffusion potential can aid the decision. If the project isolates the direct beneficiaries from the farming community, diffusion potential is lost and there is a strong case for charging all scheme cr- heads to the direct beneficiaries. Even then tpx revenues from project inccmes and from associated multiplier effects should be benefits offsetting the government's contribution. When the changes introduced have potential for diffusion, increases in tax revenues will usually be many times the government contribution. Direct beneficiaries should not be penalized for acting as catalysts for developnent. -vi - 17. Accounting conventionE of disbursement projections, often based on necessarily tenuous estimates made in appraisal, should not be allowed to compromise the institution-building objectives. RURAL DEVELOPMENT IN TANZANIA: A REVIEW OF COTTON PROGRAM IN SUKEMALAND I. INTRODUCTION: THE DEELOPMENT OF COTTON PRODUCTION AMQNG THE SURMA AND NEIGHBORING PEOPLES OP WESTERN TANZANIA 1. Between the late 1940's and the late 1960's the production of cotton in what are often called*the "western growing areas" in Tanzania increased tenfold, a rate of growth rarely observed in agriculture. With the current emphasis on rural development as a basis for economic growth in Africa, the circumstances of this rapid expansion offer pointers for planning and policies, and its repercussions on the economy indicate the need for wider criteria in rural planning. It should be made clear at the outset that this development was not the result.of a planned and imple- mented program, although prior to and during this period the government did take five major initiatives to promote cotton production: (a) cleared the tsetse menace and provided domestic water points; (b) established a cotton research center; (c) established a cotton-marketing infrastructure; (d) funded an extension service concerned with the improvement of cotton by the use of improved techniques; (e) embarked on the structural trang- formation of the agricultural sector to improve production organizatioll. Only the final initiative, part of a country-wide strategy, prompted by the International Bank for Reconstruction and Development's report on The Economy of Tanganyika (1960), is properly identifiable as a project, bE TV~ -wasih_OF-1Tfve_T._ 2. This review examines the impact of these five major initiatives taken by the Government to promote cotton production, it also describes the expansion of cotton production after 1950 and attempts to identify the reasons for it and to assess the repercussions at the farm, area, and national levels. A.TeAe and People of Sukumaland 3. Sukumaland is the area in northwest Tanzania that is ocupied by the Sukuma people and lies between 2 and 4 degrees south and 31 and 5 degrees east, bounded on the north by Lake Vidtoria. Map 1shows the six administrative districts of Sukumaland Proper, with Nzega, Musoma, and Biharamulo in addition. Greater Sukumaland, wh ch in#cudes parts of.these three districts settled by Sukuma, has over .5 million people. Road communications are good, though Smith Sound forms a substantial barrier to east-west flows with only two ferry crossings. The Central Line branch to Mwanza from Tabora and Dar es Salasm runs sotth\to north, and railway steamers continue the service to ports on Lake Victoria. produce from the -2- Lake Victoria 0 usoma cn iDistrict Cerewe S ict ~rnza Distri Bihla amfulo Geita Di Lrict, District District I Kwirba District i1 RallwhinyangAyag District boundaries District Boundary; Sukumaland proper ea Nzega District Ia : Western Gotton Areas of Tanzania Divided into Administrative DistPice -3- area--mainly cotton, lint for export, and cottonseed for Tanzanian urban markets--are moved out by rail. 4. The lake is 3,700 feet above sea level, and much of Sukumaland is a plateau at about 4,000 feet, though falling below lake level in the south and east. The lakeshore areas have 35 to 40 inches of rainfall each year, and the southern districts 25 to 30 inches; here rainfall is also less predictable. The area has a unimodal pattern of distribution, rain falling from November to May. A slight trough in January and February is a reminder of the bimodal distribution further north. June to September is a dry period often with no rain, particularly in the south and east. This rainfall distribution influences the pattern of Sukuma agriculture. 5. Topography is everywhere undulating with a catena soil sequence, from sands to alluvial, black, cotton soils that have a high clay content in the valley bottoms. The topography is more pronounced in the northern districts where the catena often begins in a granitic outcrop. In the flatter topography of the sath often only the middle ranges of the catena are found. 6. The population is overwhelmingly rural; Mwanza is the only town with over 10,000 people, having 34,800 at the 1967 census. Settlement is predominantly by farmers on their own land; small villages provide local trading and administrative facilities. Farms run from top to bottom of the catena giving each farmer a cross section of local soils and an opportunity to grow the varied selection of crops making up the Sukuma consumption pattern. In the more densely populated areas settlement is now permanent; in other areas families are likely to relocate their homesteads at intervals of from seven to ten years, moving closer to better land. This shifting tradition perhaps helps the migrations to the peripheral districts once the older areas become densely populated. 7. Land is held in usufruct under the control of the village headman. In areas where land is plentiful a family can have rights of use to all the land it can clear. The headman retains the authority to reallocate fallow land to new families as the local population increases. Flexibility in tenure and communal authority in land matters have persisted despite the removal after independence of most of the headman's authority to make way for more formal government representation at the local level. 8. Cattle play a large part in Sukuma life, but are divorced fra the arable agriculture. Again the headman and village elders control areas of communal grazing, reserving areas of grass for the dry season and allowing cattle to overgraze crop residues after, harvest. The cattle holdings in addition to the homestead housing of most Sukuma families are part of a generation link, usually extending through three generations of the family. Sons have rights to their fathers' cattle as soon as they are born, thus binding the generations together. 'Met and milk also play an important dietary role, and often several households will meet to work for a farmer who will then kill an animal and provide a meal of meat. (Other ―ㅢ 5 Farms are worked mainly by family labor. The rainfall pattern dictates the cropping season, ard the Sukuma have dovetailed their crop activities to satisfy their food and cash needs from the labor supplied by the family. The planting and weeding period, from December to February, is critical and limits the size of an area the farm family can cultivate. Many families supplement their efforts by hiring casual labor during part of this period. Again, differences in rainfall and population density between north and south bring variations in the way resources are used to grow crops. In the north, particularly on the lake islands where density is highest and rainfall best, crops of sweet potatoes or legumes follow rice in te season. In the south, where there is still ample land but where the planting period is more ccnoentrated due to the lower rainfall and a shorter season, families use ox: ploughs and hired tractors to estab- lish a crop sooner. Use of machines for cultivation is more practical because of the flatter topography; so a higher proportion of the arable acreage is planted on flat seedbeds rather than on the five-foot ridge characteri6tic of the older areas. Flat planting gives relatively poor weed control and rapid postplanting weed emergence. The labor peak in these systems shifts to a later month: Heavy hand weeding is substituted for the heavy m1tivation required by the five-foot ridges. Further details on resource use and incomes will be given later. II. THE EXPANSION OF COTTON PRODUCTION IN THE WESTERN LOTTON GROWING A EAS OF TANIZANIA A. The Rate and Timing of the Increase in Cotton Production 12. When the Germans occupied Tanganyika in 1885 they found Peruvian types of cotton in the western growing areas (then the lake region). Pro- duction there was encouraged by both the German and later 'the British administrations, but it remained around 7,000 bales. 1/ until 1930. BY 1935 it had risen dramatically to 40,000 bales. It r emained around this level until the early 1950's. when production rose from 39,,500 bales in 195n to 405,,000 bales by 1970. Table 2 shows the history of the crop since 1950 and gives the earnings from exports of lint and sales of seed when available. By the late 1960's cotton lint accounted for 20 percent of the value of Tanzania's exports; in 1966 it was the country's most valuable export commodity. Since 1966 production has been on a plateau, reaching the 400,000-bale mark only once again, in 1970. Because Tanzania's second five-year plan sees further cotton expansion as a major source of foreign exchange, there has been considerable concern at this fall off, but the concern is not necessarily justified. Projections for the crop were made in 1968 from the very high results in 1966. But 1966 was a short- teym high compared to the general trends; a three-year average would have proved a better base. Other similar short-term peaks occurred in 1952 and 1/ A bale in this review is a standard 400-paund lint. -6- Table 2: BALEAGE PRODUCED, LINT AND SEED REALIZATIONS, AND PRODUCER PRICES, 1950-70 Raw Lint Total Seed Lint Export Seed Seed Earned Cotton/a Exported Price Sold /b Price Yearly Year (cts./1b7 (tons) (cts./lb. (tonsT (Shs./ton) (Thousand f.o.r.) Shs.) 1950 25 39,100 N.A. N.A. N.A. N.A. 1951 34 41,600 N.A. N.A. N.A. N.A. 1952 46 71,000 N.A. N.A. NIA. N.A. 1953 46 44,500 220 (Est) 12,000 (Est.) 245 42,100 1954 57 90,000 217 21,300 (Est.) 377 87,281 1955 58 110,100 250 29,75o (Est.) 352 120,72 1956 53 119,000 228 32,162 (Est.) 334 119,270 1957 52 152,000 194 35,094 373 131,042 1958 52 150,6oo 210 46,315 311 140,908 1959 49 183,400 189 45,308 231 149,116 1960 52 162,900 185 5,696 362 140,708 1961 56 161,600 199 46,078 353 144,899 1962 56 196,200 198 47,253 405 174,527 1963 51 238,500 186 54,560 354 196,528 1964 47 269,50o 188 66,573 313 223,501 1965 45 338,400 187 105,000 (Est.) 353 327,588 1966 43 405,500 179 93,639 524 339,40 1967 43 374,000 168" 113,524 356 291,473 1968 45 273,000 184 73,800 (Est.) 486 236,495 1969 47 384,000 168 104,000 (Est.) 420 (Est.) 301,728 1970 47 405,000 181 109,000,(Est.) 420 (Est.) 339,000 Sources: Ministry of Agriculture (Mwanza) and Tanzania,"Annual Reports of the Lint and Seed Marketing Board, Tanzania,"1957-67 (hereafter referred to as the LS14B reports). Reliable information is contained in the LSB reports, 1956-57 to 1966-67 (available from the board). For other years no two sources give the same data though variations are usually small. Some confusion is caused by the board's financial year; for example, the 1957 plantings are covered in the report for 1958-59, the year the board handled that crop. Notes: Est. = Estimated. N.A. = Not available. a Producer prices are averages for Grades A atid 0. /b Estimates of seed sales are based on 37 percent of the(-bales sold in the same year. -7- 1959, Similarly, other plateaus have occurred within the expansion period, the period 1957-61 is an example. 13. Many have said the increased rice price has diverted farmers from cotton production. Rice prices rose from Shs. 45.75 per bag of 75 kilos for the 1965 harvest to Shs. 51 per bag for the 1966 harvest and Shs. 56.25 thereafter. (See Table 3 for paddy purchases by the National Agricultural Products Board /APB7 in the western groing areas.) Table 3: RICE PURCHASES BY THE NAPB, 1966-71 (75 kilo bags) Area 1966 1967 1968 1969 1970 1971 Mwanza 1,751 N.A. 29,473 6h,760 148,393 17,238 T7nlampaka ** ... ... 17,085 54,911 7,634 Ukerewe 16,750 10,428 19,747 h0,50 34,869 11,399 Musoma 15,782 17,727 66,531 57,6)1 133,775 50,562 Shinyanga 18,094 6,111 60,785 26,336 77,604 37,749 Nzega 117,891 41,298 2114,136 194,L09 226,605 157,552 Total 169,471 ,96147 obl$,992 400,704 676,157 282,134 Estimated Acres/P 21,200 /9,14007 52,000 50,000 85,000 35,000 Source: NAPB recordo. /a Estimated acreage uses a standard yield of eight bags per acre, In fact, the variation in yields from year to year will reduce the apparent acreage variation. Cotton and rice acreages are normally found on the same farms. Larger quantities of rice are grown for subsistence eaph year withiin the area, and increased sales may merely imply a shift from rice as a subsistence grain to take advantage of the higher prices. But assuming the figures do demon- strate an increase in acreage over the period, it is relatively insignifi- cant compared to the fluctuations in cotton production over the same period. At the extreme in 1970, there was an increase of 60,000 acres of rice; more generally the increase has been of 30,000 acres. Cotton and rice do compete for labor in the farming system, rice being planted in January when late cotton is also planted and early cotton weeded. If a one-for-one substi- tution were made it would imply a loss, usually of some 30,000 acres of 350 A 300 250 Thousand bales 200 150 100 50 c 1948 1952 1956 1960 1964 1968 Years A = Improved average yield per acre B = Inorease in acreage per farm C = Inorease in number of gröwers (base production 1948) Figure 1: Reasons for the Increase in Cotton Production, 1948-67 cotton or 3 percent of the area; at the extreme, 60,000 acres or 6 percent. Neither figure is significant in relation to variations of 20 to 35 percent. experienced in cotton production. Secondly, in Nzega, which consistently produces half the rice purchased, most of the crop is grown in an area where cotton, as an alternative, has not yet fully penetrated. 14. K. G. Brown 1/ gives a highly significant R2 eqtal to 7.5 percent for cotton yields and planting rainfall. It would be interesting to relate December rainfall to the total crop yield over time to see if this relation- ship holds over the aggregate. The key factor affecting variations in yields is almost certainly rainfall. The fact that the amount of seed distributed has continued to rise after 1966 is indicative that plantings are still increasing, although the rate of increase has fallen. (The amount distributed increased by 10,200 tons from 1961 to 1966, and by 8,900 tons from 1966 to 1972.) B. Sources of the Increase in Cotton Production 15. There have been three reasons for the rapid increase in cotton production in the western growing areas; an increase in the number of farmers growing cotton, a larger acreage of cotton per farm, and an improve- ment in yields per acre. Table 4 presents a summary of these factors for the three years 1948, 1957, and 1967. Data has been examined from several sources and the figures in the table are, at best, a compromise. 16. The most important source of increase is in the acreage of cotton grown per farm--a fivefold increase over the twenty-year period. Additionally, 55 percent more farmers were growing cotton at the end of the period, mainly due to an increased farmer population but partly due to an increased proportion of the farming community adopting cotton as a cash crop. A forty-percent improvement in per-acre yields of cotton over the period is made even more significant by the increase in the acreage grown. Population increase is a self-evident source of expansion; yield improvement and the increase in acreage grown per farm will be examined in more detail. 17. Three sources of higher yields can be identified: (a) better husbandry including the use of fertilizers and insecticides; (b) improved varieties of seed; and (c) movement of population into new areas with relatively fertile soils. K. G. Brown, "Rainfall, Tie Ridging and Crop Yields," Empire Cotton Growing Review, January 1963. - 10 - Table 4: GROWTH OF COTTON PRODUCTION BY SOURCE 1948, 1957, 1967 Item 1948 1957 1967 Baleage produced (three-year average) 34,000 162,000 372,000 Population of western-areas (in thousands) 1,5o6 1,8o6 2,489 Family size 5.6 6.3 6.7 Number of farmers 269,000 287,000 371,000 Percentage growing cotton 75 87 85 Number farms growing cotton 203,000 249,000 315,000 Acreage per grower .67 2.20 3.36 Yield per acre (lbs. seed cotton) 300 355 421 Pounds seed cotton per grower 200 780 1 ,420 Sources: The data for items listed below are taken from the sources indicated; other data are calculated from these figures: Baleage produced-- Ministry of Agriculture (Mwanza) and lMB reports; Population--Tanzania, Central Statistical Bureau (CSB), '"Recorded Population Changes 1948-67V 1968; Family size--1948: calculated from taxpayer base of 1950, found in Tanzania, East African Statistical Department (EASD),"Sample Census of African Aericulture 1950,"1953; 1957 and 1967: 6ample survey of agri- culture, found in Tanzania, OSB,"ReDort on the 1966 Survey of Cotton Growers in the Lake Region,1968; Percentage growing cotton--1967: Bureau of :fesource Assessment and Land Use Planning (BRALUP),'"Preliminary Report of the Sukumaland Inter- disciplinary Research Project,111971; Acreage per grower--1948: N. V. Rounce, "The Economic Development of Sukumaland,l Empire Journal of Experimental Agriculture, 1951; 1957: interpolated on a straight-line basis between 1948 and 1967; 1967: CSB, "Aeport on the 1966 Survey of Cotton Growers;"Yield per acre-- 1948- Rounce survey of yields in T949, rep6rted in EASD, "Sample Census of African Agriculture 1950;11957 and 1967--derived from production per grower and acreage (BRALUP, tPreliminary Report,41 gives a yield of 490 pounds per acre for a sample of 200 farms in the 1969 harvest year; the Ministry of Agriculture claims an average yield of 400 pounds per acre.) C. Better Husbandry and the Use of Inputs 18. Of the three factors increasing yields, improved husbandry is the only one requiring positive initiative on the part of the farmer. Clearly the movement to new areas requires initiative but, as is discussed later, it was probably triggered by a need for more land, particularly for grazing. An organized program for the distribution of fertilizer was started in 1962, for insecticide only in 1968. Table 5 shows how much fertilizer and insec- ticide were issued on credit over the period 1962-63 (1963 harvest) to 1971-72. Purchases by individuals or other groups were very small through- out the period, although some of the block farms (see Chapter 4) were sprayed in 1964-65 and 1965-66. The figures shown for 1964-65 and 1965-66 are inflated by the use of fertilizer on block schemes, part of the central government's strategy to transform agriculture. Similarly, 1966-67 was the first year an inventory was taken of fertilizer in stock, and the accumu- lated losses due to rehandling and unrecorded issues are reflected in these figures. The 1970-71 figures represent a more accurate level of usage, mainly in the northern parts of the western growing areas; they show that some 23,000 acres were covered with fertilizer and the use of insecticides rapidly increased over the five-year period. 19. Tests at the Western Research Center in Ukiriguru near Mwanza and their results tested on a field scale on a unit farm have suggested an increment in yielc of 400 pounds of seed cotton per acre from the use of fertilizer alone, and 60 pounds per acre from the use of fertilizer with insecticide. However, these increments have not been substantiated by farmers' results from using inputs. Taking the figures in Table 6 (assuming that when insecticide is used so is fertilizer) and combining them with the acreages covered in 1971, the total contribution from the use of inputs is estimated to be approximately 9,200 bales-of lint, relatively insignificant in terms of the overall increase since 1950. D. Improved Varieties of Seed 20. Cottonseed has always been issued free to growers. Gradually an efficient program of seed multiplication and distribution has been built up from the cotton breeding work done by the Cotton Research Corporation at the Western Research Centre in Ukiriguru; and a 'path has been created for the automatic introduction of improvements to the farmer who sells unginne d cotton and must collect new seed each season. Varietal improvement on the ortginal MZ 561 strain began under the auspices of the then Empire Cotton Growing Corporation staff in 1933. Initially selections were made, and then, after World War II, crossings with outside material allowed the intro- duction of desirable characteristics. New seed issues from the plant- breeding work at Ukiriguru were made in 1946 (UK 46), 1948 (UK 48), 1951 (UK 51), 1955 (UK $5), 1961 (UK 61), 1963 (UK 63), and 1964 (UK 64); the year denotes the time of initial bulking at the research station. -12- Table 5: THE USE OF PURCHASED INPUTS ON COTTON, 1962-72 1962- 1963- 1964- 1965- 1966- 196f- 1968- 1969- 1970- 1971- Inputs 63 64 65 66 67/a 68 69 70 71 72/b Fertilizer Sulphate ammonia (tons) 50 165 1,250 1,h15 1,815 1,312 1,232 830 1,132 1,019 Single Super- phosphate (tons) 100 335 2,500 2,835 1,869 1,367 662 190 730 514 Insecticide (acre packs) --- --- --- --- --- 11,400 17,962 28,502 35,257 38,357 Pumps --- --- --- --- --- 931 3,14 3,02 4,577 3,847 Sources: "Annual Report on the Administration of IDA Credit TA 80 for year ending March 31, 1969 (up to 1967-68),11and"National Development Credit Agency Reports," (Dar es Salaam). /a Recommendation changed from one awt. of 21 percent N sulphate ammonia and two-owt. single superphosphate, to one cwt. of each type per acre. /b Assumes mara figures at the same level as 1970-71; 1971-72 figures not yet available. Table 6: RESULTS OF AN EVALUATION OF YIELD INCREASES FROM THE USE OF INPUTS, 1968-69 No Fertilizer Insecticide Fertilizer & Area Inputs only only insecticide Mwanza Number in sample 20 20 20 19 Pound seed cotton per acre 778 832 990 1,126 Increment -- 54 212 348 Musoma Number in sample 38 -- 34 0 Pound seed cotton per acre 576 -- 663 949 Increment -- -- \ 87 373 Source: Field evaluation of the use of fertilizer and insecticide purchased under Category 1:"Annual Report on the Administration of IDA Credit, TA 80, and NDCA,t,1969. The report notes that the absolute levels of yield are likely to be exaggerated due to using ridges as plots and only harvesting 1/100th acre. -13 - 21. The major improvements introduced by the braeding program are: (a) resistance to Jassid by incorporating leaf hairiness (to some extent in UK 46, more definitely in UK 48); (b) resistance to bacterial blight by incorporating pats of Albar 51 from earlier work by the corporation in Uganda (this work has been continuous; reisstance was incorporated, in part, in UK 48); (c) improvement of yield levels; (d) improvement of ginning outturns and the ratio of lint to seed cottaiby weight (an improve- ment of from 32 percent to 33.5 percent was noted by 1966, with varieties then under evaluation showing a 35 percent outturn); (e) latterly, since 1960, an increasing emphasis on finding a relection resistant to Fusarium, wilt waich has spread outward from the lakeshore through the northern end of the growing area (the UK 62 seed issue was abandoned due to its sus- ceptibility to Fusarium);(f) an improvement of, and standardization in, staple length (see Table 7). Table 7: STAPLE LENGTHS, 1958-62 and 1963-67 Staple Length Five-year periods (32nas of an inch) 1958-62 1963-67 Percent 32 4.8 ... 33 26.8 7.0 34 47.9 77.8 35 18.3 13.8 36 2.2 1.4 37 ... ... Source: 0. K. Laurent, "Cotton Marketing in Tanzania," mimeo. Laurent makes the point that if the price difference between staple lengths 33 and 34 is three cents per pound of lint, the extra revenue from the longer staple, only 7 percent below 34 in 1963-67 and 31 -! percent in 1958-62:, almost covers the board's annual outlay on research. 1/ In addition, the more consistent staple length is a favorable buying p-int. Volume 3e of the ire Cotton Crowing Review (1961) records the results of 55 variety trials over several seasons all with a coefficient of variation lower than 25 percent, showing UK 55--still the variety planted on over half the acreage grown in 1 967--to hav-e a 31 percent yield advantage over MZ 561. 1/ 0. K. Laurent, "Cotton Marketing in Tanzania," mimeographed, 1969, -1)4 - E. Movement of Population into New Growing Areas 22. The movement of population from Central Sukumaland into peri- pheral areas increased the availability of land and was a major element leading to the increase in the acreage of cotton grown per farm. It will be described below in considering the reasons for the expansion, so it is sufficient to say here that lands were taken into arable cultivation that were more fertile than the traditional hillsands of Central Sukumaland, and at the same time, cotton yields were improved. Table 8 gives estimates of the differences among areas in the yield of seed cotton per acre. The table shows the district, the ginnery zones (within each district), and the estimated yield per acre. Several of the ginnery zones fall into more than one district, but for the purpose of this table each zone has been attributed to the district in which it has the most buying posts. (Map 1 shows the administrative districts of Sukumaland.) 23. In summary, of the three sources of higher yields, very little contribution to cotton expansion has come from the use of inputs and improved management. Most is from the use of improved seed as a result of the cotton-breeding research at Ukiriguru. The unknown element is the use of better soils in the newer growing areas, and two points indicate that this element may not be too important: (a) It will be offset by deteriorating fertility in the older cultivated areas; and (b) Ybthods of cultivation in the newer, southern areas are predominantly extensive, and husbandry is poorer than on the northern hillsands. So, overall it can be fairly concluded that the breeding of improved varieties is probably responsible for some 25 percent of the present crop. F. Increased Cotton Acreage per Farm 24. Over the period 1948-67 there was a large increase in the area of cotton grown by each farmer. According to Tanzania's Sample Census of African Agriculture 1950 (using a ratio of 1.35 taxpayers pr hoselold), te -average acreage- ofcotton was 0.68 1/ for the 22 zones covering the districts of Mwanza, Kwimba, Maswa, and7Shinyanga. The largest cotton acreage in any zone was 1.37 acres. Tanzania's"Report on the 1966 Survey of Cotton Growers in the Lake Region"gives an average of 3.36 acres grown per household and, from a sample of 120 farms in Shinyanga, 7.03 acres. This general level is corroborated by farm management studies by 1/ Similar to the average acreage indicated in N. V. Rounce, "The Economic Dpvelopment of Sukumaland,1 Empire Journal of Experimental Agriculture, 1951. Table 8: ESTIMATED YIELD IN POUNDS OF SEED COTTON PER ACRE BY GINNERY ZONE District Zone Yield District Zone Yield District Zone Yield Mwanza Bukumbi 350 Geita Buchosa 00 Maswa Luguru * 600 Nassa 5 500 Nyamililo 350 Sola 4 h50 Buyagu 400 Ngasamo 5 500 Kwimba Manawa 400 Kasamwa 5 550 Nyamnbiti 450 M4alampaka 400 Musoma Kibara 400 Shinyanga Mhunze * 600 Maga 600 Mugango 350 Usogore * 550 Ushashi 400 Nzega Nanonga * 600 Ukereve Ukerewe 300 Biharaulo Chato * 550 m Kahama Kahama [00 Source: Ministry of Agriculture (Mwanza), 1970. Note: The asterisked zones took an influx of population from the older areas. - 16 - M. P. Collinson, 1/ D. Rotenhan, 2/ and the Bureau of Resource Assessment and Land Use Planning. 3/ 25. However, the evidence available does not suggest an increase in the average size of land holdings. The 1950 sample census puts the average at 7.70 acres, which is almost exactly mirrored in the 1966 sample survey's average of 7.65 cultivated acres per holding. 4/ Furthermore, although there are wide differences in holding sizes fron area to area, the range in hold- ings is almost identical for 1950 and 1966. The only factor qualifying this is that the areas with large holdings are also areas of immigration. By 1966 there was a higher percentage of the population in southern and eastern areas. Given that holding sizes have probably remained about the same, the increase in cotton acreage must have been made possible by a change in the system of agriculture practiced: The change occurred in the proportion of cultivated areas devoted to cotton and food crops between 1950, when the average unit under cotton cultivation was only 9 percent, 5/and 1966, when it rose to h percent (see Table 9). 26. The fact that the increase in cotton acreage per farm was the result of internal changes in the farming system demonstrates the inherent flexibility of a traditional system of agriculture in absorbing changes. It raises the questions of why, given such flexibility, there has not been enthusiasm for the use of fertilizers in particular and improved husbandry in general. It also raises the question of why the change accelerated in the 1950's and not before. But before examining why the changes occurred at this time, it is interesting to look briefly at what happened to the sub- sistence aspect of the farming system in response to the increase in area under cotton cultivation. 27. Apparently the change in the farming system has been one of intensification, the substitution of high-value for relatively low-value crops. With an average family size of approximately five to six persons in 1950 the food-crop area then cultivated was 1.25 acres per head. In 1966 the family size had risen to 6.7 but the acres of food grown per head had 1/ M. P. Collinson, "Farm Management Surveys 1-4:" "Bukumbi Chiefdom," 1961; "Usmao Chiefdom, Kwimba District," 1962; 1Luguru Ginnery Zone, Maswa District," 1963; "Nyampande Cooperative Society Area, Gaita District," 1964. 2/ D. Rotenhan, "Cotton in Sukumaland," in Smallholder Agriculture and Development in Tanzania, H. Ruthenburg, ed., vol. 24., Africa tUdi as Series (Munich: IFO, 1968). 3/ BRALUP, "Preliminary Report of the Sukumaland Interdisciplinary Research Project," Report No. 40, 1971. ' Rounce, "The Economi6 Development of Sukumaland.." 5/ Ibid., which gives 11 percent cotton. -17 - Table 9: DATA COMPARING SUKUMA LAND HOLDINGS WITH COTTON ACREAGES, 1950 AND 1966 Item 1950 1966 Average acreage cultivated 7.70 7.65 Smallest average acreage in district/zone 3-50 3.80 Largest average acreage in district/zone 14.70 14.80 Acreage of cotton grown 0.68 3.36 Percent of cultivated area in cotton 9 [4 Acreage of food crops grown 7.02 4.29 Percent of cultivated area in foods 91 56 Sources: Tanzania, EASD,"Sample Census of African Agriculture 1950;" Tanzania, CSB,"Report on the 1966 Survey of Cotton Growers in the Lake Region." Note: In arriving at acreage data for 1950, which are presented in the report as TIatios" per taxpayer, the ratios are mltiplied by 1 .35 to give areas per household. These results are close to those given by Rounce in 1945, and the multiplier of 1.35 is supported by the 1966 study that found 1 .39 taxpayers per house- hold. - 18 - decreased to 0.64, about half the former level. This, in conjunction with the increase in cotton acreage, implies either that cotton was grown with resources hitherto devoted to the production of surplus foodstuffs, or that food production intensified. 28. Rounce indicates that crop sales in 1948 averaged about 314 pounds, of which about 180 pounds were cotton. 1/ (Cotton sales, at 25 cents per pound of seed cotton, were valued at Shs. 4T per family, while other crops contributed an additional Shs. 13.) The-sale of 130 pounds of food- stuffs was insignificant in terms of total food production, between 5 and 10 percent. On a proportion of farms, it probably represented grain exchanged for cattle and disposals for local brewing. The evidence ,then, suggests small sales of food crops, no more than would occur under the present system in a year when the grain acreage planted as an insurance against poor rains represents a surplus to subsistence requirements. G. Intensification in Subsistence Production 29. The evidence does suggest that food-crop production intensified. The traditional grain staples of the Sukuma were millet and sorghum. Of the 22 zones surveyed in the 1950 sample census, the 13 with the lowest population densities (predominantly in South Kwimba, Maswa, and Shinyanga) still sh'wed a dominance of these crops; these zones averaged 4.72 acres of either sorghum or millet in pure stands out of an average crop area of 8.37 acres. In other areas where population density was highest (such as in Mwanza, North Kwimba, and East Geita) and where population pressure had already shortened the resting period of the land, extensive growing of millets had given way to increasing dependence on cereal-legume-root mixtures. Farms in the remaining nine zones of the 1950 survey averaged 0.96 acre of pure millet or sorghum and 1.73 acres of cereal-root-legume mixtures. Studies in the 1960's showed that this change had spread to all areas, with maize as the dominant cereal in the mixtures. The 1966 sample survey showed average acreages of 1 .89 for maize and maize-dominated mixtures and 1 .16 for millet. A similar trend was seen in the increased dependence on roots; all roots, either pure or intercropped, averaged 0.75 acre in 1950 compared to 1.21 acres in 1966. D. Rotenhan suggests maize was substituted at the expense of the more reliable millets and sorghum. 2/ But the poor weather conditions in the late 1960's have led to a revival - of sorghums and millets as well as to a plateau in the upward trend of cotton production. 30. Due to the susceptibility of yield data collected in a single season to seasonal climatio variations, ccmparison of families' food supplies in proportion and quantity in 1950 and 'the late 1960's is difficult. Table 10 gives food-supply proportions derived from farm 1/ Rounce, "Economic Davelopment of Sukumaland.t 2/ Rotenhan, "Cotton in Sukunaland.1 - 19 - Table 10: ESTIMATES OF FAMILY SUBSISTENCE PRODUCTION, 1950 AND THE 196,t's 1950 1960's Food 'I-- - - - - - - - - - - - lbs./family lbs./person lbs./family lbs./person Maize /a 211 38 1.,240 185 Rice 180 32 494 74 Sorghum 1,951 348 265 40 Millet /b 1,16 /b 205 25 4 Foodgraln 3,M88 623 2,024 303 Roots /C 2,625 469 h,235 632 Nuts ag pulses -Not recorded; all intercropped 204 31 Sources: 1950 --Tanzania, EASD,"Sample Census of African Agriculture 1950,tt p. 31; 1960's--M. P.Collinson,"Farm Management Surveys 1-4 (1961-64): Bukumbi Chiefdom; Usmao Chiefdom, Kwinba District; Luguru Ginnery Zone, Maswa District; and Nyanpande Cooperative Society Area, Geita District." Notes: Yield figures as shown have been used for pure stands and have been halved for interorops with roots or pulses. Results from the 1950 survey may be high; cuts were taken only from pure stands, and on small plots between 1/500th and 1/100th of an acree /a Maize was unrecorded in the 1950 survey; it has been attributed the same yield as millet. /b Unidentified "cereals" in the 1950 survey have been attributed to Millet. /c A per-acre yield of 3,500 has been assumed for all root crops in both years, - 20 - management surveys in the 1960ts in three areas of Sukumaland. These proportions, with estimates for root crops added, are averaged and compared to estimates for 1950, which have been derived from the results of 1949 and 1950 yield surveys applied to the crop acreages from the 1950 sample census report. Clearly, regardless of the crudeness of these estimates, the balance of the grain supply changed from sorghum and millets to maize and rice. The quantity of grains was reduced, and the production of roots increased, supplementing the supply of bulk foods. 31. No information is available to show the foodstuffs purchased by farmers in 1950, but a household budget survey by the CSB of lake cotton growers shows food purchases for the six months, May 1 - October 31, 1966: Total costs of foods purchased were Shs. 76.70, of which Shs. 51.30 were spend mainly on meat, fish, sugar, and salt, and Shs. 24.80 were spent on maize, cassava, millet, and rice, 1/ Since May to October is the period of plenty following the main harvests, it is reasonable to assume that food purchases at least as high for the remaining six months of the year. This would give an estimated annually expenditure on food over Shs. 150, with approximately Shs. 50 for staple foods. However, the budget survey diows this to be balanced by sales of staples worth Shs. 40, suggesting internal transactions in the community between those with and those without food surpluses in the season concerned. 32. The extra cotton has been a net addition to the farming system; its greater value has created the incentive for the farmer to grow it on an increased scale. The key questions surrounding cotton development and many of the lessons to be learned from it can be found in the circumstances leading to the rise in cotton acreages grown per farm from 1950 onward. III. HE RASONS FOR THE EXPANSION OF COTTON 33. In Chapter II the sources of the expansion in cotton production were discussed and attributed mainly to increased cotton acreages per household. This chapter outlines the reasons why farmers felt sufficient incentives to expand acreages from 1950 onward, The reasons are fairly clear and can be summarized as: (a) marked increases in the price for cotton received by the grower, and, over time, better yields per acre, both contributing to high,r returns to land under cotton and to the effort expended in its cultivation; (b) a spreading out of the population into areas not previously cultivated; (c) marketing developments guaranteeing the farmer easy disposal of his crop. Other factors have played only a 1/ Tanzania, CSB,"Report on the 1966 Household Budget Survey among Cotton Growers in the Lake Region,"Dar es Salaam, 1967. -21- a peripheral role: the gradual improvement in access to consumer goods and the stimulus of a more closely controlled tax collection. 34. Since nationalization of the main merchant houses and with monopolization by the cooperatives of trade in items such as bicycles, cement, and corrugated iron, the supply of goods has been a problem. Officials of the State Trading Corporation reported that the only factor holding down turnover has been the scarcity of supplies for the area. It is possible that this shortage in consumer durables and capital goods has removed the incentive for improving cash income and, as with the increased price for rice, has been an added factor in the recent stagnation of cotton production. To evaluate this factor it would require a detailed investi- gation to trace inventory changes at the various levels of the wholesale and distributive chain. But this stagnation is only a feature of recent years, and there seems little doubt that the three factors listed were the main contributors to expansion. A. Increases in Price and in Growers' Returns 35. Evidence has accumulated during the 1960's showing African peasant farmers to be responsive to price. Cotton production in the Western growing areas is a classic example (See Table 11). Table 11: PRICES AND PRODUCTION OF COTTON IN THE WESTERN GROWING AREAS, 1943-60 Production Farmer Price Production Farmer Price Year (bales) (cts./lb.) Year (bales) (cts./Ib.) 1943 23,500 12 1952 71,000 46 194 18,000 12 1953 44,500 46 1945 35,400 14 1954 90,000 57 1946 34,000 15 1955 110,100 58 1947 32,000 16 1956 119,000 53 1948 45,000 24 1957 152,000 52 1949 38,000 24 1958 150,600 52 1950 39,100 25 1959 183,400 49 1951 1,600 34 1960 162,900 52 The period 1944 to 1954 saw a 450 percent increase in the price per pound of seed cotton, which more than doubled between 1950 and 1954. The same period also saw an improvement in farmers' yield leveis becaise of the intro- duction of better seed varieties, UK It6 and UK 48. Allowing for a 10 percent - 22 - improvement in yield--say from 300 to 330 pounds of seed cotton per acre-- this ten-year period saw a 500 percent increase in returns to cotton growing. A fundamental change in the profitability rarkings of available production opportunities accompanied this radical increase. 36. Evidence on the profitability of alternatives to cotton is very limited up to the early 1960's. Rounce implies a producer price of about 10 cents per pound for sorghum and millet likely to be sold when in surplus. 1/ Collinson gives a producer price for the same crops at 15 cents per pound. 2/ Assuming a figure of 12.5 cents per pound, Table 12 compares the returns from sorghum and millet in 1950 with probabl returns from cotton in 1944, 1950, and 1954; and it demonstrates how the return from cotton overtook returns from the main grain crops over the ten-year period. Table 12: INCREASING RETURNS FROM COTTON AS COMPARED TO SORGHUM AND MILLET, 1944-54 Yield Price Return per Crops (lbs,) (cts./lbs.) acre (Shs.) Sorghum 1950 755 12.5 94038 Millet 195o 655 12.6 81.88 Cotton 1944 280 (Est.) 12.0 33.60 1950 310 25.0 77.50 1954 330 (Est.) 57.0 188.10 Note: Est. Estimated Although the Sukuma farming system is limited by seasonal labor supply at, the time of cultivation, comparison on a return per acre basis is valid because the labor required in preparing the seedbed is similar for all crops except rice. Cotton moved from being a fairly poor crop in the 1940's to a highly profitable one in the early 1950ts, when the price moved up to as high as 57 cents pe- pund of seed cotton. 1/ Rounce, "Economic Development of Sukumaland.11 2/ Collinson, "Field Management Survey 2: Usmao Chiefdoa, Kwimyba District,tv 1962. 23 - 37. There is a lack of information on sales prices of lint before 1950, and on changes in intermediaries' margins once the board took control of marketing the crop. However, the main increases in price took place between 1945 and 1952 while the Tanganyika government continued to purchase the whole crop to sell to the British government--a practice common through- out the world markets during this period--because of the shortage of raw materials caused initially by World War II and later by the Korean War. 38. The rapid increase in the value of cotton to the smallholder was sustained until the early 1960's. After 1961 a falling market pushed the grower price down, although support from the price-assistance fund softened the blow. By this time the developments in marketing and the evolution in the farmirg system had established cotton as the major crop on over 300,000 farms in the western growing area. B. The Spreading Out of the Sukuma Population 39. Population data available show changes by district, thus masking some of the movement: For example population emigrated from the west to the east of Mwanza District and from the north and center to the south of KXimba District. Nevertheless, the movements awe so marked that the major changes are clear from the district figures set out in Table 13. 4O. Before the colonial period began under German administration in the 1890's, the distribution of population was controlled by security considerations. Most chiefdoms were independent, isolated units; people were afraid to settle in the intervening areas due to internecine warfare.1/ For defense needs, people were often more concentrated than they are today. The rule of law imposed by the Germans and continued by the British gradually removed this constraint and the population began to spread out. Movement was accelerated by periodic food shortages caused by soil exhaustion and by infestations of a weed called Striga (Imparata cylindrica). Later, as population grew, people began to move across Smith Sound into Geita District, to the northeast, east, and south. Between 19L10 and 1970 this peripheral expansion continued. h1. The British colonial administration, aware of the failing soil fertility in the older, settled areas of Sukumaland, pursued two lines of development in an effort to redistribute the population more evenly over the area. The first was tsetse clearance and resettlement of the human population, based on the knowledge that population densities of 20 to 40 persons per square kilometer were needed to prevent tsetse regeneration. The second was the establishment of water sources in poorly supplied areas, 1/ D. W. Malcolm, Sukumaland (London: Oxford University Press, 1953). -24 Table 13: ANNUAL RATES OF GROWTH AND POPULATION DENSITIES BY DISTRICT, 1948-67 Rates of Growth Population District (percent) Density per sq. km. 1948-57 1957-67 1948-67 1948 1957 1967 Mwanza -0.1 2.9 1.5 46.1 65.6 60.8 Kwimba 0.2 2.4 1.3 39.2 39.8 50.2 Ukerewe 1.0 1.5 1.3 135.0 147.5 170.6 Shinyanga 2.0 2.3 2.2 22.6 27.0 33.9 Geita 7.6 3.3 5.4 15.4 29.8 40.9 Maswa 2.0 4.0 3.0 11.5 13.7 20.1 Nzega 1.0 3.5 2.5 13.7 14.9 21.7 Musoma 3.6 4.0 3.8 9.4 13.1 19.7 Biharamulo -1.9 7.0 2.6 4.6 3.8 7.4 Source: Tanzania, CSB,Recorded Population Changes 1948-67, 1968. (see Map 2), Between 1930 and 1950 approximately fifty small dams and catchments were built, mainly by the local people organized by the headman.l/ 42. The construction of water sources and the tsetse-eradication campaigns stimulated significant local movements of people (see Table 14); internal pressures within the communities helped the administration. Large general movements of the population to the peripheral districts enabled the Sukuma to maintain the balance between arable farming and grazing. They were able to continue to maintain the fertility of the land by crop rotation as they had done traditionally; and grazing problems were relieved that had been exacerbated in the older areas by the removal of an effective natural check on the cattle population as a result of the highly successful rinderpest-eradication campaign promoted by the government in 1942. These movements also increased the availability of land, allowing the Sukuma to maintain the size of their land holdings and to increase cotton acreage grown per household. Holdirg size would have begun to contract as population increased if internecine war and the tsetse had confined the Sukuma to their traditional chiefdom-based settlement pattern. 43. The data in Table 15 show the relation between the geographical expansion of cotton production and these population movements. Of parti- cular interest are the data showing the relationships over time between population density and production per head and per unit of land area for Ukerewe, Mwanza, Kwimba, Geita, Shinyanga, and Maswa. Despite a rapid 1/ ibid. . . � . . ` • . J � ` � • � L1 • г: `4 � , • .f.� л'I ' iG .�' У Р� Т О fi' 1 А �l''���'�.., . � � ` к� 1 . . �1. J". ._,У� .r �7 � . . •, i�� , � �J�,, � е ' � 1 � � е � . 1 i. .� '9 6о�, ilнi iяи.jв if� wаси wытйстеtrьд �ц '` \� 5 �,� .�г:` r �'� -�зг�..� 1 о о°` 1 г, s�r ,-^ С�,м-�� _ s'`-rc' ^�`-' �-' �= `° р tSmfFд � 1�s р¢mp1s. 1?вю алпь4гиеsад Ь� • 4 Твн�. Gmrmrwмna0 таа таВагllаад. � �� ��, - . � © а `�., hд�` � ��У� /� ��У-� j^�� J � `l r � и+ы wmraг�ra . - а--�'- О ° д�%�`� �-� (ti'�)О ��, . $Н _.- 8ага }юl�а.� • • . . . . . . . `±� j ° О � •, �� 9 ��{ ;-� `''г.. `}: ��� � Г ,i 3 9еервув ег Sp�iag. . . f °�1�' /'� i�о •° Г���j,о �..J�-. t�,Jt _<1?"`�° / � .�'� �) � \ 'r l �` L � `.�С .f 84тд SлЬ•ин'Роаа wo4av, ' �Г `о `.Х + 1 � 1� ~ . `����1-% . (�¢ д?-�� ст � � � -- 1 � �''✓�� / °( �! i - /�q� �{�°�J�Ама4 eithia л�•S ы�1еь е4� рвгавааоа@ wafъr вr�-р11вд тМ1вh та � ° 'н С� ~ i, -,� •1,й; •!, ,l '', О° Q�Q °°� �Г �Jj ) �„J 4. r • � �� ' -�'� р Q Q �% •`�•_, цLJ adeqыata {тг alS pwQmaao. , . � 1�< ss. �` ~ 0 L ��rM� �а r \. О ° i�.�'� � У о ... } -.� ,� , � ���{ �п -�г ;-� � о - � (�"� t ¢ .� и •` �.�' -" 6 �! � i R•ваа иi7йiw а'',� мiian агр ргаhоЫб рагпа+вп4 ые4ьл сиррliви 1а '',,�О а ' �` ,�� Q f д� л� ^`_ О; 1 ^, �� • . ` `.а1 ил�.илььагд tmnd. '� °- � ¢\..lJ � ���: Г '1',, CV � � 1� !"`'t . S, � } �� Q . ;�r �. о г �Г'��( 'r�'' J �r Vl Lj _�.- �� �� � '�.• ° n�j� `_ "�l \� t �! � i О Агатв wьki�in 1 пilв vf раг.птпьп4 va4¢г ь+.рр9¢аа wtй<9� вге _ .. .�� о `.1_ ;Q• 1 ьМl4л�иФ4в }тг длwиь}fspuepaaat олФ аrиьМ ь4есL ow1y. Уkгае , . � �`0 � "" = � /, ��� / Q 1 iкlлда tha. а�лllаr uwi.uc +чт4м со7ь�п.смГ.гтгkт, muko tw ' .. - . . ' �. rl..i• • � \ .(о.' /%' •.-��-�'�r.�( '1' (� , �г�� � , ы.ь1! ааид� вFгаоп Ьгдь, iу.lтла, аваgеqаг ond ьопв , • � 7 -`f.+'ву � р��'л\�Ч-����Г г_у-� �r 1 � � � - _ rл►.вьгloas виv а.реGаь. : ' ' . � ��4...�..../• С} О ?!i� ��� l С�'):J�? (?/j � ¢ ��! ! ~� ' - ' • . . � - • ��� . о . 1 г ,; .,¢ ('\ �r, �, �.r. • - • `��,�� - р °�J ����-( �� с� `- �: + `� �.,1.1; � • . � Агrть withSл � пмlа т: ®н.п�i� цам�.рима�ыР ит4аг wppticn • . . (�,, / ,'� � �¢-- , � J�� Q "iJ • �`,`; /•` �,� р 9`��� Ч( `� ti� д� - , . ,�, вд.q..овь fas 1wмт. ыеа н¢q, - . . . \п / � f� �� 5 '� � 'У �•� � j......_. 1 г _ • -. . WRTER BVPYISE4 ' • � . `,/!•.(i° �° ,, "�Q� ��//' i,. `� \ �f � lпвдв 1д wa4e.aA аггаъ. • . , • • ' • � '� i.��'' + (J��' ` � 1 , �д з ' . � � • Т : �г о �v о °;�� о О� ``,+,г 1.�,. f--,( '1�;��- } iдРйtвУ. (. Ia 41ив дг� вилваw гiаам 9л1д теТог .е\у йв раеlь ег ьавl.ГlВад � � � � � � � � � �, � � г` �� О � - 4а.дь. Тве �..аа p..ittгn .г а..гq рл.г ьа nat �,г 4..�л. � `� �, �л,1� о� \ �, �/,'�. f�_,r\^ ` r�`.�� ��Т,. г ц 0яЬу iб�o lвмьг двма вгг .v мтtlу ь.гд Ьу uR1a. Наf:.в, ta.44, ' . ` - t / г 1� l�L��' � l�•y• l1 -� i� ^`r-� ���•г�� �^ • г j °l � ����/ '�'`J'VV ч ���/^ i� Г� ¢J � 1 у..а 6.lги, а:..1иа wfгl:ea а.е ..емУ. S..¢аа ега Uvг1a и•ад 4у 3птk bL00Q.CЮ0 . /-• . • , ! ,!}-л �\..'� ��J� 5 fa tnM1a тлд Nue мiу рг� т в..аR оевtь моiл\у � sвlиа+и вbвар mwd • , v ���' ' (` • .QuaF0. • • Pi1Iв4 и ' ' • .. �\�.% � ` , -- , ��. • . • . . . � 1: . Ма�2: Water Suppli�s in Sukumaland (ta�ar� �r•®т D. W. Маlсоlт, Sukumaland)в 26 Lable 1. -. POPULATION AND LIVESTOCK VARIATIONS AMONG SELECTED DISTRICTS BETWEEN 1934 AND 1947-48 People - Livestock* Areas 1934 1947 48 1934 1947-48 of emigration Kigoku 12 800 8,,6oo 33.,4oo 25,6oo Ngwagala 71,,6oo 46.,400 Ntuzu 463800 34.,200 --- 5- 00 Ngwdubi --- --- 31"000 2 , Buchunga --- 233600 15.,6oo of immigration Itilima 8,60o 35.,265 232400 87.,800 Sengera-ma 13,4oo 192800 --- --- Kimali MOO 19,307 --- --- Xgunghu --- 45,ooo 68.9000 Nth Sanga 65200 21$800 Source D. W. Malcolm, Sukumaland. Note: Water sources were established in areas of immigration. The time period spans 1942 when the successful rinderpest campaign was mounted. I!bLe ¶s CoTTa PRCU on An PCPUIATI O EPANSION, 1955-1970 1955 1960 1965 197 o Pounds Pounds Pounds ?ounda Population Thousand Bales Ped Population Thousand Dales need Population Thousand Bales seed Population Thousand Bales sood Ar Thousand per Blo par cotton per Thouand per 'bales per cotton per Thousand per .bales per cotton por Thaneand per bales per cotton pr District (eq.km.) people eq. ks. cotton sq.km. head /a people cq. Is. cotton eq.In. boad /a people eq. Iko. cotton &q.kn. heed /a people sq. km ctton 'q.km. hand /8 Older areas 'Xanza 3,890 177.9 66 19.2 -.93 129 187.9 48 25.2 6.46 160 223. 57 37.5 9.61 200 257.8 66 60.7 10.56 189 Kwiba 6,080 21.2 ho 18.6 3.09 94 259.9 43 33.9 5.58 157 291.3 h8 64.1 10.54 264 328.1 54 52.0 8.55 190 Ukerevo 640 92.5 165 5.3 8."6 70 97.3 152 3.3 5.02 39 105.0 164 7.1 11.09 81 114.2 178 h1 6.41 43 Total 1,0,610 511 231 k3.L/a 4.27 102 Sh5.1 203 62.4 5.87 137 619.7 269 108.6 10.24 21o 700.1 298 96.8 9.12 166 Never Areac Shinqanga 9A50 2h!.6 26 6.2 0.67 31 273.6 29 19.7 2.01 0. 36.0 32 38.7 2.10 101 343.7 36 55.3 5.85 193 Geit.A 9,070 233.4 26 2h.9 2.7" 129 297.8 33 32.4 3.57 130 33. 8 6. .5 3 o. 56.3 6.85 12 ava 21,370 280.9 13 10.A 0.50 45 328.8 I5 21L.1 1.13 88 398.1 19 66.3 3.10 200 h8h.3 23 1033 .3 26 N 7 229.4 N.A. - - - 276.7 NA. 11.2 0.82 h8 32n.2 IA. 32.4 2.36 121 Nzsam. 17,70 201.2 N.A. 2-0 16 h2 25$ NA 21.8 1.22 102 310-5 IF.A. .36.4 2.04 .141- 377*7 N.A. 32.3 1.84 10L iazvalo 10,900 h2.5 N.A. - - - So.1 H.A. - - - 70.2 N.A. 392* otl 82,450 1,215.1 N.A. 66.5e .8 66 1652 NA 8o .9 8 ppa MA 221.1 V1.68 155 2,o28.3 N.A. .298*3 3.62 176 Total 0.8/c 12 102 1,5.1 X.,3 92.0 .9:2 17.8 NA Oreator~ 9,.72 21'6 2 . .7 31 236 2 97 20 976 23*4 32A 329.7- 1,1 3.56 2,728.7 N.A 355.-1 8 9 Osuita lad 93,060 1,726.7 N.A. i1.6 1.18 76 1,980.3 .A. 160.4 3.7 93 329.5 35 68.5 7.55 27 1,72.1 1:5 62.1 6.85 182 sm,ceens production figurso--derived ram gicmcry-gon data, only available after 1953o by allocanting production to a district accordina to the rurber of senal Isaina pos%4 In that '1istrictl Papalation figamos-Obtaised frea the rates give in Table 13. /a Bmaoga is mmltiplied by 1.200 to givm po=mda at ewd wotton. /b Tbipansion into 03aita va mainlr Prior to 19%6. fgSansdiao rmpanats with adal total balcm cold arise duet to problema in allmocation to distrInft. po - 28 - increase in population, production per head in Miusoma did not rise over the given period. Nzega and Biharamulo are relatively recent cotton- producing areas, making any conclusions from the scant data available from these districts questionable. Although baleages in different areas are subject to localized climactic conditions from year to year, there are some interesting indications at the effects of increasing population densities. 44. The larger part of Ukerewe District is Ukerewe Island populated by the Wakerewe, a tribe distinct from the Sukuma, with a tendency to remain as a community, not migrating as freely as the Sukuma. The effects of their insularity and the resultant high population density ar clear: Although the intensity of cotton production per land unit is high, the quantity produced per head never achieves the level realized on the main- land. By 1955 the population density was already too high to allow the size of holding found on the mainland. (In Table 15 the 1965 figures of 81 pounds of seed cotton per head and 11.09 bales of cotton per square kilometer were the results of an enforced fertilizer program between 1963 and 1965 that broke down after rioting in the area.) The renewed upturn in rates of population growth in Mwanza and Kwimba districts after 1957 suggests that migration has slowed, and these areas will follow the pattern of Ukerewe over the next decades. The situation of Ukara Island, west of Ukerewe, perhaps points to the eventual predicament of an increasingly larger area of Sukumaland as population pressures grow and new lands become populated. H. Ludwig has compared agricultural conditions on Ukara Island, where, as a result of migration, population is now static at a density of 207 persons per square kilometer, with those on Ukerewe Island and Kwimba, documented by D. Rotenhan. Data from Ludwig's comparisons are reproduced in Table 16; results for Shinyanga, found in Rotenhan, are added. The cash crops on--Ukerewe and the mainland is predominantly cotton, but none is grown on Ukara where rice is sold. Although.the data are based on fairly small samples, the differences are large enough to demonstrate a phenomenon likely to spread to the mainland as population densities increase. 45. The data suggest that the maximum possible yield for the existing farming system and present cotton technology is ten to eleven bales per square kilometer. However, the southern parts of Geita, eastern parts of Maswa, and southeastern part of Shinyanga are probably unsuitable for cotton growing, and the area data would need refining to allow any estimates of potential production from the area. Experience in Ukerewe District offers evidence of trends of per capita production at high densities of population, but because rainfall is less favorable on the mainland, the 11carrying capacity" is probably lower and the downturn likely to occur at lower popu- lation densities. C. The Devlopment of a Cotton-Marketing Infrastructure 46. Undoubtedly the turning point in the development of cotton marketing in Tanzania was the settirg up of the Lint and Seed Marketing Board OSMB) in 1952. Because the western growing areas did not emerge as 29 - Table 16: A CONPARISON OF LAND USE WITH DIFFERENT POPULATION DENSITIES IN PAF TS OF GREATER SUKUMALAND, 1965 Ukara Ukerewe Kwimba Shinyanga Items Island Island District District People per sq. km. 207 164 /a 48 /a 32 People per holding 10.9 9.0 6.6 5.9 Livestock units 3.4 2.1 10.7 19.0 Total farm area (acres) 6.2 12.1 19.4 23.4 Fallow (acres) /b 2.0 6.0 13.7 15.3 Cash crops (acr`) 0.7 2.2 2.5 /c 3.7 Subsistence crops (acres) 3.5 3.9 3.2 7 .4 Per capita net value of production (Shs.) 117 99 210 378 Sources: H. Ludwig, "Agricultural Practice on Ukara Islandly in H. Ruthenburg, ed., Smallholder Agriculture and Development in Tanzania, Vol. 2 (MunichFOT 96F; D Rotenhan, "Cotton in sukumaland." /a Population densities for 1965 in Ukerewe and Kwimba have been inter- polated from Ludwig's data and Tanzania, Recorded Polation Changes, /b Grazing areas have been included as fallow land, as it would be mis- leading to conclude that animals only graze on the holding, at least on the mainland. /c Cultivated acres in Kwimba seem low, Collinson, in Usmao Chiefdom, Kwimba District, reported 2.9 acres of cotton and 3.7-ares subsIstence crops. the major producing areas until after World War II, and because from 192 to 1952 the government compulsorily purchased all cotton and sold it to the British Ministry of Supply, the history of the LSMB mirrors the history of Tanzania's marketing infrastructure. Nevertheless, one or two landmarks did occur during the 1930's. 47. During the 1920's, in fact, the government made an initial attempt to fix a guaranteed price for the producer. Commentators present it as a failure for two reasons: (a) Private processors were prepared to pay higher but varying prices, so the steadying effect of a guarantee never reached the farmer; and (b) seed distribution was contracted out by commercial firms who required growers taking seed to sell through them, often at a previously agreed price. 1/ 1/ Kirby, "Cotton in Tanzania," Empire Cotton Growing Review, Vol. 7, No. 3 (1930). - 30 - 48. Government initiatives in the early 1930's were the start of a prolonged dialogue with processors. Ginnery building was limited until justified by the level of potential throughput; in return for this security in investment for existing ginners, the government exercised influence on the prices paid to farmers. In addition, the cess on cotton was reduced, and ginnery-controlled buying posts were abolished; ginners had to compete for the crop. Two ginnery inspectors were appointed in 1934. Significantly, the cotton crop in the western areas rose from an average of between 7,000 and 8,000 bales over the 1920's to between 30,000 and 35,000 bales by 1933-34, where it remained until 1950. 49. Development was inhibited by the events leading to the war in 1939 and by the war itself. In 1941 the government, in an effort to main- tain production, allowed ginners indemnities on high interest and on insurance and storage charges caused by shortages of shipping. In 1942 the government took over the purchase of the crop for sale to the British Ministry of Supply and later appointed the board already established in Uganda as sales agent. The Lake Province Exporters Association in the Mwanza area acted on behalf of local processors. The government continued to purchase the crop until the LSMB was set up in 1952. As forerunners to the board, provincial cotton committees had been established--in Lake Province in 1949 and Eastern Province in 1950--to promote cotton production. 50. The Lint and Seed Marketing Ordinance of 1952 established the LSMB to put cotton marketing under central control. 1/ This review is limited to a discussion of the policies and programs adopted-by the board that might have played a part in stimulating growers to increase cotton production. Other activities of the board that have supported such policies and programs are outlined. 51. First, the present marketing structure is briefly described. Map 3 shows the ginnery zones with the ginneries and oil mills around which the marketing of cotton pivots. Primary buying is done solely by cooperatives, and each of over 590 cooperative primary societies is assigned to a ginnery zone; its purchases are channeled to the ginnery concerned. Only two of the 22 ginneries remain in private hands; the cooperatives own 20 and have a shareholding in the remaining two. The LSMB buys the crop from the ginneries and transports it, via the railheads shown on Map 3 to the seaports, mainly Dar es Salaam, but with 15 to 20 percent of the crop exported through Mombasa via Kisumu on the lake. 1/ Much information is available on the structure, costs, and performance of the LSMB: 0.K. Laurent, in "Cotton Marketing in Tanzanial (mimeo., 1969), not only covers the board, but also presents a detailed analysis of the costs and performance of the cooperative primary societies. H. G. Porter, in"Cotton in Tanzania" (Washington, D. C., USDA, FAS M219, June 1970), summariies all marketing aspects, including board activities. - 31 - WESTERN COTTON AREAS TANZANIA GINNERY ZONES Lake Victoria ugaa 4:77 bUshash' 00 Uker2v ~~~roues toKiilara Y*,assa Buchose.Nga Ct aima.gu Kasamwa<D---- uya Manawe-' ÅLuguru _Sola Malampaka O Ginneries- -roues torailead/Olsogore 'OMhunze route to rilhea Shinvangao and ports DManonga Map 3: Ginnery Zones, Ginneries and Oil Mills, and Routes to Railheads and Ports, Western Growing Areas, Tanzania - 32 - 52. The buying season begins in June and ends usually in November with most of the crop ginned by the end of the year. Farmers bring their seed cotton to the primary society where it is graded A or 0; Grade C is rain- or insect-damaged crop, about half the value of' Grade A,- and makes up 8 to 12 percent of the total crop. The society pays cash on delivery (crop finance is provided with the help of the National Bank of Commerce, the NBC); the National Cooperative Bank has recently been absorbed by the NBC. The cooperative union organizes the transport of the seed cotton from society to ginnery. The cottonseed is recommended either for milling or for redistribution. That for milling is sold to the oil millers (out of 12 millers, 10 are owned and operated by the cooperatives). The oil is packed in four-gallon tins and is sold within Tanzania for cooking; The cake is exported for cattle food;and the cottonseed for redistribution is bought by the LSMB, dusted against insect pests, sent out to all societies, and distributed free for the following season's plantings. It should be available to societies by the end of October in time for the new plantings with the rains in November. 53. From its initiation, the LSMB took a positive approach toward the promotion of the cotton industry, partly undoubtedly because growers' representatives had a place on the board from the outset, and partly because the LSMB's antecedents, the provincial advisory committees who were finally absorbed by the board in 1959, held a brief for cotton production promotion. Three facets of its activities directly stimulated the growth of production: the promotion of primary marketing activity, pricing policies, and the organized distribution of free, improved seed D. The Financing of Assets for the Marketing Cooperatives 54. Perhaps the most positive initiative of the LSMB, backed by the government, was to finance the growth of the cooperative movement as a primary marketing channel for the hundreds of thousands of small cotton growers. Cooperatives had been developing in the Lake Province since the 1930's but it was not until 1953 that these societies were allowed to buy seed cotton. This was the first breach in a field dominated by the processors, a breach which widened until, by the late 1960's, the coopera- tives controlled not only all the primary marketing but also the ginning and a significant proportion of to seed-crushing industry. Table 17 documents the domination the cooperatives so rapidly achieved in cotton marketing after 1953. 55. Prior to 1967 the cooperative hierqqrchy was a three-tier structure with the primary societies buying seed cotton from the farmers, the unions supervising the societies and transporting the cotton from them to the ginneries, and the Federation of Unions coordinating the unions. In 1967, the unions were reorganized on a zonal basis, one to each ginnery, to become area offices of the federation. The marketing zones, centered around the - 33 - ginneries, and the primary societies paralleled, both numerically and geographically, the expansion of the crop; societies were only registered when production locally was judged sufficient to support the cost of operation. Map 4 illustrates the growth of the primary marketing outlets, the cooperative societies, on a district base. Clearly, this rapid increase in buying points located according to local production gave easy access to market for thousards of small growers. Table 17: THE INCREASING ROLE OF COOPERATIVES Nuimber of Percent of Crops Number of registered purchased by Year societies members societies 1953-54 38 15,334 13 1954-55 65 33,935 32 1955-56 113 53,282 45 1956-57 198 86,627 61 1957-58 235 92,400 70 1958-59 275 N.A. 83 1960-61 335 N.A 100 Source: Cooper Brothers,"Report on the Cotton Industry in Tanzania," Dar es Salaam, 1959; supplemented by data for 1960/61, when the cooperatives were given monopoly rights for purchasing. Note: N.A. = not available. 56. The core of the cooperative movement in the cotton,areas is the primary society. The superstructure above this level is akin to that of any large-scale organization. At the higher levels a democratic element persists in the annual general meeting and in the final authority held by the committee of the union; nevertheless, real authority lies with the executive. The procedures and control required to run a large organization are not always conducive to the democratic solution of local problems. This is not true at the society level. Society membership varies between 300 and 500 growers. Each village within the society area nominates a committeeman who, together with the secretary, the only paid official and the one responsible for the day-to-day running of the society, can be taken to task and refused re-election at the annual general meeting. The chairman, vice chairman, and committeemen hoild respected positions locally, but are very much responsible to the membership in the tradition of decision- making in African rural communities. During the busy marketing season the committeemen, who receive subsistence allowances and honoraria approved at the general meeting, will do much of the buying and handling of the purchased crop. - 3h - 344 W,STEN; CMiUj ARAS TA 1LI\ ADMINISTRATIVE DISTRICTS .. kor Victoria o 1 2 -. 2 2 3 2 2 2 1 23Musoma Distric 3 40 12 3 4 Z ereWe ' ct 4 2 -anZa Distri 4 22 2 2 3 2 2 1 0 2 2 2 4 42 2 2 .Glta 2 0 1 2 2 2 2 , 3 2 2 -3 3 MU.N nulo Geita 2 2I-,la swa 4 Vi rrict, - District 2 3District 3t42a/ KWLpbap 2 2 3 . 2 4 District 0 ,itrc 2 ,2,( 4 3 3 -r 2 - 1 2 6 ) er. 1 2 2 fShinyanga 2 '. Railway 2 2hi 4 3 a - 2 Shinyanga District boundaries ..-.n.c.o. Distcict Boundary; Sukumaland proper- 4 3 Nzega District 4ap A Sch~mtre.,Representation of the Inårease in Cooperative Primary Societies in the Western Growing Areas of Tanzania, 1953-70 0 Societies registered by 1953 ) 1 fr" 1953-55 )Bach symbol 2 " 1955-60 )represents 3 "19 1960-65 )3 societies. 4 11a 1965-70 ) Note: Societies are not correetly located, 57. A high proportion of societies nro tied into the credit program for fertilizers and insecticides. Diotribution of fertilizers is done in November, December, and January, and of insecticides in February-March; recoveries parallel the marketing of the crop. 58. From its inception the LSMB accepted the responsibility for financing the cooperatives. It paid the levy for their operation from its general funds and supported capital works at each level of the hierarchy with loans from the Price Assistance Fund. Initially, on the registration of societies with the registrar of cooperatives, the LSMB made a loan, usually of between Shs. 18,000 and 20,000 for the construction of a store for seed cotton. It also promoted loans for cottonseed stores to keep itself clear of the seed cotton and to give better control of seed distri- bution. Later it made loans to the Federation of Unions to build ginneries, which helped establish it in the pocessing as well as in the marketing aspects of the crop. By the end of 1960 the board had financed storage for 15 unions and 363 primary societies and continues its policy today. During the 1950's the board deducted from the remuneration to the societies interest (usually at 4 percent per annum) and principal due. Early borrowers thus paid off their debts quickly. Latterly, as more of the marketing is con- trolled by the cooperatives, this direct recovery has lapsed. Interest and principal payments are currently badly overdue; there is a reluctance to make transfer payments within the industry when the original investments were made with funds withheld from the growers. The Price Assistance Fund, into which repayments are channeled, continues to be useful in balancing this reluctance. 59. In financing the fixed assets,.of the cooperatives the board showed a flexibility in its policy not always apparent in creditors even 20 years later. For example, it did not insist on a title being obtained for the land on which the store was to be erected; agreement between the society and the local authority was sufficient. This remained the case even when ownership of the crop before processing, the boardts security in these early years, passed to the cooperatives in the 1960's. With hindsight after 20 years, this financial support for the cooperatives seems straight- forward enough, but in the early 1950ts with no thought of an independent, government, it represented a degree of flexibility uncommon among colonial institutions. E. Control of Marketing Margins and Stabilization of Grower Prices 60. As we have seen, the tradition of official pressure on ginners to keep reasonable margins began during the 1930's. In 1952 a government committee examined the services performed by ginners, up to then respon- sible for buying the seed cotton, transporting it to the ginnery,and ginning and dispatching it to the railhead. But the average throughput in 1952 of 4,000 bales of cotton had increased to 7,500 bales by 1958. So, in view of the increased crop and the growth of cooperative societies, which - 36 replaced ginnery buying posts, the LSMB commissioned a further investigation in 1957. The results of the two investigations are given in Table 16. Table 18: PROCESSORS COSTS 1952 AND 1958 (cts. per lb. lint) costs 1952 1958 Buying expenses 8.1086 7.1922 Transport of seed cotton 3.2042 3.202 Ginning Overhead 9.0089 8.1394 Fixed costs hb50 3.3163 Variable costs 9.8255 7.2402 Transport of lint to railhead 0.5625 0.6696 Total 35-154 29.7619 Source: C. K. Laurent, "Cotton Marketing in Tanzania," mimeo, 1969. They probably formed the basis for decisions on remuneration as the processors' responsibilities were gradually divided between tP cooperative movement, the ginners, and the board itself under the close supervision of the board and the government. In 1961 in an amendment to the ordinance that had established the board, the government decreed that the institutions would agree on the costs of buying, ginning, and transporting and that the board would pay a fixed lint price to the ginners. This was the basis for a tight control over marketing margins that remains to date. Table 19 shows how margins have moved since statutory control was established. Over the period there was a 45 percent reduction in the total levies to the marketing hierarchy in an attempt to bolster a falling grower price. 61. A second important facet of board policy vis-a-vis price control is stabilization. The ordinance of 1952 provided for a price assistance fund to smooth out short-term price movements on the world market. The ordinance was given teeth by the section that provided Shs. 16 million be paid into the fund from profits accrued to the government during the period it had control of lint sabs. The fund built up very rapidly during the 1950's to a net worth of over Shs. 100 million by 1958, but with the downturn in world lint prices it fell to Shs. 38 million by 1967. At the same time, with the rapid increase in the crop, by 1967 a one-cent change in the grower price cost the board Shs- 5 million., So, whereas the 1958 Table 19: PRICES PAID TO FARMERS, LEVIES, AND GINNING FEES, 1961-70 (Cts. per lb. seed cotton) Coop ederation Export Grower price coop union of Federation Coop or Local Ginning Finance Year Grade Grade union marketing Unions' development society Gov't. council fees (per and Total A B levy/a fee levy /a levy /b levy /c levy levy lblint insurance levy 1960-61 55.0 23.0 1.0 2.0 1.0 --- --- 2.0 31.0 1.0 h3,0 1961-62 55.0 23.0 1.0 2.0 1.0 -- 5.0 --- 2.0 31.0 1.0 43.0 1962-63 55.0 23.0 1.0 2,0 1.0 --- 5.0 --- 2.0 26.5 1.0 38.5 1963-64 50.0 23.0 1.0 2.5 0.5 o.6 5.0 --- 1.0 23.0 1.0 3.6 1964-65 50.0 23.0 1.0 2.5 o.5 o.6 5.0 --- 1.0 23.0 1.0 3.6 1965-66 48.0 21.0 1.0 2.0 0.5 o.6 5.0 2.0 1.0 23.0 1.0 36.1 1966-67 46.0 21,0 0.8 1.8 0.4 0.3 3.0 1.0 1.0 18.0 0.5 26.8 1967-68 46.o 210 0.8 1,8 0.4 0.3 3.0 3.0 1.0 17.0 0.5 27.8 1968-69 48.0 23,0 1.0 1.8 --- 0.3 3.0 3.0 1.0 17.0 0.5 27.6 1969-70 50.0 25.0 1.0 1.8 --- 0.3 3.0 --- --- 17.0 0.5 23.6 Source: Laurent, trCotton Marketing in Tanzania.? /a These levies were paid to the zonal unions until the end of 1967 and to the Nyanza Cooperative Union, previously known as the Victoria Federation of Cooperative Unions, thereafter. /b This levy was mainly used for the purchase and construction of ginneries; with the reduction to 0.3 ets. per lb., a levy of 3.0 ots. per lb. on the lower-grade BR.was paid. /a Although this levy was reduced to 3.0 cts. per lb. in 1966-67, remuneration was on a sliding scale based on throughput, acknowledging the high overheads of new societies; average payment in 1966-67 was about 3.6 cts. per lb. /d Similarly, ginning fees are now on a sliding scale; 17.0 cts. per lb. refers to ginneries processing over 10,000 bales. 38 - fund was valued at 50 cents per pound of seed cotton, the 1967 fund was valued at 8 cents per pound of seed cotton. 1/ It is evident that the fund had been used for support rather than for atbilization. Since Laurent's report, several significant trading surpluses have been realized, and the fund in 1971 reached Shs. 100 million again, nevertheless representing a much narrower base for support than the same amount in 1958. However, with the general downward trend of prices the fund has also served as a stabili- zation device; as an extreme case, in one year the fund contributed eleven cents per pound of seed cotton, over 20 percent of the producer price. 62. A third facet of price control directly contributing to grower confidence is the board's policy, backed by the government, to announce the grower price before the season. This announcement of the price prior to selling has evolved from the government's association with cotton marketing: During and after the world war, the price had been announced by the government at fortnightly intervals based on the level of Liverpool futures. The importance of an early announcement is the security it offers the grower, who plants his crop in December and doesn't market it until the following June. As the crop has expanded, the board has become more worried by this policy; errors in the forecasts of both lint and seed prices have become more and more critical in terms of the board's trading position at the end of the season. But the practice still continues, and the board consistently announces the same price for the season to all producers throughout the western growing areas. Fourthly, levels of remuneration to intermediaries up to 1966-67 were based on average costs inflated by pressures from inefficient and recently established units, allowing large margins for efficient operators. After 1966-67 the squeeze on lint-export realizations brought about a change in policy: A sliding scale provision was made for those societies and processors with a low throughput and high unit costs. The early policy and the change precipi- tated by falling export prices had some interesting repercussions: (1) In the 1950's and early 1960's small differentials in marketing costs found their way back to the society member as a second payment. The possibility of a second payment to society members as opposed to producers tended to create pressures for a closed shop in the primary societies, particularly after the cooperatives were given a monopoly in buying in 1960-61. All of the crop had to be sold through the cooperative: the fewer the members, the larger the second payment. The lower remuneration after 1966-67 largely removed the second payment and with it much of the incentive for membership. Members had nonmembers alike were obliged to s'ell to the society, and members had the dubious advantage of paying a fee for registration. (2) The standard price, regardless of cost of access to markets, must have been a great help in maintaining the incentive to move to, and grow cotton in, the remoter areas. 1/ Laurent, "Cotton Marketing in Tanzania.1 - 39 - (3) The crop from the western areas has consistently subsidized the crop grown in other areas in Tanzania in which the processing and makketing infrastructure has developed in a more haphazard manner. (4) Early in the 1960's the cooperatives put pressure on processors' margins to reduce what were considered excess profits. Later in the 1960's the cooperatives, having taken over the processing of the crop, found themselves faced with the tight margins they had helped to create. 63. Lastly, despite the mounting task of financing as the crop expanded, the board stuck to its policy of providing the grower with the promulgated price on delivering his crop to the buying point. Initially the board used its general funds. In 195 it began to sell its seed from deliveries already processed earlier in the season, bringing in revenue to help finance the purchases toward the end of the buying season. It sought to move the crop quicker to the ports allowing earlier receipts from buyers. The board had always kept a proportion of its PAF portfolio in fairly liquid investments9and for the first time in 1959, and again in 1960, it negotiated the right from the government to use specified amounts for crop finance. In 1961 when control of the crop in primary marketing passed to the coopera- tives, the burden of finance shifted to them. The cooperatives used the crop as collateral for finance from the commercial banks. Nevertheless, the crop grew so rapidly and the problems of moving it to the ports became so serious in the late 1960's that the board again found difficulty in financing and was forced to turn to the banks for resources. It concen- trated on limiting the period of its overdrafts; for example, it negotiated with the National Agricultural Products Board for the mutual use of surplus funds. Their buying activities were, to a degree, complementary in timing. F. Distribution of Seed 64. As earlier noted, the institutionalized distribution of seed has led to the controlled spread of new, improved varieties throughout the growing areas. The board,on its formation, took over the responsibility for seed distribution from the provincial cotton committees. Actual dis- tribution is done by the Ministry of Agriculture, but the board handles the financing of both seed improvement and distribution. 65. Since its inception, the board has vigorously supported variety research carried out by the Cotton Research Corporation at Ukiriguru. It has carried the costs of a seed multiplication program, beginning by supply- ing a 50-acre tract at Ukiriguru, then a primary society area, then several society areas, a ginnery zone, and finally all the zones for which that issue was intended. Its program of support for road development and upkeep has facilitated distribution of seed, and its financing of special seed stores for the primary societies has aided control of distribution. In addition, the board has financed the installation.of seed-dusting dquipment at ginneries, from which seed is used for planting. It also pays for the bags and the copper-based fungicide used to protect the seed against the early Blackarm stage of bacterial blight. 66. The amount of seed distributed has risen from 10,000 to 30,000 tons over the period 1958 to 1970, representing a loss in potential revenue escalating from Shs. 1.8 million to Shs. 12 million. In addition, the costs of distribution, including dusting, storage, and transport have increased from Shs. 0.8 million to Shs. 4.4 million over the same period: Besides, the board has worried about the wastage of seed by farmers: The lint produced per ton of seed issued fluctuates around 1 bales compared to a theoretical 33 bales. The costs of the seed and its distribution represented an average of 27.1.percent of the board's total costs over the three-year period, 1964/65-1966-67. 1/ Despite this, the farmers free access to seed and the control the board-has established over quality and type, if not quantity, must be a significant factor in the expansion of the crop. G. Board Sales Policy 67. The board pursued an outward-looking sales policy: On the one hand, it made great efforts to build a reputation among buyers as a reliable supplier of a standardized product; on the other hand, the management showed a properly commercial awareness in the timing and strategy of selling after it took full control of the crop disposal in 1959. Both aspects of sales policy were instrumental in maintaining revenue as high as possible in the face of a falling world market. These policies, together with the board's control over the marketing margins and the PAF's help, allowed the board to keep the produce price attractive, particularly over the period of fastest growth from 1954 to 1964. Indeed, although the PAF has been used for price support as well as for stabilization, as long as the large surpluses realized from the 1967, 1968, 1970, and 1971 marketing seasons find their way back into the fund, the policy may prove to be justified. The line between stabilization and support is difficult to draw. 68. Some of the main measures adopted by the board to build up its reputation as a supplier are itemized below in two groups, one including measures promoting quality, the other reliability. Board Policies Aimed at Establishing a Reputation for Quality 69. 1957: The board altered its policy from one of acceptance of the cvop from the gins, as it fell, to one of negotiation with the ginners on the quality of the processed product. 1/ Ibid. 70. 1958: (a) incentive prices were offered to growers for early delivery. I is was effective in reducing the time the crop was allowed to remain on the plant, making it less susceptible to rain dulling, dust damage, and insect staining. (b) A quality bonus was paid to ginners for segregating rain-dulled cotton. Societies had been purchasing this as Grade A, and it was being downgraded by classifiers. (c) The board employed a,seed-cotton classifier to travel among the societies and warn the ginners which ones were not grading adequately. (d) The board intro- duced and financed the production of a special type of bag to reduce the amount of extraneous matter being included in the bales. 71, 1959: To any primary societies that would agree to separate rain- dulled cotton, the board offered loans for building ancillary stores. Board Action Aimed at Ensuring Timely Disposal of the Crop 72, 1955: (a) The board began to sell seed early in the season to break up storage bottlenecks at the ginneries. Early disposal of seed cleared stores for seed cotton arriving late in the season. (b) The board continued the policy of the provincial cotton committees by making signi- ficant grants for the development and maintenance of feeder roads. This has continued to date. 73. 1958: The board began a policy of investment in storage and dispatch f`iTities that continues to date. Focal points for these efforts were at the rail- and portheads. Provision of these facilites--based eventually on a policy of available storage for 50 percent of the crop-- gave flexibility in preventing a backing up of the crop to a point where processing, or even buying, would have to be suspended, with consequent delays in completing orders received. T4. The success of the board in establishing a reputation as a reliable seller, and in building up a place for what have come to be known as "Mwanzal in the trade, is reflected in the consistency of its major customers: Hongkong, Japan, Germany, and latterly, China. This success in marketing underpinned the board's local programs designed to build up the size of the crop by servicing the farmer with good seed and by facili- tating his disposal of the crop. The board inherited many of its programs from a policy of government interest in cotton production and marketing that had started some thirty years before the board's formation. Never- theless, with the authority received under the 1952 ordinance, the board could be more positive in its approach, not least due to the competent management it recruited, which used its powers effectively. During the 1960's the board fought something of a battle with the cooperatives for control of the crop to keep up standards it felt were important in the trade. This had an educntional value by bringing the top management in the cooperatives into contact with epatriates on the board who were professionals in cotton marketing. 42 - H. Summary of Reasons for the Expansion in Cctton Production 75. It can be concluded that the expansion in cotton production arose mainly from an improvement in returns from cotton growing when compared to returns from alternatives. The major reason for this improve- ment was a dramatic rise in producer prices supplemented by a less dramatic but nevertheless significant improvement in yield levels, mainly due to better varieties. The boost in incentives for cotton growing was paralleled by the establishment of the LSMB with statutory powers to control the dis- posal of the crop, initially at least from farmer to buyer. The authority available to the board was well applied both in sales policy and in the development of new channels for primary buying. The fact that the criterion for the opening of new buying posts was an estimate that the throughput of cotton from local farmers would be adequate is evidence that the development of the marketing system was a consequence of the increase in production. Nevertheless, it is clear that development would have been retarded without the opening of buying posts in the new areas of settlement. Similarly, without the significant migrations that occurred to the west, east, and south, population pressures would have required an agricultural system devoted to food production with a decreasing arable area per household. The migrations allowed the evolution of a system in which cotton gradually came to occupy 50 percent of the cultivated area. 76. The positive stimulus to the expansion of production was the boost in producer prices. The building-up of a marketing structure and the spread of the Sukuma population were factors that allowed the expansion to occur; without them the rate of growth in production would have been signi- ficantly inhibited. IV. GUENETIIITVSTO PROMOTg COTTON PRODUCTION IN THE WESTERN GROWING AREAS A. Introduction 77. In the preceding chapters we have touched on the nature and effects of three government initiatives that have influenced cotton production in the western growing areas: tsetse clearance and the estab- lishment of water-supply points, a research program on cotton, and the building-up of a marketing infrastructure. Two more remain. These first three were indirect attempts to increase cotton production; the remaining two were direct, one through the improvement of small farmers' management the other by developing the structure of agriculture on a village basis with a degree of collectivization. In terms of the rapid expansion that has occurred, neither has contributed significantly to production, and both were failures, though for different reasons. Table 20: AN EMAMPLE OF THE SUKUKA FAR4ING SYSTER: CROPS, PLANTING TIMES, ACREAGES, FOOD SUPPLY, AND CASH RETURNS Early Late Early maize, maize, Late maize, cassava, cassava, maize, legumes, legumes, Early legumes, legumes, Late sweet &weet sweet sweet sweet sweet Old Potatoes Potatoes Potatoes Potatoes Rice Potatoes Potatoes Cassava Cotton Total Planting time Nov. 1 Nov. 20 Dec. 15 Jan. 15 Feb. 10 Feb. 15 Mar. 10 Dec. 15 Acreage planted .558 1.46 0.05 1.46 0.30 0.58 - 0.05 (1.00) 403 8.51 Output Drymaize 116 511 -11 17- -- 1,300 Green maize (Mar.) 175 --- -- --- --- 175 Green maize (Apr.) 29 1h6 --- --- --- --- 175 Green maize (May) --- --- 146 -- 2917 Fresh cassava (Bept.-Nov.) -- 760 --- --- --- 70 Fresh cassava (Dec.-Feb.) --- --- - 760 --5 --- --- --- 70 Fresh sweet potatoes (Feb.-Mar.) 93 117 ---- --------- --- 200 Fresh sweet potatoes (Apr.-May) --- 117 200 --- --- --- --- --- -- 300 Fresh sweet potatoes (June-July) ------ --- 219 --- 81 --- --- --- 300 Fresh sweet potatoes (Aug.-Sept.) --- --- --- --- 299 --- --- 200 Legumes 29 73 --- 73 --- 29 ------ --- 200 Rice --- --- --- --- 300 --- --- --- 300 Cassava reserve --- 1,577 - 993 -- -- --- 1,h31 --- 4,000 Cash --- --- --- --- --- --- 912 --- Table 21: AN EXAMPLE OF THE SUKUKA FARMING SYST4: LABOR USE (KANDAYS) Labor used, by crop Early Late Early maize, maize, Late Total labor maize, cassava, cassava, maize, available legumes, legumes, Early legumes, legumes, Late Total sweet sweet sweet sweet sweet sweet Old labor Month Family Hired Potatoes Potatoes Potatoes Potatoes Rice Potatoes Potatoes Cassava Cotton used October 20 --- --- --- --- -- 7--- November 5 2 7 37 --- --- 1 --- --- 12 57 December 55 19 6 7 2 7 0 --- --- 52 74 January 55 20 --- 7 --- 29 3 --- -- 36 75 February 60 2 --- --- --- 15 13 14 --- --- 20 62 March 60 --- --- --- 5 6 2 8 12 33 April 60o- - --- --- -- --- --- --- --- May 60 --- 2 6 --- - -- --- --- 32 4h June 60 --- --- 6 4 --- --- -4- 00 July 60 --- - -- --- 3 2 -- --- 28 33 August 60 --- 9 --- --- --- --- --- 12 21 September 60 --- --- --- - 6 --- --- ------ 6 Family total --- --- 22 66 2 63 38 22 2 15 201 47 Hired total. --- 43 --- --- - --- --- --- --- 3 --- Capital 150 -- ---------- --- --- --- 150 150 78. Tables 20 and 21 present a model of the Sukuma system of farming in the mid-1 960 Is. They will help in evaluoting the improvement and transformation strategies attempted by the government. Data were collected in Geita from 1963-65, and the figures in the tables are averages of obser- vations. The farm represented is slightly larger than is typical of Suku:maland; however, the bases of the system are those of farming thoughout the western cotton-growing area. This model farm supplies most of the household foods, mainly from intercrops, and produces cotton for cash. Some resources are hired to help out at the peak labor period. 79. Table 21 shows family labor supply over the planting period, which limits the scale of the system. A unimodal rainfall pattern demands that most crops be planted by the end of February. For this particular area, rainfall is over 35 inches per annum. In the southern and eastern areas the long-term average falls below 30 inches and there is greater year-to- year variation. The peak in the season in these areas is even more pro- nounced, which may account for the increased use of oxen and the hiring of tractors there, with high returns to rapid crop establishment. B. Government Efforts to Improve Cotton Yields per Acre as a Means of Increased Prodd-ction 80. Higher yields per acre were the core of the colonial government's direct efforts to increase cotton production. Agricultural extension was an important section of the administration with a hierarchy of staff under the agricultural officer (AO), a graduate responsible for agriculture at the district level. Under the AO were two or three lower levels; at the bottom were contact staff, directly in touch with the individual farmers and one or zwo intermediate supvervisory levels. This hierarchy persisted after independence and is found today, although the level of qualification of the executive staff, particularly at the district level, is lower due to a shortage of local agricultural graduates. There was a reaction against extension after independence, partly because it was considered not to have achieved a great deal and partly because its plodding nature had little political appeal. The government consequently drained the extension hierarchy of its middle-rank staff, which is one reason for low morale and a lack of planning and direction. 1i 810 The content of extension in the western cotton areas was derived from the experimental work carried out by the Western Research Centre. It focused on increases in yields per acre and recommended managerial changes to achieve this0 The main recommendations were: (a) early planting (best 1/ As reported by R. Hulls in "An Assessment of Agricultural Extension in Sukumaland, Western Tanzania," ERB Paper 71.7 (Dar es Salaam: Economic Research Bureau, University of Dar es Salaam, 1971); and by G. Saylor in "An Evaluation of Agricultural Extension in Sukumaland," (Paper to the East African Conference of Agricultural Economists, 1970). results are obtained when planting is done by mid-December); (b) close spacing (on five-foot ridges, cotton plants should be two rows to the ridge with 18 inches between plants in the row); (c) thinning to two plants per stand and early, clean weeding; (d) tie ridging, that is, forming small dams between the ridges to hold rainfall in situ, ensuring maximum availability to the crop; (e) use of fertilizers, two hundred weight of 21 percent sulphate of ammonia and one cwt of 20 percent P20 single superphosphate (the single superphosphate was to be applied in the seed bed, and the sulphate was to be top-dressed after thinning and weeding, at about three weeks old, then applied to the hillsands in the north and central growing areas, an estimated 300,000 to 400,000 acres of cotton); (f) the application of insecticides (a six-spray regime should be used, beginning the tenth week after germination and spraying at ten-day intervals, using DDT, 75 percent wettable power, against boll- worm in the north and central areas, and DDT/SEVIN against the bollworm stainer complex in the south and eastern, drier areas; the spraying recommendations were only made in 1965). 82. Other recommendations incluOed the need to uproot and burn the 'plants at the end of the season as a mianitation measure against the carry- over of pink bollworm. C. Tie Ridging 83. Tie ridging was not emphasized in extension programs during the later 1960's but formed the core of preindependence efforts, part of a preoccupation of the British administration with soil conservation. The campaigns for the ridging were based on the work of Peat and Brown. 1/ Experiments showed tie ridging gave an average 18 percent increase in yield per acre. The British tried to enforce tie ridging by ordinance during the late 1940's and early 1950's. As soon as ordinances were lifted, tie ridging faded out, except, significantly, in Ukerere where Rotenhan, in 1963, noted that the practice had persisted. Clearly, it had persisted there because, even in the 1950's, the scarcity of arable land reduced the effectiveness of old rotation techniques, and an awareness was growing among farmers of falling yields and land scarcity. 84. The extension service recommended that five-foot ridges be tied at intervals of six feet. In addition to preventing physical erosion from accelerated run-off during the rains, the bonds formed by the ties held the water, allowing its fTll assimilation by the zrop. This is particularly important since all of Sukumaland is subject to a trough in rainfall in January and February. To .give maximum effect the ridges had to be tied as soon as possible after being made, in practice in December. 1/ J. E. Peat and K. J. Brown, "Yield Responses of Rain-Grown Cotton in Lake Province, Tanganyikax "Empire Journal of Experimental Agriculture, 1962. 85. The labor requirement for making ties in ridges has been estimated at 6.0 mandays per acre.- Intensive and extensive use of labor during this season are alternatives. The net return to employing extra labor for tie ridging in the model system is Shs. 79, which must be compared to the net return of Shs. 1161 from the ridging of a further 0.80 acre of cotton using the same amount of extra labor. The extensive expansion of production is much more attractive to those farmers with suitable land available. As most Sukuma farmers were in this position, there had to be considerable induce- ments, either by subsidy or by legal enforcement, to override the economic conditions facing the farmer. D. Insecticides and Fertilizers 86. An organized program for the use of insecticides among small farmers began in 1967-68, but the recommendations for the use of fertilizer have a longer history, going back at least to 1962. The history of the fertilizer program is distorted by the use of political authority to oblige farmers to use it; the experience repeats that of tie ridging to a large extent. Consumption of fertilizer on Ukerewe Island showed a dramatic increase between 1962 and 1966 when political pressure was on all farmers to use it, or to be shut off from seed supplies. This resulted in rioting on the island and the eventual abandonment of the enforced program. The effect of fertilizers can clearly be seen in 'bhe increase in baleage per square kilometer and in production of seed cotton per head, rising from 1960 to 1965, then falling again by 1970 (see Table 15). Uptake of ferti- lizer on the mainland rose much more slowly. 87. The fertilizer recommendation was part of a package including timely planting in the first half of December, close spacing, thinning, and clean weeding. 'While pressures could be put on the farmers to use ferti- lizer, supervision to ensure the complementary management improvements was rarely available. As was discussed in Chapter 2, an increment of 400 pounds of seed cotton per acre was expected from the use of fertilizer and 650 pounds from the use of insecticides and fertilizers. Table 22 shows the cost return ratios of these expected results with the results from farmers' fields shown in Table 6. The only results available for the use of ferti- lizer alone show a net loss to the farmer, implying that the qualifying management conditons are not being met. Indeed, the whole question of using purchased inputs involving early planting and the use of extra labor in applying the fertilizer has complex ramifications vis-4-vis the farmers' existing allocation of resources. 88. Using results from experimental work, Collinson interpolated each of a set of possible innovations as an extra activity into a linear program- ming matrix representing the model in Table 23., The increase in the objective value represented the potential increment in the system from that - 48- Table 22: COST RETURN RATIOS FOR RESEARCH-STATION AND FARMERS' RESULTS FROM THE USE OF INPUTS Yield Increment Value in Cost of Value and Input per acre Shs. (50 Inspections Cost Ratio (lbs.) per lb.) (Shs.) /a_ Fertilizer Only Experimental 400 200 40 5.00 Mwanza Field Results 54 27 40 0.70 Insecticide Only Experimental N.A. N.A. N.A. N.A. Musoma Field Results 87 43 30 2.40 Mwanza Field Results 212 106 30 3.50 Fertilizer and Insecticide Experimental 750 375 70 5.36 Musoma Field Results 373 185-5 70 2.68 Mwanza Field Results 348 174 70 2.49 Source: Field evaluation of the use of fertilizer and insecticide purchased under IDA, TA 80. /a Shs. 7.50 are allowed for pump depreciation in spraying costs. Note: N.A. = not available. Table 23: ALTERNATIVE IMPROVEMENTS INTERPOLATED INTO THE TRADITIONAL SYSTEM BY A LINEAR PROGRAMMING TECHNIQUE Extra net Extra net Extra net Input income income income Planting ertilizer Inecticide (Shs.) (Shs.) (Shs.) date use use (A) (B) (C) December 7 No No 197 197 403 December 7 Yes No 141 233 638 January 1 Yes No 0 61 439 December 7 No Yes 118 345 584 December 7 Yes Yes 2 493 1,290 Source: M. P. Collinson, Farm Management in Peasant Agriculture (New York: Praeger, 1972); column a ae Note: Column A: farmer's income when total capital available for input is Shs. 150; Column B: farmer's income when credit facilities allow increase in total capital available for inputs; Column C: farmer's income when expense for inputs is subtracted from resulTing profit. - 49 - innovation. 1/ Table 23 gives selected results from these interpolations and shows the addition to the net cash income of the traditional system (Shs. 762) while maintaining the established food supply pattern. 89. In the solutions under column A, the input must compete for the limited capital of Shs. 150 available to the farmer. It is significant that the first innovation, not using either purchased input, offers the highest return. The limited increases as a result of using purchased inputs demonstrate how close returns are at the intensive and extensive margins. Except for the use of insecticide without fertilizer, the increments achieved are probably insufficient incentive to the farmer to adapt his management techniques. The importance of successfully adapting management is shown by the loss in net increment in income when planting time is delayed to January 1st. 90. Under column B, credit facilities allow for an increase in total capital employed. The Improvements requiring purchased inputs show up more favorably: The incremental income resulting from the extensive use of the same capital as in column A is much higher. 91. The problem of evaluating possible improvement in technology is more conplex than placing a value on the increment in yield and deducting the cost. Column C in Table 23 shows the result of doing this for 4.03 acres grown in the7traditional system, using the yield increment assumed for the five situations. Results from the system-analysis approach in columns A and B are very much lower. 92. Clearly, without a credit program for fertilizer until 1965-66, and for insecticide until 1967-68, and with the complex reorganization of labor routines required, there was little incentive for farmers to adopt inputs. Even with the credit programs for both fertilizer and insecticide, the requirements (timely planting, close spacing, and clean weeding) necessary to realize the potential from the use of fertilizer and insecti- cide would require close supervision and guidance in management and labor reorganization from the extension services. E. Extension Organization 93. Studies by G. Saylor and R. Hulls'have indicated that the exten- sion services are ineffective. 2/ Hulls' study investigated the extension contact among a sample of 221 cotton growers. Two of his tables are summarized in Table 24, which shows both thn concentration of contact staff in four districts and Hulls' findings on the degree of contact in these districts. He distinguishes between advice given personally by the contact staff and advice received from secondary sources. 1/ Collinson, Farm Management in Peasant Agriculture (New York: Praeger, 1972). 2/ Hlls,"Assessment;" and Saylor, "Evaluation." F. Credit Programs 95. Table 23 demonstrates the importance of credit programs in increasing the incentive for cotton growers to adopt improvements in farm practice. Since 1960 the government has sought to supplement research and extension services with credit facilities for smallholders in the cotton areas as in the rest of the country. In 1961 the Agricultural Credit Agency (ACA) replaced the Land Bank, an organization that financed large- scale agriculture on registered land. The ACA had a responsibility to reach small farmers operating under the traditional tenure. Loan committees made up of government personnel were established in each district to administer the ACA's program locally. 96. In the cotton areas, as in most other areas of Tanzania, the small-farm credit situation poses a classic problem: very small farms, no land titles, low loan requirements giving nominal interest revenue for each loan, yet loans demanding as much evaluation and supervision as larger ones if the investment is to be suocessful. The district loan committees (DLC's) of the ACA, set up in 1961, ran into this type of problem. The DLC's were composed of officers from the district executive ministries and were chaired by the representative of the administration, with the AO as the executive secretary. They received loan applications from individual farmers and usually interviewed applicants. This rapidly preoccupied district officials, usurping their regular responsibilities. Despite this administrative attention, only 4,716 applications were processed and 2,686 loans approved throughout Tanzania in 1962, the busiest year under this system, represent- ing perhaps 0.3 percent of the rural population. In addition to this lack of penetration, recovery procedures were naive in the extreme; borrowers were expected to bring payments to the executive secretary or local revenue office at the appropriate time. This was particularly inconr aus for the district AO, who became both debt collector and adviser. 97. Lending through the DLC's began to drain the funds of the ACA. The average size of district loans, about Shs. 1,000,clearly indicated that the DLC system was not reaching the small peasant farmer, to whom this would be an excessive level of capitalization. As Table 25 shows, the agency increasingly began to lend to cooperatives. The number of loans remained low but the amounts loaned were high, giving the agency a reason- able operating margin. 98. This trend toward cooperative lending was crystallized by the National Cooperative and Development Bank Act No. 38 of 1964, which disbanded the DLC's and established the National Cooperative and Development Bank Group (NCDB). Within this group, the National Development Credit Agency (NDCA) focused lending through the marketing cooperatives to reach the small farmers. Collinson estimated that by 1969 over 100,000 farmers per season were being reached through the cooperatives. 1/ 1/ Collinson, ?Tanzaniats Co-operative Movement and Farmer Credit in the 1960s,17 mimeographed, 1970. 5 2 -I Table 2F: AOA LOANTS, 1962-64 District loans Executive loans /a Cooperative Loans Year Number Shs. Number Shs. Number Shs. approved thousand approved thousand approved thousand 1962 2,686 3,186 328 h,osh 7 392 1963 2,559 3,295 197 1,908 33 1,387 1964 999 956 50 1,300 20 6,101 /a Executive loans were large loans, over Shs. 10,000, handled directly by the head office of the agency. 99. Although the act laid down broad policy objectives for the NDCA, there were no detailed procedures for cooperative lending. The NDCA loaned funds to cooperatives with little interest in how these funds were utilized. Technical assistance, as part of the IDA credit TA 80, to be disbursed by NDCA through the cooperatives strengthened the agency's manpower, and cooperative-loan procedures were worked out between 1967 and 1969. In 1971, while the rest of the NCDB group was absorbed by the nationalized commercial bank. NDCA was renamed the Tanzania Rural Development Bank (TRDB) with enlarged responsibilities, particularly vis-4-vis the development of rural industries. 100. The cooperative primary society is today considered a viable unit for effective supervision and also a unit close enough to know the farmers. Loan application. There are two overlapping phases of loan application. At the society's annual general meeting application in prin- ciple is approved; then the society indicates to its cooperative union that it wants a loan program for whatever range of requIsites have been requested by the membership. If there is sufficient support from its societies, the union applies to the TRDB. Most applications are for purchased inputs so they have to be made well in advance of the season to allow time for purchase and distribution. Once a loan is approved, members are invited to apply as individuals for hard credit from their societies. The union's application and the TRDB's appraisal of it are usually based on current usage in the area, and the total of the individual member's applications tends to equal the estimate of the union. Appraisal. TRDB's appraisal of the union's loan application has two facets: (aF atechnical side, which evaluates the impact of the inputs on the local farming system and the capacity of the local extension services to supervise implementation; and (b) an administrative side, which evaluates the capacity of the cooperative societies included to administer the loans. The questions on the application form must be supplemented by two reports; a technical one from the regional AO and an administrative one from the regional cooperative officer. The decision of whether the union should have the loan and which societies should participate rests with TRDB. Disbursement. Farm supplies are purchased in bulk by the TRDB, which aggregates requirements of all unions that have loans approved for particular inputs and puts the total out to contract. Successful contractors are paid by NDCA and arrange for delivery to the cooperative unions on the date specified as a contract condition. Almost all credit is given as supplies; the unions distribute them to their participating societies and the societies to the farmers. Recovery. The societies recover funds by withholding payment for crop deliveries from farmers. The rate of recovery depends on the cost/revenue relationship, but it is rarely higher than 20 percent of the value of deliveries and should be agreed upon by the membership when they approve a loan application at the annual general meetings. Societies are responsible for repayment out of their general funds of recoveries not made from individuals and must make remittance to the unions, who repay TRDB. 101. Supervision of the cooperatives that borrow from TRDB is a joint function: The agricultural extension services cover the use of supplies on the farm; the cooperative development division covers the financial and general administration of the cooperative movement; and the TRDB staff covers the loan channels from the bank to the farmer and back. Bottlenecks discovered at any stage--in application, disbursement, or recovery--are reported to the executive ministries concerned. Follow-up visits are made to report on action taken, and this position is considered when further applications are received from the particular cooperative involved. 102. Problems remain and are partly attributable to a political enthusiasm for development, which doesn't take into account the over- whelming manpower constraints in the economy. Pressure for development, for disbursement of credit, is felt strongly by the bank's management board, mainly civil servants. Locally, civil servants are under the same pressures; they become overoptimistic about the demand for credits from small farmers, and they oversupply inputs. Further pressures result: interest accumulates from the carryover of stocks, and lack of recovery penalizes the cooperative's general funds and, in bad cases, producer prices for the crops concerned. This creates a further political squeeze: overdues accumulate, and the bank becomes the sufferer. 103. Additionally, the local extension services cannot supervise the use of the inputs that reach farmers, often due to a lack of organization, but also because they are diverted to more specialized developments: In the early 1960's these were the big transformation projects, latterly, the Ujamaa villages (see section below). Poor results by farmers trying out the inputs not only mean poor recovery but also no sustained acceptance of these improvements in farm practice. 104. Overstretched manpower occurs at all levels of the extension and cooperative development services and in the cooperative movement, where it is exacerbated by a wide variety of new roles being tacked on to crop marketing in an effort to widen the cooperative element in the economy. Input distribution is of course one of these additions. 105. There has been a basic failure in the improvement approach mounted by the government throughout the period under review to reconcile its own objective, the increase of cotton production, with the resource-endowment pattern and with the objectives of the farmer, particularly his overriding goal of a secure food supply. The criterion adopted by colonial and post-colonial programs alike has been an increase in yield per unit area in a system with access to plenty of land in most areas, but with severe seasonal shortages of labor. G. Government Efforts to Transform Production Organization in Agriculture 106. The postindependence policy of transformation probably sprang from both the expert opinion set forth in the IBRD report The Economy of Tanganyika (1960) and from a political impatience with the diffuse plodding, and unspectacular improvement of the existing approach. Although it was recognized in the IBRD report that improvement efforts should cDntinue, the report conclusively pinpointed the advantages of a transformation approach: ItThe ultimate ends to be arrived at by 'improvement' and 'tra nsformation' programmes are the same. However, the mission concludes that quicker pro- gress towards these ends is likely to be made... by planned settlements of empty areas than through exclusive concentration on the improvement of methods in settled areas" (p. 131), 107. The independent government apparently saw this emphasis in strategy as a vehicle for two policy objectives: (a) to achieve minimum standards of housing and services in the rural areas; and (b) to introduce the machinery and sophisticated technology it associated with the wealth of expatriate farmers. There was, and remaTs, a widespread belief that farmers can be serviced more effectively if grouped together: Ectension contact is easier, the distribution of inputs easier, and the supply of central services, such as water, schooling, and medical facilities is certainly more convenient. 108. The government established a Rural,Settlement Commission, a body corporate, to implement transformation projects. In 1963 it was renamed the Village Settlement Agency. The agency was in charge of a major program in the first five-year plan, 1964-69: Its function included planning, organization and management, and it required qualified staff on both the college graduate and career diplomat levels. It drew staff from the executive ministries, particularly from the Ministry of Agriculture, leaving large gaps in the hierarchy of the agricultural extension services. Although villagization as it came to be called, was abandoned by 1966, it was not until 1969 that overall reoponsibility for agricultural development was restored to the Ministry of Agriculture. Villagization was abandoned with only five of the proposed schemes started. (Neuwiger recorded the average capital cost per family at over E1,000. 1/) One of the five established was at Rwamkoma: Its cash crops were coTton and groundnuts, and it used machinery for the initial cultivation. 109. A second transformation project, this time under the auspices of the Ministry of Agriculture, involved the introduction of large numbers of tractors into the cotton-growing areas. Problems of underutilization due to the difficulties of organizing hiring services for small farms prompted the minister to inaugurate a tblock farm" scheme. Under this scheme, large blocks of land, from 300 to 500 acres, were cultivated by the tractors, ensuring their utilization; the areas were then allocated to local farmers who planted, fertilized, weeded, sprayed, and harvested a cotton crop under the close supervision of the extension services. 110. The potential for increasing the scale of the farming system by using extra labor or capital resources over the cultivation period has been well documented by Collinson and C. K. Kline. 2/ Over the whole of Sukumaland, wide use is made of hired labor (see Table,21), and in the southern and eastern areas, much use is made of ox ploughs and tractor hiring services. In these latter areas the peak labor requirements are increased due to the shorter growing season; the land is flatter and the potential for mechanical cultivation is more straightforward. Rotenhan reported that all farmers surveyed in Shinyanga region owned and used ox ploughs. 3/ Collinson reported that 50 percent of the farmers surveyed in a part of Mas7a region used ox ploughs, and the other 50 percent supplemented their family labor by hiring tractors for cultivation. 4/ In the older areas where tie ridging is the dominant method of cultivation and the topography is steeper, mechanical cultivation is much less suitable. This can be demonstrated by using the data from Table 26. 111, The initial limitations on cotton cultivation when using only family labor are the weeding and thinning requirements in January, operations which tractors cannot perform. The use of tractors immediately involves significant re-timing 6f labor routines and jeopardizes traditional patterns of food-supply cultivation. Complications in management arise similar to 1/ Neuwiger, 1968, 2/ Collinson, "Lugurn Ginnery Zone, Maswa District;H and 0, K. Kline, Ag tural Mechanization in Equatorial Africa (East Lansing: Michigan 3/ Rotenhan, "Cotton in Sukumaland," 4/ Collinson, "uguru Ginnery Zone, Maswa District.?' Table 26: HIRED LABOR NEEDS RESULTING FROM THE USE OF MACHINERY 11N COTTON CULTIVATION (mandays) Nov. Dec. Jan., Feb. May June Available family labor 55 55 55 60 60 60 Food-crop labor needs 45 22 39 42 12 10 Residual family labor 10 33 16 18 48 50 Cotton labor needs (per 1.80 acres) Cultivation 5 23 -- -- -- -- Weeding -- -- 16 9 -- -- Harvest 14-- -- - 11 18 Remaining family labor 5 10 -- 9 34 32 Total labor needs using mechanical cultivation for next 3.20 acres /a -- 16 16 16 26 32 Labor needs for mechanical cultivation -- 6 6 8 10 Hired labor needs -- 6 16 /b 7 -- -- /a To the limit of the family harvest labor. /b Part of the January weeding requirement is brought forward into December. Mechanical ridging allows earlier emergence than does hand ridging. those that occurred when fertilizers and optimal planting times were instituted. The use of labor, on the other hand, has flexibility in both timing and the operation performed. Given the reorganization of the system, there is scope for the use of tractors in ridging, the most labor intensive of the cultivation operations, but if the traditional food-supply pattern is to be maintained, tractors must be supplemented by hired labor for weeding. 112. The use of machinery will allow the family to grow five acres of cotton, an increase of one acre. The farm budget, with no change in the food-supply pattern, is as follows: Gross cash income (5 acres x Shs. 226.30) = Shs. 1131 .50 Tractor costs (3.20 acres x Shs. 50.00) = Shs. 160.00 Labor costs (29 mandays at Shs. 3.50) = Shs. 101.50 Total costs = Shs. 261 .50 Net cash income = Shs. 870.00 The net cash income of Shs. 870 is an increase of Shs. 108 over that of the existing system. Tractor use involves an increase in outlay of Shs. 111.50, but this additional outlay would have hired 32 more mandays of labor than in the existing system, which could have expanded the cotton area to 5.40 acres giving a gross cash income of Shs. 1222, net Shs. 960.50. In addition to hired labor being more flexible in timing and operation, it gives a better return on capital employed than machinery. When a supply of casual labor is available it pays the farmer to use it. Nevertheless, in spite of the ocwplexities outlined, there is scope for mechanization in the farming system of Sukumaland. The concept of a mechanical component in transforma- tion was not ill-conceived. With this in mind, the village-settlement and block-farm programs are examined in more detail. H. The Village Settlement Scheme: Rwamkoma 113, Rwamkoma, following the ooncept of the village settlement program, was planned to have 250 families with their houses grouped in the village. The village was supplied with water, a school, a dispensary, and a social center. Each family was supplied with a concrete-based house that had a steel frame and corrugated-iron sides and roof. Each family was to culti- vate six acres of cotton and four acres of groundnuts for cash, an acre of grain, and small garden plots for vegetables. The land was to be prepared by machinery, but planting, much of the weeding, and all the harvesting was to be done by family labor. 114. Before the scheme was fully established it became clear that the families could not cope with the work involved, particularly the harvest of cotton and groundnuts, which coincided in May. In addition, there was friction between the field officer, an expatriate appointed to manage the scheme, and the growers' committee,which eventually was to take over manage- ment. The manager found his actions frustrated at the critical development stage of the scheme. Because of difficulties in the day-to-day decision making and because of the acute labor shortage, operations were delayed and neither the planned acreages were planted nor expected yields achieved. By 1966 Rwamkoma was abandoned as a village settlement. 115. The failure can be attributed to two main reasons. First, the scheme was a classic case of planning a system of production organization from welfare objectives, instead of evaluating the overhead that the production opportunities and available technology could support. Nice houses and additional nentral services could only be carried by a farming system whose labor capaci.ty exceeded that of the typical farm family. This was exacerbated by the fact that relatively young families were attracted to the scheme, first generation families with lower levels of available labor. The use of machinery was planned to alleviate tle scarcity of labor, but the production level required to support the general overhead was beyond the increase possible from mechanizing the cultivation operations. This being so, the heavy additional investment in machinery to cope with acreages the families could not manage aggravated the overhead, burdening the scheme. Secondly, although it was an explicit long-term objective that the village settlement run their own affairs, allowing infant committees to hinder the authority of the manager during the development period stifled the energy and decision needed at this time. I. Block Farms 116. In terms of their planned contribution to cotton production, the block farms were more succes3ful than the village settlements. In announc- ing the blocks as the basis on which the tractors, newly brought into the area by the Government in 1963, would be organized, the Minister of Agri- culture proposed to cultivate 45,000 acres for the 196h-65 season, an average of 300 acres per tractor. Extension service staff were given the responsibility of selecting blocks of land suitable for cGtton growing and tractor cultivation. The administration had the task of recruiting local farmers to grow cotta on the cultivated areas; the extension services were to supervise planting, weeding, and the use of inputs. Not only were the schemes to give an increased cash-crop acreage per participating family,but the yields on the increase were to be high, estimated at 1,200 pounds of seed*cotton per acre, almost three times the local average. Because labor requirements for harvesting have been shown to limit the extent of supple- mentary acreage, the increase-yield element would have been significant in planning the allocation of block acreages to individual farmers. 117. The program never achieved expectations. According to De Wilde, 41,000 acres were selected for 1964-65, of which 13,686 acres were cleared and 7,987 acres planted. The impetus of the program was lost as problems in administration emerged in the 1964-65 season, and the acreage of blocks throughout the cotton areas never exceeded 10,000. Most blocks reverted to fallow or to their previous owners by 1968, although one or two persisted and were incorporated in the Ujamaa program after the Arusha declaration in 1967. 1/ 118. J. Pupius has recorded the effects of the block schemes on cotton production in the lake region and has calculated the yields necessary to meet or improve iarmers' present incomes (see Table 27). His report stresses that farmers tended to substitute cultivation of the block farms for that of cotton acreages on their own home farms, although the degree of substitution varied from area to area. Pupius estimated that costs on the blocks were Shs. 275 per acre of cotton compared to Shs. 51 .40 for a sample of local farms. 2/ De Wilde reported that cotton yields on 4,812 1/ J. C. de Wilde and others. Experiences with Agricultural Development in Tropical Africa, Vol. I Synthesis, Vol, II:_Case Studie (Balimoe John Hqpk#s Press, 1967). 2/ J. Pupins, "The Results of Block Mehanization Schemes in Sukmaland," mimeographed, 1965. Table 27: RESULTS OF TTE BLOCK SCMIES IN TIREE DISTRICTS Cotton Acreage Yield levels to Non Rate of mafntain incomes Partici- partici- Substi- Present District pating pating Increase tuti6n incoibes + 10% + 20% farmers farmers/a % (Shs.) Geita 5.h9 4.27 1.22 39 802 882 962 Kwimba 4.03 3.23 .80 67 918 1,009 1,100 Mwanza 2.79 2.75 .04 98 1,059 1,165 1,271 /a Cotton acreage for nonparticipating farmers is from farm management surveys in the districts concerned, acres in Mwanza region averaged 41 pounds per acre, and on 902 acres in Shinyanga region 266 pounds per acre. To ensure farmers that their incomes would not suffer through participation in the block schemes, they were allowed the first 400 pounds per acre without deductions and, since costs in some areas were as high as Shs. 275 per acre (or over 500 pounds of cotton), there was little chance of recovering funds invested, 1/ 119. Although, as De Wilde noted, mechanical cultivation coald relieve the labor constraints that limited the scale of the existing system, more easily in the southern and eastern areas than in the north, the problems associated with achieving this were aggravated by hasty improvization. Some of the major obstacles were: (a) Little planning was done. It was intended that farmers should have 4.00 acres each on the blocks; the actual average acreage was 3.18.2/The difference between supplementing tie farmers' labor resources and substituting expensive capital for cheaper labor was never appreciated. As Pupius noted, the block schemes "involved a shift out of low cost production profitable with low returns to high cost production requiring high returns;" sufficiently high returns did not materialize. 3/ 1/ De Wilde.and others. Experiences with Agricultural Development. 2/ Pupius, "Results of Block Mechanization Schemes.1y 3/ Ibid, -60- (b) A high proportion of tho block acreage was in areas of denser population, Nwanza, Musoma, and Kwimbn. Tn these oreas substi- tution was most marked. Also, establishing the blocks often involved taking over people's land, which alienated the partici- pants to a point that cooperation was reduced. (c) For the organizers, faced with a target of 45,000 acres, and only six months of preparation, a major problem was that where land was available there were no people and where people were available there was difficulty in taking over land already in use. Local communities tended to unload pieces of land known to be infertile. (d) Very few deductions were made for the recovery of debt: As the blocks, by and large, were in the middle of other cultivated areas, farmers sold the crop with their home-grown cotton through their primary societies, claiming it was from their home fields, even though to encourage farmers to market through the special channels provided they were allowed the initial 400 pounds per acre without deduction. 120. Clearly, given the limited time for preparation and the lack of resources available for what amounted to a management task on the blocks, there was no hope oi' a successful scheme. It will have been a worthwhile experience if future efforts at mechanization take note of the failings due to overhaste in implementation and a lack of planning. 121. The village settlements and the block schemes were the antecedents of the Ujamaa philosophy currently dominating Tanzania's efforts in rural development. The welfare objectives have been made secondary goals of the investment program, and the emphasis has shifted from villagization to living together on a communal basis as is traditional in the African way of life. Mechanized means of production have been dropped, and the new emphasis is on the use of family labor. 122. Neither of the government's positive strategies for development during this period had any significant impact on the rate of expansion in cotton production. The improvement appreach was feasible but handicapped by program content that was only marginally appropriate to the economy of the farming system during that period. Latterly, the program, although more and more appropriate in the more densely populated part of Sukumaland. has been sacrificed, particularly in terms of high-level personnel, to the more politically appealing transformation projects. The early transforma- -tiun effort at villagization was conceptually misguided by an attempt to give politically desirable welfare standards to participants with insuffi- cient thought for the production base that had to sustain the level of investment supplied. The block schemes, on the other hand, were unsuitable to the social structure and tenure system on which they were superimposed, and although the principle of an increase in scale of the cultivated area fitted the historical pattern of cotton production, the insistence on 61 - intensification based on substantially higher costs was ill-conceived in relation to the size of the increase in acreage. Heavy pressures were put on the family labor supply, reflected both in the substitution of block- cultivated cotton acreage for farm acreage and in an increase in the use of casual hired labor. V. BENEFITS FROM THE EXPANSION OF THE., COTTON CROP 123, Three major problems arise in assessing the benefits from the increase in cotton production over a twenty-year period; first is the lack of data, particularly on secondary effects; second is the vagueness of many of the linkages; and third, and perhaps the most awkward, the impossibility of speculating on the opportunity costs of concentrating on cotton to the exclusion of other known and unknown alternatives. Two major sets of linkages can be discussed qualitatively and data used to give some broad qua tification: the industries and services that process and market cotton and the industries and services that encourage increased spending from the higher incomes earned by cotton farmers. Whether this second group would have arisen had cotton not been developed is difficult to say; some alternative source of income is likely to have emerged had cotton not been adopted. A. Farmers' Benefits 124. By using data quoted by Rounce 1/ as a bench mark of farm incomes in Sukumaland in the late 1940's and early 1950's, and comparing it with data from several sources for the middle 1960's,we can compute the increased cash income from cotton over the period. Rounce's data gives typical farm cash income as Shs. 92 per family. Of this, Shs. 4h were from cotton sales, Shs. 13 from other crop sales, and the balance of Shs. 35 from other sources, predominantly livestock. For the 1960's only the Central Statistical Bureau's household budget survey in 1966 covers all sources of income and gives averages for Sukumaland as a whole. 2/ Other studies concentrate on agri- cultural income, sometimes only fro crop sales and usually only for small areas of Sukamaland. 3/ The CSB is household budget survey, however, covers only a six-month peri-d in 1966, including most of the cotton harvesting and marketing, but making it difficult to estimate annual totals from other sources. Table 28 summarizes the results,for income and expenditure from the household budget survey in 1966. 1/ Rounce, "Economic Development of Sukumaland." 2/ Tanzania, "Report on the 1966 Household Budget Survey among Cotton Growers in the Lake Region." 3/ Collinson, "Farm Management Surveys 1-4;" Rotenhan, "Cotton in Sukumaland; and Bureau of Resource Assessment and Land Use Planning, "Preliminary Report." - 62 - Table 28- INCOME AND EXPENDITURE FOR A SMALLHOLDER FAFN IN SUKUMALAND, MAY 1 - OCTOBER 31, 1966 Income Expenditure Item Shs. % Ite Shs. % Agriculture 633.70 76.7 Food 82.5o 18.1 Cotton 562.4o 68.0 Meat 19.20 4.2 Maize 18,20 2.2 Fish 15.70 3.4 Cassava 17.10 2.1 Maize 10.10 2.2 Finger millet 5.30 0.6 Cassava 8.40 1.8 Tobacco 4.80 0.6 Sugar 8.10 1.8 Groundnuts 3.50 0.4 Finger millet 4.10 0.9 Bananas 3.40 0.4 Salt 2.60 0.6 Sisal 3.00 0.4 Rice 2.20 0.5 Sugar and cane 2.60 0.3 Other foods 5.70 1.3 Other 13.60 1.7 Maize or rice milling 5.80 1.3 Outside meals .60 0.1 Livestock 93.70 11.2 Sale of stock 74.60 9.0 Drink 22.10 4.9 Hides and skins 10-80 1.3 Pombe 19.40 4.3 Meat 5.4o 0.6 Bottled beer 1.50 0.3 Milk and ghee 2.60 0.3 Moshi 1.20 0.3 Eggs .40 0.0 Egs.0 00 Tobacco 4.00 0.9 Other local produce 38.30 5.8 Beer 23.80 4.0 Clothing 112.40 24.6 Fish 4.50 0.6 Medicine 3.00 0.4 Shoes 4.5o 1.0 Other 7.00 0.8 Housing and household 58.90 12.9 Other Income 48.70 6.8 Furniture/equipment 28.90 6.5 Wages 44.40 6.4 Building and repair 2' .90 4.8 Selling used bikes 2.30 0.2 Fuel (mainly kerosene) 8.10 1.6 House rental 2.00 0.2 Farming Expenses 74.70 16.4 TOTAL 814,50 100.0 Purchase of livestock 33.30 7.3 Hired labor 22.70 5.0 Transport 13.00 2.9 Machine hire 5.40 1.2 Other .20 0.0 Miscellaneous 97,10 21.2 Taxes 38.10 8.4 Personal transport /a 29.40 6.5 Medical care 11.00 2.4 Personal effects,etc. 7.30 1.4 Education 6.8o 1.5 Recreation 2.80 0.6 TANU 1.70 0.4 TOTAL 456.20 100 /a Personal transport includes bus fares and the purchase and repair of bicycles. 63 125. The survey ittempted to estimate this data for a full year on three assumptions: (a) that all cotton payments are made within the studied period; (b) that total incane from "other soreces" is the same for this period as for the rest of the year; and (c) that total expenditure is the same for this period as for the rest of the year. Other studies suggest a higher income from cotton, between Shs. 700-1,000 gross. 1/ November is still a cotton marketing month, suggesting that some extra cotton income is due then. Some categories of income are other farmer's expenditures, notably sale of livestock, beer, medicine, and meat. These expenditures plus others such as most miscellaneous items, household expenditures, and clothes are most likely taken out of money from the cotton harvest, the largest inflow during the year, and are probably not repeated in the other half of the year. At the same time, the period covered was the harvest period for subsistence crops, so relatively few foods would have been purchased. Other studies show an average expenditure of Shs. 100-to 150 for hired labor or tractor services, mainly in December and January, during the period covered. Putting these three factors together, an estimate can be made for annual income and expenditure on main items (Table 29). Table 29: ESTIMATED ANNUAL INCOME AND EXPENDITURES FOR A SMALLHOLDER FARM IN SUIUMALAND IN THE LATE 1960's INCOME EXPENDITURE Item Shs. Agriculture 927,60 84 Food 247.30 Cotton 700.00 64 Drink and tobacco 33,10 Other crops 71-30 6 Clothes and shoes 116.90 Livestock 115.90 10 Housing and household 66.10 Other stock products 40.40 4 Farming expenses 18.oo All other 178.20 16 Livestock purchase 3.0o Other local produce 57.45 5 Labor, transport, Other incomes 120.75 11 and machines 150.00 All other and livestock 334-50 30 Miscellaneous 14.60 Total 1,10r.8o 100 Total 979.00 In Tible 30, a comparison of the very crude estimate of annual inccme in Table 29 is made with Rounce's estimate for 1950. Table 30: MAIN SOURCES OF CASH INCOME FOR A SMALLHOLDER FARM, 1950 AND 196& Yeaon Other Crops Other sources Total Shs 1950 48 14 38 92 1966 64 6 30 1,106 1/Ibid. - 6 - 126. Of the much larger cash income accruing to the typical Sukuma farmer, cotton revenue shows an increased share, clearly the main benefit of the expansion of cotton production. Secondary effects of this primary source for increased cash are apparent; trading between farmers, and by third parties acting as intermediaries between farmers,has been stimulated. Examples in Table 28 are: trade in staple crops and meat, livestock transactions, .brewing and the sale of local beer and the exchange of wages insofar as they are earned by laboring for other farmers. Attempting to reconcile the income and the expenditure sides of the table suggests that a minimum of 15 percent of gross farm income in the 1966 figures are generated by interfarmer transactions, a sum of about Shs. 167, significantly higher than the Shs. 92 total cash earned in 1950. 127. Farm surveys have shown a significant range in incnes in Sukumaland. 1/ Unlike in many developing economies, these ranges are unrelated tostructural characteristics of the community, which are effecj- tively undifferentiated. Farm surveys show incomes are related to available labor in the less densely populated areas, and to available land area in the most densely populated areas. Landholding is influenced"primarily by family size, but in areas where land is becoming scarce, fragmentation due to division of inheritance is creating a larger.disparity in holding size than was characteristic in traditional society. 128. The data from the household budget survey is from a 776-farmer sample throughout the area and necessarily masks localized specialities. Purchases of foodstuffs will be greater in areas with relative land scarcity, and sales will be greater in areas that supply those that are deficient. The crude data in Table 30 on food production:patterns for 1950 and 1966 suggest a switch in the diet from grain to root staples, perhaps with adverse nutritional effects. However, the data is too approximate to justify any comparison. 129. The data in Table 28 point to the farmers' increased contacts with the outside economy. Income data shows contacts with the national marketing institutions (LSMB for cotton, Tanganyika Packers for livestock, and the National Agricultural Product Board for grains) or with the urban areas (-for example, Geita supplies Mwanza and Ukerewe with large quantities of cassava). Expenditure data show significant contacts with clothing traders, furniture makers, building-supply agents, bicycle sellers, and maize or rice millers. These categories absorbed 46 percent of the expenditure for the six months covered by the budget survey and 25 percent of the estimated annual expendi- ture. Contacts with government bodies for tax payment, medical care, education,and political subscription cover 12.5 percent of all expenditures shown for the six-month survey period and 9 percent of estimated annual expenditure. Finally, the purchase or hire of resources for use as inputs on the farm, mainly livestock and labor from within the farming community, 1/ Ibid. supplemented by transport and machine hire from outside, account for 16 percent of the survey expenditure figure and 19 percent of estimated annual spending. B. Area Benefits 130. Many of the area benefit; of cotton expnsion spring from the increased contact between farmers and the other secbors of the economy, particularly tho conttcts built up as spending increased on consumer goods and farm services. Over the period of expansion to 1966 cotton beA been rn export crop, with cnly primary processing done locally. Latterly, with the development of a domestic textile industry- using Tanzanian cotton as raw material, wider secondary benefits have begun to be felt. At the same time, seed as a by-product of cotton lint has led to a domestic cottonseed oil industry. Although for the emergent spinning and weaving industry the supply of raw materials is probably not a vital factor in location, for the oil extraction industrybecause the final product is a low proportion of the raw material,locating the mills within the cotton areas is bene- ficial. Finally, the marketing infrastructure, and to a lesser extent the infrastructure for farm supplies ansociated with expanding cotton production, has been an importrnt local development. 131. Quantifying the effects of cotton expansion is hampered by a lack of data. The fanming system shows some developnents in food and livestock, though this is of less imnediate interest. Comparing the incame and expen- diture sides of the budget survey, meat and fish represent items brought into the farming community through third parties. There are no comparative statistics to show whether the trade has become more specialized over the last 20 years. It is probably that barter trade has been replaced by cash transactions over this period. At the same time, data show significant amounts of maize and cassava moving out of the household ecconmy. Several channels are available for surplus foods. Since 1965 the National Agricul- tural Products Board has marketed maize and rice on a national scale. The urban market, mainly in Mwanza and Shinyanga, is very limited in terms of the farm population available to supply it, although particular areas do produce for urban needs: For example East Geita produces cassava and bananas for Mwanza. Other minor localized cash crops find special outlets: Chickpeas are grown off-season extensively in South Mwanza and exported through the cooperative hierarchy. 132. The earliest available data classifying processing and manufacturing industries on a regional basis are the Surveys of Industrial Production for 1956, 1958, and 1961. These have two significant limitations: (a) Since the latest 1969 survey, classifications Ive been formalized and coded so that little comparability remains with the earlier surveys. (b) The early surveys were based on voluntary response, and the sample basis was poor. For example, the 1958 survey covered units with more than five employees; 1,442 question- naires were sent out, but only 336 capleted forms were analyzed. In 1961 the survey covered all units irrespective of numbers employed; out of 2,872 qontacted, 1,610 were eliminated and 1,0h responded and could be analyzed. - 67 estimated to employ 200 workers each. The Dieotory also gives the date the plant began manufacturing (See Table 32). Table 32: THE GROWTH OF MANUFACTURING INDUSTRY ASSOCIATED WITH COTTON PRODUCTION Pre-195 19-0 1971 Industry mployees Eployees Employees No. (Est.) No. (Est.) No. (Est.) Ginneries 2 400 19 2,200 23 4,,6oo Textiles .. .. ... 3 /a 530 Oil and soap mills 5 725 10 01l 3 soap 1,440 Metal-box ... .. * 1 35 Sourcet Tanzania, Central Statistical Bureau,"Directory of Industries 1971.11 /a Two of the textile manufacturing establishments are for shirtmaking with an estimated 1 employees each. The third is a spinning and weaving mill with over 500 employees. By comparing these estimates with the total employment registered in manu- facturing by the Survey of Industrial Production for 1969, the establishments related to cotton processing are shown to account for between 75 and 90 percent of employment opportunities in the area. With the emphasis previously on lint exports, only relatively recently, with the growth of oil processing and spinning and weaving for the local market, has cotton expansion begun to have full multiplier effects in the local economy. 133. Effects from increased farmer cash spending are difficult to trace. Major purchased items on the budget survey are clotheo, bicycles, furniture, and building materials. On the whole, establishments making clothes and furniture are small, employing under ten workers, and are not recorded in the T969 survey of industries. Some evidence of the effects of increased demand is apparent from the official statistics. The first industry to make furniture, with over ten workers, was established in Mwanza in 1957, and in 1970 the seventh was established. In 1965 a local bicycle-assembly work- shop was set up, reflecting the heavy local demand for cycles. However, most bicycles continued to be imported from the United Kingdom and Japan. In view of the expenditures for building materials by farmers two factories of particular interest indlude a manufacturer of corrugated iron sheets, established in 1954, and a cement factory established in 1966. Similarly, since 1967 three large-scale textile mills, dependent mainly on cotton produced in the western growing areas, have been established, one in Arusha and two in Dar-es-Salaam, where most large-scale establishments are located. 68 Present capacity could absorb an estimate d 75,000 bales, or 20 percent of the crop, once the rirket builds up 'o this lnvel; in 1972 the mills processed betwoen W0,000 and 50,0C) bales (dnta in Table 33 show consump- tion of' raw material and value of output for cottonseed oil milling in 1969). Finally, the raeently increased use of fertilizer and insecticide on cotton has helped to support a range of small insecticide companies and the establishment of a fertilizer plant in 1971. Table 33: LOCAL USE OF DOMESTIC COTTCN AS A ROT MATERIAL '11:67 -Q-u---------- Value- Induitry nw aterials (Shs. thousand) output (Shs. thousand) (tons) (tons) Spinning and 5,269/a (ravy 22,001, (raw ecotton) cotton) 148,88 1, 0, (yo rn) 8,67h (yarn) E,dible oil- 79, 565 35,079 10,266 (oil) 25,621 (oil) mil,ing W0,096(cake) 16,606 (cake) products Source Tanzania, Gentral Statistic Bureau, 11,Survey- of Industrial Production, 1969. /n Equivalent to just under 30,000 four-hundred-pound bales. 134. Comparison is difficult over time for oil millirg. In 198 and 1961 all oil milling and soap manufaoture were grouped, and cottonseed cannot be distinguished from coconut as a raw material. No textile establishments were registered in 1958. In 1961 there were five with a value of material inputs (all imported) cf Shs. 2.h million, or 8 percent of the 1969 figure, and a gross output of Shs,, 4.2 million, under 3 percent of the 1969 value, C. National Benefits 135. Benefits at the national level have come mainly from cotton as a foreigh-exchange earner. In 1966 cotton had the highest value of any commo- dity exported. The government only took direct income fron cotton for a short period, 1966 to 1969, when a levy was inposed on export. 69 - 136. Government revenaes from increased monies from cotton production will be given an extensive boost by axpansion. IBRD estimates an average tax incidence on incomes of 17 percont. 1/ With cotton lint, the main product, moving out of the country, the Tajor group affected are the farmers. IPRD estimated incidence on their cash incomes at 12 percent,with 20 percent on salaries and secondary incomes. With the development of the oil-milling industry, and1latterly in 1967, a teKtile industry based on spinning and weaving of home-grown cotton, multiplier effects have been enhanced and secondary incomes increased, giving higher revenues to the government. As product-Lion of home-produced cotton piece goods builds up, incleasing import substitution will also benefit the economy. 137. The momentum of expanded cotton production and a thirty-year history of government interest in marketing were the foundations for, first, the succeso of the I3MB, and, second, the establishment and development of the cooperative movement in the western growing area. 138. The IMM is the oldest and has been the most successful of the marketing boards, passing on much of its marketing and pricing experience to the newer boards, particularly the National Agricultural Products Board, formed in 1965 to handle a wide range of relatively minor cash crops. In part, the LSMBts success is attributed to the buoyant finances it experi- enced in its initial years, which allowed consolidation without immediate financial pressures. Happily, and unusually in these days, the LSNB came into operation when the world market price was rising, so it did not have to operate on a shoestring to keep up producer payments to encourage con- tinued production. In part, also, success is attributable to capable management and to the positive activities of the board; it realized that in the long term its prosperity depended on its continuing to offer incentives to the farmer in terms of both price and access to market. The board proved a valuable training ground for local staff; the last of the expatriate management left in 1970, handling the reigns to Tanzanians with many years' service on the board, 139, The cooperative movement grew from 37 registered primary societies in 19i53 to over 590 in 1971. There is no doubt that the pace of this expansion outran the availability of suitably qualified men to run the societies, creating also prcblems of bookkeeping and administration. But while there may have been doubts about control at the union level of the hierarchy, there is no doubt that the structure of the primary society gave the membership participationiin the handling and disposal of their own produce. During the colonial period encouragement of the cooperatives was seen as a channeling of political activity into a productive, as well as political, outlet. Nevertheless, it fitted closely with postindependence aspirations and methods. Each society committee has only one salaried / IBRD, The Economy df Taganyika, (Baltimore: John Hopkins University Press,, )~~ 70 - employee, the rest of the positions, seen as responsible and desirable among the community, are filled by elected memibers. All these positions are accountable at the annal general meeting to the members who can and do hire and fire. This creates a grass-roots control which is not necessarily affected by exogeneous criteria--whether the balance sheets are up-to-date or the institution is as cost effec+ive as are possible alternatives. When the government has sought to impose new roles on the marketing societies-- as distribution points for farm requisites, as organizers of production units, as retail outlets, usually hastily cnceived and implemented--the inadequacy of the manpower available beccmes apparent. VI. LESSONS FROM THE SUKTMA EXPERIENCE 140. A key question is: Are we today in any better position to exploit the cotton situation in Sukumaland if it were as it was in 1952? Could the opportunity be foreseen and the development planned be as successful as the expansion which has occurred? Clearly, variety improvenent was already a feature of the government's program in promoting cotton production, the extent of its effectiveness dependent on an increase 'in the acreage of the crop grown. Only a monitoring of the farming system and a knowledge of the resource relationships in the system would have revealed: (a) the manage- ment problems associated with timely planting of cotton--the way the farmerst objective of a secure food supply was clearly manifested by his priorities in resource allocation; (b) the possibilities for extension of acreage by the hiring of resources at critical periods; and (c) the trend toward intensification of food-crop production to accommodate an increasing cotton acreage. 141. In 1952 the methodology for monitoring traditional systems of farming was not available in Africa. Today, the methods are available but still make no significant contribution to agricultural development strate- gies. There is still no organized effort to diagnose existing farming systems. Allan has developed the concept of "human-carrying capacityt' of the natural resources of an area, 1/ Analysis of the relationship between the existing system of farming, the7farmers' resource endowments, the crop opportunities he chooses, and the methods he uses, and the natural resources of the area in which the system is practiced allow the evaluation of development potentials and the key factors in achieving these. Where there should be a cycle, farmer-researcher-farmer traffic is mostly still one-way. A. Reinforcing or Reversing System Trends 142. From the history of cotton expansion, there is evidence that reinforcing observed trends in the present agricultural system is a possible 1/ Allan, 1965. 71- basis for developnent stratsgv. In the Sukuma case, the trend toward cultivation of maize, root, and legume mixtures and away from extensively grown sorghums and millets could have been reinforced and perhaps influenced to allow a more rapid adoption by Ghe farmer population, whom the change has since reached. Certainlr the trend toward greater cotton acreages coul have been reinforced by supplying resources to aid the farmer in cultivation. Throughout the period, actual government policy was on on opposite tack, to increase cotton yields per acre. 143. On th obber hand, judging by the recent reversion to sorghums ,nd milaots because of relatively poor weather, it may have been possible, if desirable, to reverse the trend to maize-based intercrops. The recent move into sorghums and millets has been on the heavier soils, with tractor or ox cultivation providing the necessary power. Given government assistance, the move onto these soils could have been an alternative to the introduction of maize, a less reliable grain crop under rainfall conditions for most of Sukumnlond. 144. A clear diagnosis is only possible when the present farming system and its constraints are relrted to the natural resources of the area. In the Sukuma case, the followi.g factors would have been important and, to the extent that the system is still evolving, remain important: (a) The scale of the system is limited by to need for crop establishment Ir a short period, dictated by the unimodal rainfall pattern. (b) There are large areas of fertile, heavier soil- excluded from the system mostly because working them is too much for hand-hoeing. Machinery can cope with these soils. (c) Cotton, at the present producer price, gives a higher return for resources emp2oyed than any other crop; and it is a desirable crop on a national basis. (d) Maize is basically unsuited to the rainfall pattern and is subject to high interseasonal yield variations. Sorghum and millet are more reliable. From these major points, a program of cultivation aids for the heavier soils would have allowed an increasing acreage on the farm and the use of soils otherwise relatively idle to grow the preferred staple grains of the area. Land would have been released on the lighter hillsands for the expansion of cotton acreage. B. Short- and Long-Term Potential 145. Such a policy would probably have led to an increased acreage per family and reveals the need to reconcile short-term opportunities with long-term trends, particularly in relation to population density and its ccusequences for soil-fertility levels. Any gain from encouraging or reversing present trends in the farming system, or from other development strategies, must be weighted by resource balances expected in the long term. 72 146. It may be judged that ohort- or medium-term exploitative strategies can generate surpluses beyond those necessary for eventual resource renewal; thus the immediate economic gain is worthwhile. Policy must be reversed to begin to stifle such strategies before a critical point is reached. Ukerewe Island is an example where a falling per capita income is an apparent result. A conscious decision for exploitation is required, which includes some estimate of the period over which exj,loita- tive strategies can be pursued, usually being dependent on the rate of population grm!th. The experience with fertilizer in the western growing areas shows that, even with subsidy, farmers who have adopted cotton as a cash crop, adopted ox ploughs, and have had the initiative to undertake extensive migrations, will not adopt an innovation that does not give some benefit by creating adequate incentives. It is,important, then, to evaluate the level of benefit required to encourage adoption in the context of the available local economic opportunities. Again a decision is dependent on knowing the resource endowments, relationships, and productivities in the existing system and their relation to the farmers' own priorities. Unless government and farmer objectives are in harmony, or at least recon- cilable, little impact is likely. 147. It is of value in an area like Sukumaland, with a range of population densities, to use the experience in the more dense areas as a basis for decisions on strategies in the less dense areas. The point of transition from increases of scale to intensive measures poses a parti- cularly difficult problem. As has been intimated, u single generation of farmers is unlikely to become awtre of the looming soil fertility problem due to increasing population density. Planners must determine the critical point, and the government must be prepared to stabilize the economic relationships as planned once the critical density is reached, to bring farmer choices into line with government objectives. Clearly there is no inherent barrier to intensification; the agriculture of Ukara, Ukerewe, the transition to maize based interorops, are all examples that show Sukumaland's great flexibility in adapting to uhanges in the production environment. 148. The government's overall strategy for an area should delimit the program considered in planning extension as well as focus research and investigation, both technical and economic. The use of extension services in implementing selected changes is supplemented by price policies, subsidies, and, if necessary, supply cmtrol of alternatives, to channel the system into the chosen development path. C. The Sukuma Migrations 19. The movement of thousands of Sukuma into peripheral areas acted as a safety valve, ensuring that the farming system would not be cramped due to pressures of population in the older, settled areas. Again, this could be observed by regular monitoring of the farming system and could be reinforced where desirable. It is of interest that the established - 73 system of land tanur was easil ad(apted to the newly inhabited areas. Farmers had rlghts over any Iand tiy cleared initiall-y, but as population density grew the hedmn hAd the right to reallocate fallow land not being, used if increasing loca' 3emand justified it. The traditional tenure asteri i.- very flexible to changing, densities, whereas more formal tenurial arringements often involving Lhe registration of holdings of a fixed size, have failed to nllow for differences in family size or farmer motivatiois and abilities. rigidities have frequently been created in schemes set up on this basis that inhibit the rate of development of the agricultural economy. D. less.ons from Government Efforts to Encourage Cotton Production 150. The lessons to be learned are from the shortcomings of the improvement and transformation approaches adopted by government, rather than from any success. At the risk of repetition, although organizational failures in implementation contributed in both cases, improvement of cotton husbandry and transformation of production organization were il- conceived. There was a lack of appreciation for the importance of the relationships between lond, labor, and capital, which neither strategy altered and therefore had to operate within. Such relationships only become apparent from a description of the existingr farming system. 151, The two strategies are opposites. Tn improvement, the emphasin is on speeding evolution within polcy guidelines ari with stnictu.rn] changes to break bottlenecks inhibiting evolution. In transformation, structural change is seen as creating opportunities for development. Special conditions are created for production--in the village settlement and block schemes government resources were concentrated on supplying services, including tractors, sprayers, and managem ent, to the mass of peasant farmers on the islandsv-but these conditions raise many problems for replicability and for diffusion: (a) Diffusion is inhibited, often copletely. Structural changes overcone some of the limitations in the production environment of the ordinary farmer, but the improvements introduced are dependent on the change and tend to be irrelevant for him. (b) Cost per farmer settled tends to be very high because of the increased services that he is provided. Coverage, because of the limita- tion of the agricultural budget, is low, and penetration of the rural population is very limited. Together with a lack of diffusion this high capital cost allows only a small proportion of farmers to be affected. (c) Reorganization of production on this scale requires managers, found by taking staff from the middle ranks of the extension services. This need for executive staff puts a heavy drain on the output from diploma and graduate institutions and makes little use of the lower-level personnel produced in abundance by the educational system. In the case of the village settlements, there was one agricultural officer or field officer per 250 farmers on the scheme. 152. The lessons frm the improvement strategy adopted are more positive. The experience in Sukmaland was unsuccessful partly due to poor content and parly to poor organization of the extension services. Improvement can work within the context of the farmers' production opportunities: (a) It -eads to better use of highly qualified staff, allowing them to deal with large numbers of farmers through intermediate staffirg levels. (b) The content can be selected by its relevance to large numbers of farmers By the process of diffusion, innovations will be carried to farmers other than just those in direct contact with the extension services,as long as ancillary services, credit, and supply can be repeated on a sufficient scale. (o) Recent writers have shown there iz no reason why the extension hierarchy-- albeit with modestly qualified staff at the lower levels--should not be organized using the modern technique of placing control and supervision at each level. 1/ 153. The analysis of the reasons for failure of the two government efforts to stimulate production suggests that the failure of transformation is more a result of the Skmaland context, where the existing farming structure is still stable and is still meeting the needs of farmers. By implication, transformation is the last resort, the only way for develop- ment once the increase in population has broken down the effectiveness of the traditional or existing structureb At this stage the high capital costs associated with a structural change are unavoidable. By catching the situation early, before a crisis point is reached, a well-founded improvement approach may forestall the need for this ultimate step, probably only feasible, or indeed desirable, once the authorities them- selves see the traditional structure beginning to crack from the strain of a population density beyond its capacity. E. New Varieties: A "Captivel Innovation 154. The rise in prices was an obvious sou-re of better returns to cotton growing and of incentives for increasing acreage. Nevertheless, the fact that average yields rose by over 30 percent points to a signifi- cant contribution to incentives from improved varieties. Particularly over the late 1950's and 1960's the better ginning percentage and disease resistance in the field countered the depressing effects of a gradual fall in the world market price. 155. New varieties in this case were a "captive innovation.'s Obvi- ously the replicability and transferability of this experience are limited to crops whose seed is sold. In this case the situation was made particu- larly easy because of the monopoly on the sale of seed; outlets after 1961 were limited to the cooperatives. Because the cooperatives were an instrument of government policy they could be used as seed distribution channels. Kulp, 1970; and Collinson, Fam anagement in Peasant Agri lture. - 75 156. It may be feasible to induce this "'captive" element into crop production in any area, in parUciular by promoting the substitution of control processin for home processing. It would only be effective where all the crop is sold ai.ay from the fa-. At the same time, the obvious secondary benefits derived when seed is subject to continuous improvement may be sufficient to jtstify free seed issue as an 'neentive to take the new seed on offer instead of using seed retained from.the prdvious crops. Success would depend on contirnaity and reliability as well as on finding a channel for distribution, perhaps one that the farmer had to use for other purposes. F. Lesons fra the Lerience with the LSMB 157. A case can be made for or against a marketing board interfering with private enterprise. An early clash between Asian traders and pro- cessors and African smallholder growers showed signs of flaring in Sukumaland. As far back as the 1920's the government had to interfere to set a floor price for cotton. This clash escalated after the war, climaxing in the late 1960's when the cooperatives edged out the Asian processors. In this situation an arbitrator's role was important. It is difficult to imagine the smooth continuity of cotton marketing since independence and through the 1960's,when the traditional middlemen were replaced by the cooperatives, without an overall body that could place the interests of the industry forenost. Given that a political commitment to cooperatives had been made, it is difficult to conceive the old middle- men taking the positive attitude toward cooperative development and price stabilization that the board has tken. 158, The conditions listed below are reasons for the success of the LSM in the Tanzanian situation and limit the transferability of the experience. Unduiltedly other factors, attributed to the utility of marketing boards in general, also contributed to its success. (a) The LSNP inherited a history of government intervention in the marketing of the crops. From 1940 onward, there were no alternative channels for selling the crop overseas; the government ias the sole buyer. (b) It dealt with a crop that had no domestic use. All production was exported. The board was never forced into the position of fbuyer-of-last-resort' with its inherent uncertainties, nor did it have to face illegal local and interfarm sales; for the local producers, the board was the sole consumer. (a) The board was formed at a period when a booming world market allowed it to consolidate its establishment with no financial problems, creating a security never threatened until perhaps the near exhaustion of its Price Assistance Fund in 1966. I99. A final point to be made is that the IMB did not give the responsibility of trading cotton to the cooperatives, thus remov4 Ag the commercial sting that has ofter been the donifall of cooporativ/.- operating in competition with private traders. The cooperatives remain purely a collection channel for the crop. - 76 - 160. The question of what would have happened to the industry with- out a board is at best hypothetical. Given the facts that processors were a minority groap having no social contact with producers, that the cooperative movement had its roots in the ineffectiveness of the indivi- dual producer against a cartel of buyers, and that the country was now independent, it is unlikely that there would have been a satisfactory alternative to the board for the 1950's ani 1960's. G, Lessons from the Cooperative Movement in SEumaland Cotton 161. The circumstances of cooperative development have been touched on in discussing the value of the board. The tendency to compare cooperatives to entrepreneurs on the basis of commercial criteria perhaps underrates theLr role in Africa. As is implied above, their origins are more nearly comparable to those of a trade union, and the analogy of management and workerf applied to Asian processors and African small prodcers is a pertinent one. It is clear there was a local reaction against existing buyers who were also ginners. Whether the reaction was based purely on commercial grounds is difficult tD assess. While there is evidence that ginners' margins have been reduced considerably during the tenure of the board, there is no data on actual margins before the existence of the boardo The board took over the commercial role, and it may be significant that the cooperatives expanded only after the board was established despite the emergence of cotton cooperative marketing as early as the 1930's. Certainly the fact that processors were a racially distinct minority added to the reaction that grew among producers, and it probably helped to create a common purpose among them that might have been dissipated had the distinction between cultures not been so clear. 162. But whether there was a purposeful channeling of enthusiasm and energy into the cooperative movement as a diversion from political objectives is difficult to say. Certainly during the 1950's the government was planning the cooperative movement. The establishment and running of societies was closely controlled and carefully supervised. After 1961 cooperatives were given a monopoly on cotton buying. Prior to this, they had marketed over 80 percent of the crop, but with strong government encouragement in the mid-19509s and the official inhibition of expanded participation in marketing by the processors, there was no open competition. 163. Hovaver, the impetus for cooperatives did come from within the traditional communities0 There were strong local precedents: communal pasture reserves, communal grazing of animals, communal cultivation societies, and house-building groups. The tradition of the baraza, a meeting of the village men to thrash out community matters by consensus, contributed to a heritage of cooperation, closely mirrored in the organizationa of the primary societies. 77 - 164. It remains doubtful whether the cooperative movement would have developed had the trading and processing of cotton not been dominated by a cultural minority. Had private enterprise developed fran within the rural community, it would have formed powerful interests, much more difficult to unseat, This suggests a limitation in tho transferability of cooperatives to rural societies where strong commercial interests have emerged among the local people. The stronger the established commercial interests, the more difficult the emergence of cooperatives; the society will have tended to grow away from the community spirit required. The exception to this, at the risk of racial animosity, is where commercial interests are concentrated in the hands of a minority divorced from the local society0 In such cases other forces can be mobilized to create the common purpose required to establish cooperatives. 165. Given the postindependence emphasis on social welfare policies, the early establishment of cooperatives was,a satisfying developrent; they were chosen as a vehicle for African socialism. Again, this political context of low income differentials and a wide sharing of the benefits of development limits the replicability of cooperatives. Without a strong political commitment to equal income distribution, experiences elsewhere suggest that cmmercial enterprise will be more efficient in financing than cooperatives, more efficient in the short term though not necessarily in the long term. The desire for income disparities as in advanced countries creates economic warfare in developing countries; without a developed framework of law and authority, these pressures create more violent reactions. 166. As it has developed in the cotton areas the cooperative is an institution that, by its function, reaches to the individual farmer, also it is an institution under close government supervision and control. These two conditions make it a useful vehicle for rural development efforts. There has been misuse of the movement, though, and some of the points to be avoided are outlined below. 167. As one of the few local bodies generating funds, the cooperatives have had development decisions, made by political bodies, village and district development comnittees, and the central government, foisted onto them. Examples are tractors, compulsory use of fertilizers, distributive and retail services. The decisions about these added roles have not usually been made within the general meetings of the societies, and when they have, the general meeting has merely served as a rubber stamp. Financial responsibility for the outcome of many of these developments has been left with the cooperatives concerned, and thus divorced from the decision to undertake. the programs. 168. Many of these developments involved stretching the capacity of the manpower within the cooperative movemento Societies increased very rapidly in the cotton areas between 1953 and 1966, and suitably qualified personnel were not available in sufficient umbers to handle the expansion. 78 - Similarly, the government could not expand its snpervisory services fast enough to maintain standards. Society balance sheets are several seasons behin, and books are so badlr kept that the annual audit often requires a rewriting of the societies' accounbs, It was on this overtaxed structure that an inereasing number of ancillary roles were imposed, often imperiling the functioning of the societies and unions chosen for development programs. Consolidation and revitalized supervision by the central government remain a major need. 169 Only in 1967 vas a cooperative college established to train personnel for the coperative movement. This demonstrates the government's lack of a:warenss of the iplications for manpower that both the rate of expansion in coopnrative numbers and the imposition of a role wider than the original marketing function have, H. Lesons for Smallholder Credit Administration 170, One of the ancillary roles thrust on the cooperatives after 1964 was the distribution of farm inputs on credit and the recovery of principle and interest fron the marketed produce of borrowers. Much of the procedure devised is transferable to other areas; the obvious advantages of the cooperative as a credit vehicle can be duplicated: (a) The cooperative has a high degree of penetration into the rural areas, resulting in an overhead that a credit organization could not cover from the low earnings on the small loans. (b) It also has monopoly buying rights to the cotton crop and single-channel access to markets, as well as an institutional basis for repayment capacity--in terms of records of past production--as security for borrowing and as an indication of the size of loan that it can absorb. 171. The democratic base of the cooperative movement, the annual general meeting, gave the society less obvious advantages: (a) Elected committeemen from each village area know the integrity and capacity of ndividual applicants from their own areas; this was a good basis for personal evaluation of smallholders, which would be administratively very expensive to achieve by a bank organization. (b) The right of the members to elect their committeemen, and re-elect them each year, created pressures on them to carry out their selections corocientiously. At the annual general meeting, it was noted which msobers were debtors, from which villages they came, and what this would cost the society in terms of a reduction either in divident payments, transfers from reserves, or, as a last resort, a special levy. This created a social pressure on officials because of the possitility that committeemen with an undue share of debtors would not be re-elected to their positions of authority in the comAanity. 172. As discussed, manpower available to the cooperative movement in the cotton areas was drastically overtaxed in the mid-1960's, when IDA credit TA 80 was mada available for disbursement through cooperatives to smallholders. The IBRD and the central government were so eager for disbursement that the implementing body, the NDGA, had to make a conscious decision to disbarse through cooperative societies when it was known that its secretary could not cope with the bookkeeping involved. The policy - 79 - adopted was to build up a credit administration around a flow of funds rather than in vacuo. The resulting procedures are of interest: (a) Ultimat1`,IToans and interest were recovered from the farmers' profits. (b) The chain of recovery, farmer-society-union-NDCA, was par- ticularly weak at ne society level. The society at its annual general meeting was to propose the loan application to members. At the same time if members wanted the loan they would have to: (i) agree that the society as a whole would be responsible for repayment from society income; and (ii) agree on the level of deduction to be made as repayment from borrowers' deliveries. 173. If the NDOA was doubtful about the capacity of the society staff to administer the loan, the agreement would specify society income as the source of repayment. This safeguarded the agency. At the same time the agency concentrated on building up credit administration between the society and the farmer. Once the agency was satisfied that the society had the capacity to disburse and recover from individuals, the basis of the loan agreement could be changed with the society passing on funds collected from the farmers. 174. Two NDCA procedures seem to have aided this process: (a) A "minimum visit" program of supervision0 Credit supervisors visited each society during each of three critical phases of loan administration: loan application, when details of requirements were being collected; disburse- ment, when inputs were being distributed; and recovery, when repayments were being collected. Provision was made to record progress pro forma on each set of procedures. (b) Long-term agreements, over a nine-year period, for working capital funds. This had several advantages that would have been inhibited by a renewed application and appraisal each year: (i) it allowed NDCA to take the initiative each season, ensuring timely ordering and delivery of requirements; (ii) it gave continuity to 1he institution- building element in establishing procedures at the society for credit administration; (iii) it allowed a carryover of stocks on hand at the end of the season, whereas with a single-season agreement the cooperatives would have had to use their own funds to pay for unissued inputs. 17. Three further aspects of the IDA credit are noteworthy: (a) The prices of fertilizer, insecticide, and pumps were reduced by the contract procedures specified in the loan agreement between IDA and the Tanzanian government. Purchases were by direct import in considerable quantities. The savings achieved enabled the farmers' price for fertilizers to be maintained at the previously subsidized level without a subsidy being paid. Although there is controversy about subsidized inputs, there is certainly a case for subsidizing down to the price expected once demand has built up to the level allowing bulk purchases. 80 (b) obtaining finance at the concessional interest rate of 5-1/4 percent from institutions normally working in the 6 to 8-1/2 percent range had repercussions at all levels: (i) the NDCA came under local political pressure to lower its interest rates on loans funded from these other sources; and (ii) the coopera- tives, borrowing at 6 to 6-1/2 percent for crop finance, tended to use farmers' repayments, which should have been channelled back to NDOA, to offset their overdrafts, saving 3/h to 1-1/4 percent and, at the same time, reducing the financial turnover in NDCA's sis;ter organization for crop finance, the NOB. Similarly, cooperatives with outstanding loans from the NDOA.,. at the current local interest rate of 8 to 8-1/2 percent tended to repay these loans with IDA repayments, saving the difference in interest. (c) The loan agreement, signed by the Nyanza cooperative union on behalf of its affiliated societies, included an arrangement for the accuuldation of capital funds so that financing of inputs could be assumed by the cooperatives at the end of the nine-year- loan term. A percentage of each year's returns had to be paid into a loan account. Effectively, this raised the rate of interest and appeared to remove nuch of the advantage of low prices obtained by contracting. There seemed to be little logic in raising a fund for working capital requirements, and the arrangement was implemented grudgingly by te cooperatives. It created more problems than any other part of the long-term loan programs. This was particularly tre because of th difficulties in maintaining producer prices in 1966, the year the agreements were negotiated. Every marketing levy was being trimmed, and this commitment was seen by the unions as jeopardizing their own revenues. 176. IDA credit TA 80 provided a particularly timely injection of qualified manpower into the smallholder credit program. Although the 1964 act had reoriented loan policy, directing credit to farmers through the cooperative movement, up until 1966 (when TA 80 became effective) the NDCA had remained isolated from the cooperatives, its customers. The NDCA had no effective local representation and dealt solely with the cooperative unions, the secondary level of the movement. The agency concentrated on security and recovery; it had no interest in the effectiveness of its lending in terms of farmer benefits; and it had no local supervision to monitor the way in which the unions used the funds. TA 80 provided man- power at the center to evaluate lending in terms of the ultimate benefi- ciary; and it provided manpower in the regions to supervise disbursement and recovery through the movement. Although policy was refocused on cooperatives by the 1964 act, there had been no changes in procedures. The unions still used the same application forms as did the individual borrowers from the old ACA. The injection of manpower allowed the agency to adapt its procedures to cooperative lending: Procedures were tightened, 81 - and the agency was given the capacity to examine the outcome of loan programs for the farmer. This stiffened the agency's resistance to out- side political pressures to grant loans, especially those arising locally. Without this ability to examine the development potential of a loan program, there could be no real basis for resistan-le to a political proposal. With an agricultural and economic appraisal of the loan's impact on the farmer, res!7tance was justified, particularly since propo- sals are usually presented as benefiting the farmers in the areas concerned. 1 77. However, TA 80 was not without its bad aspects: (a) The lack of continuity in the Banks' supervisory personnel led to some contradictions, as well as to casiderable repetition by the lower-level staff. Several missions criticized the use of IDA-recruited staff for locally funded loan programs, but this seemed a logical development, when the building of an effective credit institution is kept in mind. (b) As already mentioned, the loan's objectives clashed: The pressure for rapid dis- bursement was not compatible with the aim of institution building. The program required that agreements be negotiated with different cooperative unions on many subprojects, Negotiations took as long as 15 months at the beginning of the loan period, and disbursements during this period were slow. The policy of disbursing through inefficient societies while building up their administration and safeguarding recovery by the estab- lished method of repayment was an attempt to satisfy both sides. Missions criticized disbursement of funds to cooperatives who did not yet have adequate loan procedures. Yet, had disbursement been postponed until sound accounting procedures were established, few categories would have been drawn on. (c) This compromise situation remained unsatisfactory for the first two years of disbursement and was never wholly adequate. Certainly the local staff situation was critical, but the credit was presumably appraised and approved with this in mind. The effectiveness of the loan, both as credit and as an institution-building exercise, - depended on the continuity of the staff, but it should have been more closely dependent on the prior nomination of acceptable counterpart personnel and their commitment to the program after a probationary period of not more than a year. 178. Despite these failings, the TRDB is in a significantly stronger position now than the NDOA was in 1966; it has a range of technically qualified personnel, both centrally and locally, and it has potentially effective procedures at both levels. Most of this improvement must be credited to IDA TA 80. I Secondary Benefits Arising from Cotton Expansion 179. Given further time, quantification of the secondary benefits from the multiplier effects cf the ginning and cottonseed oil industries, where the linkages are straightforward, will be possible; and these 82 benefits will be valid additions to those already accruing to the direct beneficiaries of the program. Usually such direct vertical linkages are part of the development prograim itself. With the local textile industry, the linkages with domestic production of cotton are less straightforward; at ihe extreme, it is clear that one cannot assume that secondary effects from spending increased incomes would not have occurred without the expansion of cotton. Other sources of income in 1966 provided 31 percent of the total, and livestock, in particular,had a large development pottial. It is valid to show the multiplier effects accruing to incomes of the project's direct beneficiaries. This, then, raises the question of how far the multiplier effects of expected expansion of the traditional system would correctly be a cost to set against the secondary benefits generated by the project. The rapid evolution in cotton production reveals the fallacy of using a static situation in traditional agriculture as a basis for estimating the incremental benefits of a project whose establishment phase will last probably ten years. It is fairly common to use figures on reduced production--the falling yield from neglected tea, for example--in estimating incremental benefits. An expansion on the order of that experienced in Sukumaland cotton is likely to make any other project appear highly inferior. The success of this expansion is a further indica- tion of the importance of knovirg which farming trends are occurring in an area before attempting to channel the area in a planned direction. 180. In a similar way, the additional government revenues that are generated by linkage effects increasing incomes for services and by secondary benefits, provide added justification for government investment in the project. The tendency to charge the direct beneficiaries for costs of services that are normally provided by the government and are considered as infrastructure ignores the fact that the programs undertaken should be the most effective way of deploying national resources. These c-sT- should be met by public revenues. Tax revenues from the direct beneficiaries can validly be supplemented by the tax revenues from secondary incomes to show the viability of a povernment contribution. J. Two Prerequisites for Bank-Lending Policies 181. (1) The existing farm system ast be evaluated, as a basis for development programs and to determine the diffusion potential of changes introduced. The history of the development of cotton production in Sukumaland demonstrates the potency of diffusion as well as the capacity of the traditional farming system to react to changes in the production environment. Both the spread of cotton production to most farms and the cultivation of a larger acreage of cotton on the individual farm contributed 'to the rapid expansion of production; and both stress the importance of diffusion potential in changes to be introduoed to traditional farmers. The history of government efforts to expand cotton production demonstrates the differetce in impact of changes with and without diffusion potential: 83 (a) Tsetse eradication and the development of water sources were aimed at alloviating a basic imbalance between population increase and land availability by changing the settlement pattern. They were successful because of a correct diagnosis of the existing situation and because of the availability of new land areas to settle. (b) Villagization was unmuccessful because of its failure to take into account the resource relationships in growing cotton. In addition, the structural reorganization involved isolated the changes introduced froa the production conditions of the community, and the strategy had no diffusion potential. Had it been ful1y implemenVad, villagization would have affected under I percent of the raral commnity, growing by about 2,, percent each year, (c) It was correct to diagnose intensification, the increasing ofthe productivity of the land, as a long-term needof Sukumaland but, for most areas at least, its introduction was and is premature. The use of fertilizer is not compatible with the economic situation that faces the farmer. Although it gave a fertilizer subsidy,, the government failed to alter conditions enough to create incentives for the farmer to adopt intensifying practices. It is important to stress that the (relatively) slow-changing relationship between the land and the rural population is what affects the evolution of the farming system. Exhortation and advise, which contradict these trends, must appear impertinent to farmers and can only decrease the standing, and eventually the morale, of the advisory services. More emphasis should be placed on monitoring the existng farming system and the local economic conditions affecting farmers' decisions as a basis for the selection of ap ro riate improvements. Gavernent intervention to "arrange" the producton conditions should be carried out with an eye to both long-term trends in the agricultural econoy and to diffusion potential of changes throughout the whole community. If the innovators are isolated fram the comrtmity by a structural reorganisation that only alters production conditions, diffusion is inhibited. The Bank has appeared to pay little attention to these points. In its early (1960) recommendations for villagization as a strategy for agricultural development, it ignored the lack of diffusion potential as well as the difficulties of replicability in terms of manpower and cost. In its appraisal of the cotton fertilizer program, although it had anticipated a long-term need of Sukuma farmers, it did not examine the resent farm-system relationships Nor did it evaluate how conditions in the local economy would have to be influenced to create a level of incentive for adoption competitive.with existing possibilities for income expansion. 84h 182. (2) Infrastructural costs must be recovered from direct beneficiaries. One reason for the lack of success of villagization was that settlers' incomes had to cover the overhead of the schemes, which included supervisiong water supplies, and access roads. If a program is so lucrative for the direct beneficiaries that they can repay infrastruc- tural costs without endangering their incentives to participate, then recovery is reasonable, but if such charges threaten to stifle sustained acceptance of the changes in the program, there is a good case for the government to carry the costs. Capital is no longer the key resource in development efforts in agriculture; qualified manpower is probably the scarcer, Recovery of capital and recurrent expenditure from the project is unrelated to the opportunity costs of either capital or manpower, but particularly of manpower. The decision to allocate resources to a program should be made on the basis of their opportunity costs. The government chooses to use its resources in this way because it is seen as the best alternative available for the economy as a whole. On these terms it is not the direct beneficiaries who are receiving special treatment, and if the government decides in principle to generate revenue through taxes for development purposes, the program represents use of that revenue, whether in the form of capital or manpower. Diffusion potential is also an important criterion here. When the program changes spread throughout the commnity, being appropriate b large numbers of farmers, the direct beneficiaries, the innovators are contributing a service and should not be penalized for it. When the program is based on structural reorganization, and diffusion of the improved farming practices is unlikely, the beneficiaries are then being given special treatment, and there is a stronger case for recovery of Infrastructural costs from them. In such cases the capital and manpower requirements are usually so high that the beneificaries amount to only a small proportion of the rural community. The program then becomes a questionable choice by the government because of the opportunity costs of the manpower required, a factor in no way offset by the recovery of the expenditure involved, There is a case for including government revenues that are operated by development programs as project benefits to offset the infrastructural costs. The case is strengthened when the diffusion potential of the changes involves multiplying such revenues without concomitant increases in costs. 183. The institution-building role of bank lending should dLinate program implementation and supervision. The weakness of the institutions with a central part to play in development efforts in Sukumaland, parti- cularly the cooperative movement and the agricultural extension services, emphasizes the need for less of a compromise of the institution-building element in bank credit programs. The counterpressures arising for disbursement of TA 80 and the buildup of control, accounting, and super- vision procedures tended to be resolved in favor of disbursement. BIBLIOGRAPHY Brown, K. J. "Rainfall, Tie Ridging and Crop aields." Epire Cotton Growing Review, January 1963. Bureau of Resource Assessment and Land Use Planning (BRALUP), University of Dar-es-Salamm I'Prelimina7 Report of the Skumaland Inter- disciplinary Research Project," Report No. h0, 1971. Collinson, M. P., Farm Manageient in Peasant Agriculture. New York: Praeger, 1972. "Farm Management Surveys 1-4:" "Bukumbi Chiefdom," 1961 "Usmao Chiefdom., Kwimba District," 1962; "Luguru Ginnery Zone, Maswa District," 1963; "Nyam:pande Co-operative Society Area, Geita District," 1964. SItTanzania's Co-operative Movement and Farmer Credit in the 1 896s" Mimeographed. 1970. Cooper Brothers. "Report on the Cotton Industry in Tanzania," Dar es Salaam: 1959. De Wilde, J. C., and Others. Experiences with Agricultural Development in Tropical Africa Vol. 1 : 'Yn es S,W II: Case Sesd ti, tm-ore: J-ohns Ho11pkins ni-vers, ity Poress,197 Hulls, R. "An Assessment of Agricultural Extension in Sukumaland, Western Tanzania," IERB Paper 71.7. Dar es Salaam: Economic Research Bureau, University of Dar-es-Salaam, 1971D International Bank for Reconstruction and Development, The Econory of Tanganyika, Bpltimore: Johns Hopkins University PFres,T6. Kirby, "Cotton in Tanzania." Empire Cotton Growing Review Vol. 7, No, 3, (1930). Kline, 0. K. Agricultural Mechanization in Iquatorial Africa. East Lansing: Michigan StaEU t Laurent, C. K. "Cotton Marketing in Tanzania2" Mimeographed, 1969. Ludwig, H. "Agricultural Practice on Ukara Island." In Smallholder Agriculture and Development in Tanzania, edited by H. RatEhenFurg. Volhof Africa t1ies SeriLes, Ranich: IFO, 1968. Malcolm, D. W. Sukumaland. London: Oxford University Press, 1953. Peat, J. E., and Brown, K. J. "Yield Responses of Rain-Grown Cotton in Lake Province, Tanganyika." Empire Journal of Experimental Agriculture 1962. -2- Porter, H. G. "Cotton in Tanzania," Washington, D.C.: USDA, FAS M219. June 1970. Pupius, J. "The Results of Block Mechanisation Schemes in Sukumaland," Mimeographed, 1965. Rotenhan, D. "Cotton in Sukumaland." In Smallholder Agriculture and Development in Tanzania, edited by RutFenYurg. Vol. 24. of =Arica StucHes Serles.7M-17.: IFO, 1968. Rounce, N. V. "The Economic Development of Sukwaland." Empire Journal of Experimental Agriculture, 1951. Saylor, G. "An Evaluation of Agricultural Extension in Sukwualand." Paper to the East African Conference of Agricultural Economists, 1970. Tanzania, Central Statistical Bureau. "Directory of Industries 1971.11 Dar-es-Salaam, 1972. . "Recorded Population Changes 1948-67." Dar-es-Salaam., 1968. . "Report on the 1966 Household Budget Survey among Cotton Growers in the Lake Region." Dar-es-Salaam, 1967. "Report on the 1966 Survey of Cotton Growers in the Lake Region." Dar-es-Salaam, 1968. . Survey of Industrial Production. Dar-es-Salaam, 1969. Tanzania, East African Statistical Department. "Sample Census of African Agriculture 1950.1 1953. Tanzania, Lint and Seed Marketing Board. Annual Reports. 1957-67.