“ .1 _ Ill . I": -‘ } af-‘uI—jfl' MSU LIBRARIES ‘ __-_ -.Unn-'l'bl swxm'm h“ J ' ' RETURNING MATERIALS: Place in book drop to remove this checkout from your record. FINES will be charged if book is returned after the date stamped below. ECONOMIC AND TECHNICAL ASPECTS OF SMALLHOLDER MILK PRODUCTION IN NORTHERN TANZANIA By Thomas M. Zalla A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements fbr the degree of DOCTOR OF PHILOSOPHY Department of Agricultural Economics 1982 COPYRIGHT BY THOMAS M. ZALLA I982 This stu: munt Kilimanja- tion systems in saallholder mill Data for 1 farm management the coffee-banar analyzed in the theory as well a nical data are s mags-meat pract Ina study accounted for a: i" 1973 even tit: am]: 3'5. Latte 265“ came an: l “483- RE'jUC; VEUer' 1”W an: ABSTRACT ECONOMIC AND TECHNICAL ASPECTS OF SMALLHOLDER MILK PRODUCTION IN NORTHERN TANZANIA By Thomas M. Zalla This study provides data on smallholder milk production on Mount Kilimanjaro. It analyzes alternative smallholder milk produc- tion systems in current use there and outlines a strategy for expanding smallholder milk production on the mountain. Data for the study were gathered in 1973 from a single visit farm management survey of over 680 randomly selected households in the coffee-banana zone of Mount Kilimanjaro. These are discussed and analyzed in the context of post-Hicksian hep-classical economic theory as well as the political and social context of Tanzania. Teche nical data are supplemented with information on production traits and management practices for cattle available from elsewhere in Africa. The study estimates that zebu cattle and grade cattle each accounted fOr about one-half of cattle milk production on the mountain in l973 even though only l2 percent of all cattle were grade dairy animals. Lactation milk production averages around 380 liters fior zebu cattle and 1,470 liters for grade cattle, net of milk suckled by calves. Reducing calf mortality and the calving interval by making veterinary and breeding services more easily available to farmers wear to be ti tial increases carg‘r‘al value grain produCts erginal factor suite significa extension servi the variation i The econo: systems shows ti ”it are conside the returns avai ash mm Sophis mt' The ijr 'm Dmduction firm‘0’"."9 Exte- WHES are als services and res. 811m . n . appear to be the only two factors that are likely to lead to substan- tial increases in milk production by zebu cows. Fbr grade cows the marginal value product of resources invested in providing water, salt, grain products and forage are all considerably in_excess of their marginal factor cost and the potential fer increasing milk production quite significant. Overall, variables over which a well-organized extension service can have some influence explain about 70 percent of the variation in milk production fbr grade cows. The economic analysis of returns to alternative milk production systems shows that upgrading zebu cattle yields returns to resources that are considerably above their opportunity cost and not far below the returns available from a pure grade dairy enterprise that requires much more sophisticated management and over twice the capital invest- ment. The major constraint on expansion of upgraded and grade cattle milk production is a dearth of good quality grade bulls and poorly functioning extension and artificial insemination programs. FOrage supplies are also a problem. Improvement of extension and veterinary services and research into forage crops that can be integrated into existing farming systems merit immediate attention. Educat‘ from teacners such a way t.’ e‘eys obvio'. few such pee; out in any we. bered but equ gratitude to leagues at tin insights and ; flinch, thug?»- lire-found in; ACKNOWLEDGMENTS Education involves the assimilation of ideas and observations from teachers, colleagues, family, students, scholars and others in such a way that their contribution to our knowledge and skills is not always obvious. I wish to publicly acknowledge the contribution of a few such people to the intellectual content of this dissertation with- out in any way diminishing the contribution of others less remem- bered but equally helpful. In particular I wish to express my gratitude to Mark Hurtz, Clive Thomas, Manuel Gottlieb and other col- leagues at the University of Dar es Salaam fOr their stimulating insights and patient explanations of dimensions of political economy which, though no longer so evident in the body of this paper, have had a profound impact on my understanding of economic and social dynamics. I also wish to express my appreciation to Lester Manderscheid in helping me to grapple with some of the methodological issues raised by this study and in methodically reading and commenting on the many drafts of various components that have risen and fallen over the past several years. Tjaart Schillhorn-van-Veen has played a similar and equally valuable role over a somewhat shorter period. Glenn Johnson was another persistent and loyal critic who persevered in his efforts to keep me from wandering too far from the fold. Though the intellectual contributions to a work such as this are usually the most obvious they are not always the most essential. ii Special gratituc'e frustrations, ac: min; and cocpl special thanks 2 Cynthia, Laura a fellow graduate and patient 53-... not have been p: Flfially I the Rockefeller My. the Uni national [9,910 Stimrt during take“ Place. Special gratitude is owed those close to us who share directly in the frustrations, accomplishments, joy, tensions and relief that under- taking and completing a dissertation involves. In this regard I owe special thanks to Mary Howard, our sons Matthew and Christopher Zalla, Cynthia, Laura and Mike Morton, Janet Munn, Pat Eisle and innumerable fellow graduate students. Carl Eicher has also been a steady, faithful and patient source of support and encouragement without which it would not have been possible to complete this work. Finally I would like to thank the FOreign Area Fellowship Program, the Rockefeller Fbundation, the Swedish International Development Agency, the University of Dar es Salaam, the U.S. Agency for Inter- national Development and Michigan State University for their financial support during portions of the lengthy period over which this work has taken place. LIST or F135?“ 15‘! TC EXCHANGE 34.391133 1_ HISTSR A. In' B. N .7997” 3"pr N—‘ TABLE OF CONTENTS PAGE LIST OF TABLES ........................ viii LIST OF FIGURES ........................ xiv KEY TO EXCHANGE RATE AND CURRENCY EXPRESSIONS ......... xv CHAPTER I. HISTORICAL FRAMEWORK ................. l A. Introduction and Purpose ............. l B. Geographical Description ............. 3 l. Tanzania ................... 3 2. Kilimanjaro .................. 5 C. The Colonial Heritage ............... 7 l. German Commercial Policy ........... 8 2 The Commercialization of African Agriculture ................ 9 3. Influence of Christian Missionaries ...... 12 4. Population Growth and Land Pressure ...... l4 5. Kilimanjaro at Independence .......... l7 D. Post Independence Political and Economic Development .................. 18 l. Prior to the Arusha Declaration ........ l8 2. The Arusha Declaration and Its Aftermath ................. 2l E. The Impact of Historical Fbrces on the Kilimanjaro Farming System and Diets ...... 23 F. Summary and Implications ............. 29 II. ECONOMIC PERSPECTIVES ................. 30 A. Pareto Optimal Market Economics .......... 30 8. Implications for the Analysis of Milk Production in Kilimanjaro ........... 36 III. KILIMANJARO AND ITS DAIRY INDUSTRY .......... 38 A. Demographic Characteristics ............ 38 8. Reflections of Health ............... 4l C. The Agricultural Economy ............. 41 D. Alternative Milk Production Systems ........ 46 l. All-Zebu Cattle Enterprises .......... 48 2. Grade Dairy Enterprises ............ 48 iv 1.14‘llulll‘! I III 1' I. .I ‘I It 4‘ up. ill al' .l ‘l ‘l r; -‘| F: I at ll.‘ 9 v 3. IN In 3... n A3 35 ..I\,. v; o 0 Fr. o a n3 rub . 0W. In .- c EFG MI ”an A B C Anon ABC Urnu A BC v. H'- I I VI CHAPTER IV. VI. VII. PAGE 3. Mixed Zebu-Grade Dairy Enterprises ...... 49 4. Goat Enterprise ............... 50 E. Herd Size .................... 53 F. Cattle Ownership Patterns ............ 56 G. Extension, Veterinary and Artificial Insemination Services ............. 59 l. Extension and Veterinary Services ...... 59 2. Artificial Insemination Service ....... 60 H. Credit ...................... 63 1. Summary ..................... 65 HERD STRUCTURE AND HERD DYNAMICS ........... 67 A. Herd Structure .................. 67 l. Zebu Cattle ................. 67 2. Grade Cattle ................. 73 8. Mortality Rates ................. 77 l. Zebu Cattle Mortality ............ 78 2. Grade Cattle Mortality ............ 83 C. Summary ..................... 85 AGGREGATE MILK PRODUCTION. MILK MARKETING AND RELATED PUBLIC HEALTH ISSUES ............. 86 A. Aggregate Milk Production ............ 86 8. Milk Marketing .................. 91 C. Milk Consumption ................. 95 l. Public Health Aspects of Milk Consumption . in Kilimanjaro .............. 97 MANAGEMENT PRACTICES FOR ZEBU AND GRADE DAIRY ENTERPRISES .................. l02 A. Breeding Practices ................ l02 l. Zebu Cattle ................. l02 2. Grade Cattle ................. l08 3. Summary of Reproduction Coefficients ..... ll2 B. Calf Rearing Practices .............. ll2 C. Feeding Practices ................ 115 l. Zebu Cattle ................. ll5 2. Grade Cattle ................. ll7 3. Feeding Practices for Cows .......... ll9 D. Milk Yields and Lactation Histories ....... IZl E Veterinary Practices ............... l24 VARIABLE AND CAPITAL INPUTS FOR ZEBU AND GRADE CATTLE ENTERPRISES ............... l28 A. Labor Inputs ................... l28 B. Other Variable Inputs .............. l36 C. Capital Investments by Enterprise Type ...... l38 A. The B, Ze:. C. Gra: D. Inf. i IX. ENTERFRI uhfi“lfi’l V-‘rn Iv-u-d.‘ Bsd. Sufi En: f Inc Ohm) E. Po‘. X. EXPARII on _ (I I _ I? F, 8:.) I (l‘ n¢ncn l‘: CW l CHAPTER I. Housing ................... 2. Cattle ................... 3. Other Cash Investment Costs ......... VIII. DETERMINANTS OF MILK YIELDS ............. The Model .................... Zebu Cows .................... Grade Cows ................... Implied Marginal Productivity of Selected Inputs ............... Summary of Determinants of Milk Yield ...... m DOW) IX. ENTERPRISE ACCOUNTS AND INCENTIVES FOR UPGRADING THE DAIRY ENTERPRISE ........... Budgeting Techniques .............. Summary of Procedures and Coefficients ..... Enterprise Types and Results .......... Incentive for Upgrading the Dairy Enterprise .................. Potential of the Various Alternatives for Increasing Milk Production .......... DOW) m X. EXPANDING SMALLHOLDER DAIRYING IN KILIMANJARO . . . . A. The Political Economy Context .......... 1. Returns to Milk Versus Coffee Production 2. Macro-Economic Effects of Increasing Milk Production ............. 3. The Local Context .............. The Technical Context .............. A Strategy fer Developing the Dairy Industry in Kilimanjaro ........... Further Research ................ Epi109ue .................... MU OW BIBLIOGRAPHY ......................... APPENDIX A: DATA COLLECTION AND ESTIMATING PROCEDURES . . . . Sampling Methodology .............. Data Collection ................. Data Quality .................. Sampling Bias .................. Statistical Inference .............. Computational Compromises ............ Conclusion ................... CD'HNUOW) vi PAGE I38 I45 I45 I48 I48 I52 I56 I67 I69 I72 I72 I74 183 187 I92 I94 I94 I94 I98 200 202 204 207 208 209 A-7 A-II A-I8 A-ZZ A-24 AFEEICZX D LiiiliCIX E. APE DIX F: APEZIIDIX G: Inn? "'USIXH' #3? er L:.lIb.x 1: team a lF-‘E‘czx K- PT‘" Harv NIL CHAPTER APPENDIX 8: APPENDIX APPENDIX APPENDIX APPENDIX APPENDIX APPENDIX APPENDIX APPENDIX APPENDIX C: D: ADJUSTMENTS TO REPORTED LACTATION MILK PRODUCTION ................. CATTLE UNITS AND ASSOCIATED LIVEHEIGHTS ..... DERIVATION OF UNIT LABOR INPUTS FOR VARIOUS SIZES OF ALL-ZEBU, ZEBU-GRADE AND ALL-GRADE DAIRY ENTERPRISES ......... CATTLE VALUES BY AGE, SEX AND BREED ....... CONSTRUCTION OF PROGRESSIVENESS AND HEALTH INDICES ............... Progressiveness Index .............. Health Index .................. l. Housing Index ................ 2. Household Possession Index ......... 3. Health Index ................ MEAN VALUES, STANDARD DEVIATIONS AND SUMS OF SQUARES FOR THE ZEBU AND GRADE CATTLE MILK PRODUCTION MODELS ............. DERIVATION OF THE OPPORTUNITY COST OF FAMILY LABOR IN KILIMANJARO ......... THE ECONOMIC VALUE OF MANURE .......... SAMPLE ENTERPRISE SIMULATION .......... ENTERPRISE BUDGET SUMMARIES ........... vii PAGE B-I D-I E-I 3.l 3.5 3.6 3.7 4.] ‘1 L3 44 Sales of Human Po: Kilimanga of Chagga Percenta; Agricul ta Proportio Agricul to Size of 1973 Size of as Inclu: Estimate Survey A, Seurce 0 by Type TABLE I.I l.2 3.1 LIST OF TABLES Sales of Crops in Tanganyika, I945, I952 and 1960 . . . Human Population in Mainland Tanzania and Present-Day Kilimanjaro District, l92l-l967, Including the Number of Chagga ....................... Percentage of Households in Kilimanjaro with Selected Agricultural Enterprises or Growing Specific Crops Proportion of Kilimanjaro Farmers Using Selected Agricultural Practices, l973 ............. Size of Goat Holdings in Kilimanjaro by District, I973 ......................... Size of Cattle Holdings in Kilimanjaro by District as Included in the Survey Area, 1973 ......... Estimated Cattle and Goat POpulation in Kilimanjaro Survey Area, l973 ................... Source of Cattle Holdings in Kilimanjaro in l973 by Type of Animal ................... Number of Artificial Inseminations in Kilimanjaro District, l964-l972 .................. Age and Sex Composition of the Zebu Cattle Herd in Kilimanjaro, l973 ................. Number of Zebu Cattle Transactions Made During the Twelve Months Preceding the Interview, Broken Down by Sexual Maturity .................. Age and Sex Composition of the Grade Cattle Herd in Kilimanjaro, l973 ................. Number of Grade Cattle Transactions Made During the Twelve Months Preceding the Interview, Broken Down by Sexual Maturity ................ viii PAGE IO IS 44 55 58 62 75 TABLE 4.5 4,5 6.l 6.2 6.3 6.4 5.5 5.6 ibrtality‘ Zebu C319 Portall‘d. Grade 6a.. comparison Prod-action tonsurptlc Estimated in Kilima' Underrepor Prices of 1963-73 Grams of Household Preparatii Incidence Consmpti in Kilima Size of I TYPE in r Refiroduc: Zebu and PErcenta~ ReceivirI 0f Fora; pmpm-ti I Highest l Receivir. of pom! l973 . ' AVE Fa Se Length . ‘5 Hell pe'CQHte Vactina .1 PAGE Mortality and Age Specific Mortality Rates for Zebu Cattle in Kilimanjaro, l973 ............ 80 Mortality and Age Specific Mortality Rates for Grade Cattle in Kilimanjaro, l973 ........... 84 Comparison of Two Estimates of Annual Milk Production in Kilimanjaro with Reported Consumption, l973 ................... 87 Estimated Fluid Milk Production and Consumption in Kilimanjaro in 1973 by Source, Adjusted for Underreporting ..................... 90 Prices of Milk and Coffee in Kilimanjaro, 1963-73 ........................ 92 Grams of Protein Obtained from Milk by an Average Household in Kilimanjaro by Type of Product and Preparation, l973 ................... 96 Incidence of Digestive Problems Related to Consumption of Milk and Other Common Foods in Kilimanjaro ..................... lOO Size of Individual Cattle Herds by Enterprise Type in Kilimanjaro, l973 ............... 103 Reproduction of Coefficients fer Average and Superior Zebu and Grade Cows in Kilimanjaro, l973 ........ ll3 Percentage of Zebu and Grade Cattle in Kilimanjaro Receiving Hater, Salt and Grain and the Percentage of Forage Coming from Selected Sources, 1973 ...... llB Proportion of Average and the Twenty-Five Percent Highest Yielding Zebu and Grade Cows in Kilimanjaro Receiving Hater, Salt and Grain and the Proportion of Forage Coming from Selected Sources fer Each, l973 .......................... l20 Average Lactation Milk Production and Lactation Length fer All-Zebu and Grade Cows in Kilimanjaro as Hell as For the Top Producing Twenty-Five Percent of Each Type .................. l25 Percentage of Zebu and Grade Cattle in Kilimanjaro Vaccinated and Treated with Acaricide, l973 ...... l27 ix r .n TAaJ ll 7] LI 7A 75 L5 L7 LB 73 II 3.2 83 Iota1 an Cattle E Kilimafi.‘ Total an Zebu-5Fa in Killm Total an- Cattle E Kilimanji chrs of for All-. of Diffe' Labor SC} The Ave re for All-L Enterpri: Expendi t: Location by Cattle bxof: in Kills; Value of of Housi Value of Of Enter Average CaItIe a AlLGra: Sum-Wary Variatl: ln KlIi~ Summary Val‘latj1 I" Kili- IDCreaS . ”DBri c ”I the the As: Of the meUC: TABLE 7.I Tbtal and Cattle Unit Labor Inputs for All-Zebu Cattle Enterprises of Various Sizes in Kilimanjaro, l973 ................... Tbtal and Cattle Unit Labor Inputs for Mixed Zebu-Grade Cattle Enterprises of Various Sizes in Kilimanjaro, l973 ................. Tbtal and Cattle Unit Labor Inputs fer All-Grade Cattle Enterprises of Various Sizes in Kilimanjaro, l973 ................... Hours of Labor Per Heek Required Per Cattle Unit fbr All-Zebu, Zebu-Grade and All-Grade Cattle Herds of Different Sizes, Including the Proportion of Labor Supplied by Homen ................ The Average Value of Annual Variable Cash Inputs fer All-Zebu,Mixed Zebu-Grade and All-Grade Dairy Enterprises in Kilimanjaro; Tbtal and Cattle Unit Expenditures, 1973 .................. Location of Cattle Housing in Kilimanjaro by Cattle Type .................... Type of Structure Used as Housing fbr Cattle in Kilimanjaro by Type of Cattle ........... Value of Cattle Housing in Kilimanjaro by Type of Housing and Type of Enterprise ........... Value of Cattle Housing Per Cattle Unit by Type of Enterprise and Size of Herd ............ Average Tbtal and Cattle Unit Investments in Cattle and Other Cash Items fbr Zebu, Mixed and All-Grade Dairy Enterprises in Kilimanjaro, 1973 Summary of Regression Coefficients Explaining Variation in Milk Production Among Zebu Cows in Kilimanjaro .................... Summary of Regression Coefficients Explaining Variation in Milk Production Among Grade Cows in Kilimanjaro .................... Increase in Milk Production Between Average and Superior Grade Cows Explained by Technical Variables in the Milk Production Model for Moshi District with the Associated Cost of the Added Inputs, the Value of the Added Output and the Implied Marginal Productivities Per Unit of Expenditure ........ X PAGE I29 I31 I35 I42 I46 9! A4 A-Z A3 A4 34 A2 coefilcll Summary ' Value 0; Month To and Ail-- Alternat Sumter! I to Esta: Enterpri Proporti Arising Two Heas of Cattl Hale for Types Ur salysls Recordin Analysis Grade ca Comparis for Elev Random 5 Seaple Ccmparis ESIITate PCDU‘iati Comparis Hoducti by Age ( AVEra 35. Dry Ha: ASS LATE: CoeffiC Pmduc: Labor 1 and Ag; °f Cat: TABLE 9.I 9.2 9.3 9.4 A-4 B-I C-I C-2 Coefficients Assumed fbr the Enterprise Budgets . . . . Summary of the Cost of Variable Inputs and the Value of Variable Outputs Per Cattle Unit Per Month fer Various Sizes of All-Zebu, Zebu-Grade and All-Grade Cattle Herds in Kilimanjaro Under Alternative Levels of Management Summary of Initial Capital Investment Required to Establish One Cow Zebu and Grade Cattle Enterprises in Kilimanjaro Proportion of Undiscounted Simulated Returns Arising from Milk, Manure and Meat Production, Two Measures of Investment North and the Proportion of Cattle Units in Ongoing Operations Hhich Are Male fer Eight Simulated Cattle Enterprise Size/ Types Under Two Different Operating Assumptions . . . . Analysis of the Variance in Enumerator Recording Patterns Analysis of the Means and Variances of Separate Grade Cattle Sample Populations ........... Comparison of the Means and Their Standard Errors for Eleven Cattle Variables as Estimated From Random Sample A and A Composite Grade Cattle Sample Comparison of Heighted Random and Two Stage Estimates of the Standard Error of Selected Population Estimators ................ Comparison of Various Estimates of Annual Milk Production from Cattle in Kilimanjaro in l973 . . . . Average Liveweight of Zebu Cattle in East Africa by Age Group Average Liveweights, Approximate Consumption of Dry Matter and Adult Zebu Equivalent Cattle Units Assumed fer Zebu and Grade Cattle in Kilimanjaro Coefficients Used in Calculating Maintenance and Production Period Labor fer Enterprise Budgets Labor Inputs Per Cattle Unit for Carrying Hater and Applying Acaricide fer Various Types and Sizes of Cattle Enterprises in Kilimanjaro xi PAGE I75 I79 I88 A-l 4 A-I7 A-23 3-4 '33. 8-2 as J-l 5-2 M 3-4 H K.z 1973 Cat! Sex and 3 Per-38.9149 ProgrESSl specifies Mean Yalt Explainir the Firsfi Mean ValL Explainir the Firs‘ Cars by | Sum-ary E Productic Smary 1 Zebu-Era: Presence Simulate Cattle E Nature H Female Derivati for a l. "lib the Simulate Cattle E i‘ature y DeTIT/at" for a 1 VIth LL; Te“ Yea AII-Ze: Initiaj . Ten Yec AH'ZE"; One Hei YEar F: TABLE PAGE E-l l973 Cattle Values in Kilimanjaro by Age, Sex and Breed ..................... E-2 F-l Percentage of Random Sample Population Having A Progressiveness or Health Index of the Value Specified ....................... F-4 6-] Mean Values and Standard Deviations for Variables Explaining Variation in Milk Production During the First Six Months of Lactation For Zebu Cows . . . . G-2 G-2 Mean Values and Standard Deviations for Variables Explaining Variation in Milk Production During the First Six Months of Lactation fer Grade Cows by District ................... G-3 6-3 Summary Statistics fer Zebu and Grade Cow Milk Production Models ................... 6-4 0-1 Summary Sheet for a Simulated, l.6 Unit, Mixed Zebu-Grade Cattle Enterprise with Required Presence of a Mature Male in the Herd ......... J-3 J-2 Simulated Herd fer a 1.6 Unit Mixed Zebu-Grade Cattle Enterprise with Required Presence of a Mature Male in the Herd and First Calf is Female ........................ J-S J-3 Derivation of the Salvage Value of Dead Animals fer a l.6 Unit Mixed Zebu-Grade Cattle Enterprise with the First Calf a Female ............. J-B J-4 Simulated Herd for a 1.6 Unit Mixed Zebu-Grade Cattle Enterprise with Required Presence of a Mature Male in the Herd and First Calf is a Male . . . J-lO J-5 Derivation of the Salvage Value of Dead Animals for a l.6 Unit Mixed Zebu-Grade Cattle Enterprise with the First Calf a Female ............. J-l3 K-l Ten Year Capital Budget fer a Simulated, l.6 Unit, All-Zebu Cattle Enterprise with Purchase of Initial Heifer Only .................. K-2 K-2 Ten Year Capital Budget for a Simulated, l.6 Unit, All-Zebu Cattle Enterprise with Purchase of One Heifer in Year Zero and Another Heifer in Year Four Hhen First Calf is a Male .......... K-3 xii L4 {-5 L6 K-l Ten Yea cattle Hith Pr Ten Yea Cattle Vith Pr Ten Yea ttle No Hale Ten Yea Cattle ND Hale l‘ana gen Ten Yea ttle Require Ten Yea Cattle No Male TABLE K-3 K-4 K-5 K-6 K-7 K-8 Ten Year Capital Budget for a Simulated, l.6 Cattle Units, Mixed Zebu-Grade Cattle Enterprise Hith Presence of Mature Male in Herd Not Required . . . Ten Year Capital Budget for a Simulated, l.6 Cattle Units, Mixed Zebu-Grade Cattle Enterprise Hith Presence of Mature Male in Herd Not Required . . . Ten Year Capital Budget for a Simulated, l.6 Cattle Units, All-Grade Cattle Enterprise Hith No Male Required in Herd .............. Ten Year Capital Budget fer a Simulated, l.6 Cattle Units, All-Grade Cattle Enterprise Hith No Male Required in Herd and Hith Superior Management ..................... Ten Year Capital Budget for a Simulated, 2.2 Cattle Units, All-Grade Cattle Enterprise Hith Required Presence of Mature Male in Herd ...... Ten Year Capital Budget for a Simulated, 2.2 Cattle Units, All-Grade Cattle Enterprise Hith No Male Required in Herd .............. xiii PAGE K-4 K-G K-7 K-8 L2 6.3 (h 2;- 65 Number 0 Kilimanj Number 0 Interval Parturit Number 0 Parturit Lactatio Kilimanj Lactatio Kilimanj FED of S Average Producti of Appli EXperime FIGURE 6.I 6.2 6.3 6.4 6.5 A-I I-l LIST OF FIGURES Number of Births to Zebu and Grade Cows in Kilimanjaro by Month of the Year, l972-l973 . . . . Number of Births Occurring at Specific Intervals Between the Last and the Previous Parturition fer Zebu Cows in Kilimanjaro, l973 Number of Months Between the Last and Previous Parturition for Grade Cows in Kilimanjaro, l973 . . . . Lactation Milk Production for Zebu Cows in Kilimanjaro, l973 ................. Lactation Milk Production fer Grade Cows in Kilimanjaro, l973 ................. Map of Survey Area ................ Average and Marginal Increases in Cereal Production per Tbn of Manure at Various Rates of Application as Indicated by Several Experiments in Michigan and Africa ........ xiv PAGE I05 I07 III I22 I23 jg: : One Tanzania 73:: = Seven srm KEY TO EXCHANGE RATE AND CURRENCY EXPRESSIONS l/= 8 One Tanzanian Shilling = lOO Tanzanian Cents = $.l4 U.S. 7/l4 = Seven Shillings And FOurteen Cents = $l.OO U.S. XV This s smallholder as it existe the grade ca it percent 0 research was deity incur: the ISEQS. seen as the ate an inc re CHAPTER I HISTORICAL FRAMEWORK A. Introduction and Purpose This study investigates political and economic aspects of the smallholder dairy industry on Mount Kilimanjaro in Northern Tanzania as it existed in 1973-74. The mountain contains about 60 percent of I on small holdings in Tanzania and produces about the grade cattle 30 percent of the country's coffee-—its principal export in 1973. The research was stimulated by government concern over the rapidly rising dairy imports and declining coffee prices which were occurring during the 19605. Expanded smallholder milk production in Kilimanjaro was seen as the antidote fer both these problems while helping to allevi- ate an increasingly visible young child nutrition problem in the areas. Specifically, the study has three objectives: 1) to provide farm management data on smallholder milk producers in Kilimanjaro fer planning purposes; 2) to analyze alternative smallholder milk produc- tion and 3) to outline a strategy fer expanding smallholder milk pro- duction in Kilimanjaro should policy-makers decide that it is desirable in light of the analytical results obtained. 1Grade cattle in East Africa refer to dairy animals with varying amounts of exotic genes. Fer the purposes of this study a grade dairy animal is defined as one with sufficient exotic genes to have lost the hump typical of the Tanzania Short Horn Zebu, the dominant type in Tanzania. The hump generally disappears when zebu cattle are cross- bred to European breeds of cattle. l Input-output visit farm manage: Juana zone of the and weaknesses of attention to him esaecially as the agricultural time This chapte- I-mt of the Sr W“ 5099 pers: HINTS the co“: lIl outlines imp; the Survey area ( (“WT—jam. I‘. IS Ell GS CUFre 355m Ty detai Ms as "fill at based on data SE e155 ag Sre 96 te | sects 0f arm has 2 Input-output data fbr the study were gathered from a single visit farm management survey of over 680 households in the coffee- banana zone of the mountain. The study also reflects on some strengths and weaknesses of post-Hicksian modern market economics and gives some attention to historical and contemporary political economy issues, especially as these relate to defining a viable strategy fer promoting agricultural development in Kilimanjaro.1 This chapter outlines the geographical, historical and political context of the smallholder dairy industry in Kilimanjaro. Chapter II offers some perspectives on post-Hicksian neo-classical theory and delimits the context of the analysis of enterprise returns. Chapter III outlines important social, economic and political dimensions of the survey area and begins the discussion of the dairy industry in Kilimanjaro. It also outlines the various production systems in use as well as current government assistance provided to the industry. Chapter IV details the structure of both the zebu and grade cattle herds as well as offtake, mortality and other dynamic aspects of each, based on data gathered in the farm management survey. Chapter V esti- mates aggregate milk production, presents a brief summary of milk marketing on the mountain, and discusses some public health aspects of encouraging interfarm sales of unpasteurized milk. In Chapter VI management practices for each of the herds are outlined 1Almost the entire mountain and surrounding lowlands lie in what, until 1973, was Kilimanjaro District. In 1973 Kilimanjaro District was split into two new districts, Moshi and Rombo Districts. It is this area we refer to when speaking of Kilimanjaro. It is to be distinguished from Kilimanjaro Region which includes, in addition to Moshi and Rombo Districts, a third district, Pare District. Pare District covers the Pare Mountains to the southeast of Mount Kiliman- jaro. CM ”1 I In. T? bet-teen received British F cal amid 3 and lactation milk yields are estimated. Chapter VII looks at labor and other inputs used in milk production. In Chapter VIII the determinants of milk production fer both zebu and grade cows are analyzed using ordinary least squares multiple regression. Then, in Chapter IX eight separate enterprise size/type combinations are simulated and capital budgets measuring their economic profitability are analyzed. The potential of each of these systems fur increasing milk production is also discussed. Finally the last chapter, Chapter X, summarizes the findings of the study, outlines interventions fbr upgrading the smallholder dairy industry in Kilimanjaro and specu- lates on the impact of an upgraded dairy industry on coffee production. It raises some unresolved policy issues and concludes by suggesting areas needing further research. 8. Geographical Description 1. Tanzania The United Republic of Tanzania is a loose union formed in 1964 between Tanganyika, a farmer British East Africa Trust Territory which received independence in 1961, and the Island of Zanzibar, a farmer British Protectorate which received independence in 1963. The politi- cal evolution of Tanzania is very much dominated by its charismatic president, Julius Nyerere. A one party state, Tanzania nonetheless experiences some sharp and remarkably candid debate, centered in TANU »(Tanganyika African National Union), the only political party on the mainland. Tanzania has an area of 363,000 square miles, about the size of Texas and Oklahoma combined. The country is bounded on both north and south by want and 501m of IE Africa. Its v luv movie-ti" ficiently varl a base for a c Tanzania an aggregate g of this total pozulation is the largest 01 leinland papal tion is non-A‘ Like mos iural with 93 ”'5 90 Percen' m" °CCUDatii e"'te+"ill’l'.ses b. File (445.), ‘ “55m (285} 4 south by mountains and highlands which have a rather temperate climate and some of the most fertile soils and highest population densities in Africa. Its vast interior is largely dry to semi-arid bush with very low population density. Overall, the country's land resources are suf- ficiently varied in soil type, topography and climate so as to provide a base for a diversified and productive agriculture. Tanzania's 1973 population was estimated to be 14.4 million with an aggregate growth rate of about 2.7 percent per year. About 410,000 of this total are located in Zanzibar [Egero and Henin, 1973]. The population is composed of more than 130 ethnic groups of African origin, the largest of which, the Sukuma, comprise about 13 percent of total mainland population. Only slightly more than 1 percent of the popula- tion is non-African [Lucas and Philippson, 1973]. Like most African countries, Tanzania is predominantly agricul- tural with 93 percent of the mainland population living in rural areas and 90 percent of the economically active segment describing their main occupation as agriculture [Tanzania, 1971b]. Major agricultural enterprises by order of the jarm_value of output are sugar (450), maize (440), cattle (385), bananas (385), coffee (355), cotton (290), cassava (285), millet and sorghum (I75), sisal (140), cashew (130), paddy (110) and beans (100).1 Major agricultural exports in l973, 1Figures in parentheses are the farm value of output in millions of shillings. Quantities are 1972 harvested production for maize, bananas, cassava, millet and sorghum, paddy and beans; l973 marketed production for sugar, coffee, cotton, sisal and cashew; and 1973 total offtake for cattle. The value fer cattle does not include the farm value of milk, hides or manure. Producer prices are those prevailing in the 1974-75 crop year except fer beans, and sugar. Fbr the latter two crops the 1972 price was used. Prices and quantities were taken from International Bank fer Reconstruction and Development [1974]. Additional prices imputed by the author were 320 shillings per ton fer gin; F05 Valh '222) and “5" wrdi72) iS 1 sailings or a Thirty-nine PE the subsistem cultural subs‘ growth rate it 4.4 percent fc EDP growth ra‘ adeinistratio: ay1973. Aft. duuous proiu+ consumption a' tion over thi The act P 4' bane“as and 4.9 Orders 0 l . 1‘.“ Fl guy-e :' DEUCESSed .Tanste ‘ . rs [: La : I97 “3"“ far ’3 n Her L‘Q 5 using FOB values, were coffee (495), cotton (333), cloves (233), sisal (222) and cashews (141). Diamond (170) exports are important but sisal cord (72) is the only manufactured export of any significance. In the aggregate, raw and semi-processed agricultural products accounted far 80 percent of total export earnings in 1973.1 1973 GDP fer mainland Tanzania was estimated to be 11,257 million shillings or about $112 per capita at exchange rates then prevailing. Thirty-nine percent of this originated in agriculture and 21 percent in the subsistence sector which includes both agricultural and non-agri- cultural subsistence activities. Over the 1968-73 period the real growth rate in agriculture averaged about 2.4 percent as compared with 4.4 percent fer GDP and 2.7 percent for population. Noteworthy in the GDP growth rate, however, is the 8.6 percent growth rate in public administration and other services accounting far over 9 percent of GDP by 1973. After allowing fer growth in these and other services of dubious productive value, it appears that overall goods available for consumption and investment barely kept up with the growth in popula- tion over this period.2 2. Kilimanjaro The actual area covered by the farm survey includes only the coffee-banana belt of Mount Kilimanjaro. This area lies between 3000 bananas and 350 shillings per head for livestock. These figures indi- cate orders of magnitude only and do not include Zanzibar. 1Figures fer cashews do not include 32.7 million shillings worth of processed cashew products. All figures include intra-community transfers [East African Customs and Excise Department, 1973]. 21973 national accounts and growth rates taken from International Bank fer Reconstruction and Development [1974]. and 5530 feet gistrict and tion of the 5 36,580 house? The big dominate the fall has a b“ slopes. It ‘1 ishes therea‘ ICUT‘JITI rant 3230 feet 19' Iiies up the mounts fall between 56.3.; eastern and ‘ IS dry and S "'9’“ pet: 10"” temper Ethnic Survey arise 507:! 400 C15 333490 year tent anl'ra‘al Ilmst entir- ihe banana 3 1\ a.“ .Stai‘, fg-ul‘ ‘5 the FE; 6 and 6000 feet but includes about 95 percent of all households in Rombo District and 77 percent of those in Moshi District. The 1973 popula- tion of the survey area is estimated to have been 475,000 persons in 86,500 households. The high volcanic peaks of Mount Kilimanjaro (19,390 feet) dominate the climate, topography and soils of the two districts. Rain- fall has a bimodal distribution and is concentrated on the southern slopes. It increases with altitude up to about 10,000 feet and dimin- ishes thereafter. Mean annual rainfall on the southern side of the mountain ranges from 34 inches in Moshi Town, which lies at about the 3000 feet level, to over 90 inches at Kibosho Mission (5000 feet), six miles up the mountain from Moshi Town [Brevin, 1965]. Still higher amounts fall in the uninhabited rain fbrests which circle the mountain between 6000 and 10,000 feet. Rainfall diminishes moving along the eastern and western slopes in a northern direction and the north slope is dry and sparsely populated. Mean annual temperatures fellow a reverse pattern with a mean value of about 74° in Moshi Town and much lower temperatures on the upper slopes [Tanzania, 1972]. Ethnically practically all of the population included in the survey area are Chagga (99 percent), historically an amalgamation of some 400 clans which have come to inhabit the mountain over the past 300-400 years [Marea11e, 1952]. The Chagga are noted fbr their excel- lent animal husbandry and agricultural practices. Their cattle are almost entirely stall fed.I and the accumulated manure is valued for the banana groves surrounding the homestead-bananas being a staple of 1Stall feeding is a zero grazing system. All the forage and water is brought to the animals, usually by women, often carried as far as three miles. 55.21135”;i the 51995 were be‘ available years to ' impressive the Che 953 Turn the politi of particu- the comer. Christian 1 The first 1 social and and betweer t3‘37 Drovid 3"- Part a r these force “hey ha ‘76 II 7 the Chagga. Banana by-products, in turn provide an important source of energy, roughage and water for their cattle. The Chagga irrigate their farms by means of an intricate and sophisticated network of gravity fed irrigation furrows which rise up the sides of steep river valleys and crisscross the entire coffee— banana belt. At the present time this system has reached the limit of available dry season water supplies and little has been done in recent years to harness greater volumes of water. Nevertheless it remains an impressive tribute to the organizational and technical ingenuity of the Chagga. C. The Colonial Heritage, Turning toward historical forces which have left their mark on the political economy of Kilimanjaro and its dairy industry, feur are of particular interest to the present study: German commercial policy, the commercialization of African agriculture, the influence of Christian missionaries, and the growth in population and land pressure. The first three have contributed in a fundamental way to the extreme social and economic differentiation which exists within Kilimanjaro, and between Kilimanjaro and most other areas of the country. As such they provide important background to Tanzania's socialist objectives- in part a reaction to the differentiation and unequal opportunity which these fbrces created. Together with population growth on the mountain they have led until recently to a secular decline in the dairy industry in Kilimanjaro-—per capita milk production in the area today probably amounting to no more than 25 percent of what it was 50 years ago. Clearly, any effort to reverse this trend requires an understanding of these forces and how they developed. The | Africa '35 Kiliranja’t of the box! close off I the vest. lower slo:£ Calvert, l3 Serra economy of orientation cess of ca; rth agricj The 9r aim , JPbEd thlf ”Eli-ates! i 8 l. German Commercial Policy The main thrust of German colonial and commercial policy in East Africa was the development of plantation agriculture. Early in the German period a large concentration of settler agriculture arose in Kilimanjaro. A few coffee estates were established on the upper slopes of the mountain in areas inhabited by Africans, but never so many as to close off African expansion in the same way as it did on Mount Meru to the west. The majority of settlers in Kilimanjaro located on the drier, lower slopes where sisal, grains and legumes did well [Iliffe, l97l; Calvert, l970]. German commercial policy had two lasting effects on the political economy of Kilimanjaro: it introduced and firmly established an export orientation and through this, was a factor in setting in motion a pro- cess of capital accumulation in an area already very favorably endowed with agricultural resources. The Tanga-Moshi rail link was completed in l9l2 opening the area to world markets and establishing the metro— politan-periphery communication links which have dominated Tanzania's economy up to this day. By the beginning of the First World War the plantation sector had gained a powerful position in the Tanganyikan economy as it supplied the needs of Europe with the resources of Tanganyika.1 The emerging African commercial agricultural sector quickly adopted this export orientation, guided by effective demand which originated in Europe. Hut and poll taxes, initially imposed to fbrce Africans to provide wage labor fOr the estates, exposed African 1For an excellent account of the dynamics of this process see Iliffe [l97l]. farmers cregsa f‘ EurOPean’ Stagga f3 Nanters Cooperati relative culture a sector in hl'narjar 9 farmers to cultivation of exotic export crops. By the late l920's Chagga farmers no longer were willing to work on the estates of Europeans and began to produce coffee on their own farms. In l925 Chagga farmers formed the first Association of African Coffee Planters which grew into the powerful present-day Kilimanjaro Native Cooperative Union [KNCU, l963]. Kilimanjaro thus got an early start relative to the rest of Tanzania in the development of commercial agri- culture and the accumulation of capital made possible by it. 2. The Commercialization of African Agriculture By the late 1920's African cash crop production in Tanzania had become firmly established and quickly grew to rival the plantation sector in economic power and significance. Coffee production in Kilimanjaro continued to expand, even during the depression, as did cotton production on small holdings in Sukumaland [IBRD, l96l]. The major expansion in African commercial agriculture in Tanganyika occurred in the period l945-60 when the gross value of agricultural exports increased 6-l/2 times or more than l3 percent per year compounded [Fuggles-Couchman, l964]. Coffee production on small holdings in Kilimanjaro did not increase so rapidly, doubling roughly every l0 years after l935, but the advance was, nonetheless, impres- sive. Maize production in the Northern Province also increased rapidly as household consumption skyrocketed and marketed production 1 Table l.l outlines increased 3-l/2 fbld between l945-47 and l965-67. the growth in marketed production of these and other crops for Tanganyika as a whole during this period. 1FOr the l945-47 production see Tanganyika Territory Department of Agriculture Annual Report, l952. The l965-67 figures are taken from Tanzania [l974]. C'oc / 35% Pets ‘ ilssm jar-0t (”"15 ,,. fgpxr ll. f in: _ Pu 730‘! . hat -;:“ee rc‘m} It}: Estate “'11“ hard ' {was frat" Essential Oils ‘fc U30! B‘seec-s m a" Casts-r seed Lac-re Evita-seed Scams 01‘. pa'l Gil lane‘s Seas-e see: Smile-er «1 has him 'exmrzs‘ 533M ism-er "a'lti 251;»;- “IN' (exports; a!“ lens .93“ a9. a.“ "de9 FEE-tum: _ F' (39¢ hum .Jflfip‘ .“ a - .7- if 0 -iies 3‘ .t “it; t. 10 TABLE l.l (LONG TONS) SALES 0F CROPS IN TANGANYIKA, l945, l952 AND I960 Crop l945 I952 I960 Cashew nuts 2.669 10.809 55,252 Cassava (maniac) 4,0803 29,277 29,659 Cereals Finger millet (eleusine) ... ... 12,800 Maize 20,450 42,3l6 77,946 Rice (paddy) 3,739c 15.639 34,859b'd f Hheat 3,975e 5,264 ll,66O +Coffee (clean) Mild Estate 2.290 7.359 7.170‘I African 3.462 H.208" Hard 8.2l3 9,l22 9.746 tCotton (mm 7.512 14.109 34.2419 Essential oils (exports) 10 3 ‘ 3 Kapok l40 443 986 Oilseeds and oil Castor seed 428 7,239 l0,925 Capra l0,000h H.893h 9.544 Cottonseed 7.600 15,000 52,817 Groundnuts 4.756‘ l0,882 22.352 Oil palm Oil 65 35 95 Kernels 42 334 76l Sesame seed 3.810 1,899 9,482 Sunflower seed 2.2881 12.292 7.314 Onions 2,654 2,160 8,058 Papain (exports) lOl 43 72 Potatoes (European) 2.000 l.880 3.000h Pulses 4,480 20,75l 24,443 tPYrethrum 799 240 l,OlO Rubber (exports) 2,320 2 ... tSeed beans 490 l.65l 5,620 tSisal llZ,2l8 l62,l85 204.868 tSugaP 7,300 9,666 28,624 flea 580 l.ll7 3,722 tTobacco Flue-Cured S95 l,074 l,80l Fire-cured 248h 597 498h Vegetables 2,000 .. 3,000 Agricultural Change in Tanganyika: 1945-1960: SOURCE: N.R. Fuggles-Couchman. 1964. Stanford: Food Research Institute. p. 40. °Data from Tanganyika, Dept. Agr., Annual Reoorts. 1945-60. +Sa1es of these crops may be considered as total production. '1946. bCrop in western Region was five times that of 1945. 6Above average season. 'Eacludes 6,837 tons produced by the Government Har wheat Scheme. fConverted to long tons. assuming bales weign approximately 400 pounds. 9The 1959 crop was 36,579 tons. hEstimated. 31948. the first year of production. ‘Drought in the Lake Region. This rapid farmers m to ) different areas timing to “PC". ecologically 00?" lest Lake and hue auction. The nor transportation of ccwolation of n ticnal, locally i effects eventual? local consumers. established an e: ever increasing logical advantag Cash crop: tion within the live for accum- voder cultiva‘. severe tonfl it of access. Th mml'dro, or 7!. provides ev It was to: 1‘ . ll This rapid expansion of commercial agriculture among African farmers led to increasing economic differentiation between and within different areas of the country. Nhile development in those areas con- tinuing to export labor to the plantation sector was stagnating, areas ecologically more favorable for agricultural production-—Kilimanjaro, Nest Lake and Mwanza-—became more actively engaged in cash crop pro- duction. The world market, which received most of their crops via the transportation network established fbr the plantations, facilitated .accumulation of wealth and capital in a way not possible in a tradi- tional, locally interdependent, essentially closed system, where price effects eventually shift the gains of increasing relative output to local consumers. The first areas to concentrate on cash crops thereby established an early lead over other less favored areas and added an ever increasing capital accumulation advantage to their existing eco- logical advantage. Cash cropping led to increasing social and economic differentia- tion within the major producing areas themselves as well. The incen- tive fbr accumulation which it offered stimulated an expansion of land under cultivation which led, in turn, to favoritism, increasingly severe conflicts over land, and abandonment of traditional guarantees of access. The large number of court cases involving land disputes in Kilimanjaro, one of the most highly differentiated areas of the coun- try, provides evidence of this.1 It was the larger, more commercial farmers who led the rapidly growing cooperative movement during this period (1945-60). Frequently 1For an analysis of court cases including land disputes see Mar'o [l974]. ge coax“: 3;portuniti€ of the less menent in ers v'tich, a services wit-1' Early the more ter and Kilic‘en“ :inct sphere The C? leans of in; 519” enrolls First Horlc‘. ll: mlSSlon 3 that high 1“ its lead 0w Early Stage lilSSir ,. '1!!! lei S 12 the cooperatives became tools of these elite farmers and provided opportunities fbr further accumulation of capital, often at the expense of the less educated, less knowledgeable membership. The cooperative movement in Kilimanjaro to this day is dominated by this class of farm- ers which, as we shall see later, has been able to initiate cooperative services which benefit themselves at the expense of other members. 3. Influence of Christian Missionaries Early mission activity in Tanzania tended to be concentrated in the more temperate, agriculturally more favorable areas of the country and Kilimanjaro was not overlooked. Mission activity there began in 1884. By 1910 missions had carved up the mountain into separate dis- tinct spheres of religious influence. The Chagga quickly grasped the importance of education as a means of improvement within the colonial structure. As a result, mis- sion enrollment showed remarkable progress in the period preceding the First World war. By 1914 Shann [1956] counts 20,000 children enrolled in mission schools in Kilimanjaro. Although this figure may be some- what high it does indicate that this area of Tanganyika established its lead over the rest of the country in access to education at a very early stage. Mission education throughout Tanzania was, at one and the same time, divisive and a stimulus for improvement. Generally strict, it promoted European ideals of behavior-—individualism, enterprise and material improvement. In so doing it weakened many traditions which tended to make material improvement a collective as opposed to an individual affair. As a result it encouraged a process of internal differentiation which continues to the present. Attacking many tradition37 important r balance—ml societies. llaji-Haji L and their c The \ op;ortunit3 The Christi mdern man. employment in it a bat ceived in ' the technir lEt whatevr Cation for System led Kanc 3 running tr. It cerned 1920'S diffict lSlam DDOr the ' 13 traditional customs as uncivilized-—sometimes not realizing the important role they played in maintaining a delicate techno-ecological balance-mission teachings often had the effect of dividing African societies. The seriousness of this division became apparent in the Maji-Maji uprising which very clearly was directed against missions and their converts.1 The very fbrces which divided traditional societies, created an opportunity for personal improvement which previously did not exist. The Christian educated convert came to be regarded as an example of modern man. Some were finding in modern education access to wage employment and the beginning of capital accumulation. Others fbund in it a basis for social and political authority. Still others per- ceived in it the means of overcoming what they had come to believe was the technical weakness of their own societies 31§_g_yj§_the Europeans. Yet whatever the motivation fbr individuals and families seeking edu- cation for their children, mission education with its fareign value system led to growing economic and social division. Ranger [1969] very succinctly sums up the feelings which were running through Tanzanian society at that time: It can be seen then that as far as religious ideas are con- cerned there was a great deal going on amongst Tanzanians in the 1920's and 1930's and that not all developments were in the same direction. Some people were finding in orthodox Christianity or Islam a way of entering the new colonial world and of making opportunities fbr themselves. Other people were reacting against the inequalities of the colonial world, often drawing upon the Bible and the Koran to do so. There was tension between the 1The Maji-Maji uprising was an armed uprising against German rule which took place in 1905-07. An estimated 120,000 Tanzanians died befbre active resistance was crushed. Tanzania's total popula- tion at this time was only 4 million. FOr more detailed accounts of the uprising, see Gwassa [1969] and Iliffe [1969]. imprOVETS a: argumnts 5‘ NCO“ iii-ml It is H homers v4 50139 of U181 but under s1 lity and d1 customs and They were 5 Christian 0 remain in e 4. Table 1.2 in Kilimanjaro to 1921, the fi onvard, the fig The rate Deriod 1921-31 Tanzania, Fro: increased to a fill the mainla 1'11 the distric ”t' °f growth [hegga llve ti": 1“ Spite EXlSted prlOr 0f the 20th CE Penn 397‘ l4 improvers and the egalitarians. In this way these religious arguments started many of the political arguments which have become important in independent Tanzania. It is important here to realize that the critics of the improvers were not all of them blindly reacting against change. Some of them, like the African National Church, wanted change but under some sort of control so that it did not bring inequa- lity and division with it. They also wanted to protect the customs and ideas of African society from unnecessary attack. They were saying in their various ways that a man could be a Christian or a Muslim or adopt new economic practices and still remain in essentials an African. 4. Population Growth and Land Pressure Table 1.2 describes the growth in the population of Chagga both in Kilimanjaro and on the mainland as a whole. The figures date back to 1921, the first year for which an estimate is available. From 1948 onward, the figures are actual census results. The rate of growth of population in Kilimanjaro District for the period 1921-31 was 1.9 percent, about the same as fbr mainland Tanzania. From 1948 to 1967 population growth in the district increased to almost 3.2 percent per year as compared to 2.5 percent fbr the mainland. However, the growth rate of the Chagga population in the district was not as great as that fbr the district as a whole, amounting only to 2.9 percent per year from 1948 to 1967. The same rate of growth no doubt prevailed in the survey area since few non- Chagga live there. In spite of the fairly high rates of growth in population which existed prior to 1940, land pressure in Kilimanjaro was not a pressing social problem, although it was in evidence. During the first quarter of the 20th century ample land for new settlement was available within k 1Ranger [1969, p. 184]. 111111.111 roe-.11 KlLlV; __________ Ki 1 1 n3 n leer Cha§§5 hi 12mm5 1923 143,01? 1131 154,860 on 228,412 so 11.1. 1961 393,101 Carla, 1911c], non]. SOURCE: 1i aEstinatez'. hrstrict. l1nd odes cExcludi' dlnClude etxclud- 15 TABLE 1.2 HUMAN POPULATION IN MAINLAND TANZANIA AND PRESENT-DAY KILIMANJARO DISTRICT, 1921-1967, INCLUDING THE NUMBER OF CHAGGA Kilimanjaro District Mainland Tanzania Year Chagga Total Chagga Total 1921 128,0003’b 136,ooo""b’c N.A. 4,124,328 1928 143,013b 147,447":c N.A. 4,740,706e 1931 154,860 162,057d N.A. 5,063,660 1948 228,412 259,646 239,215 7,480,429 1957 N.A. 351,255 318,167 8,188,466 1967 393,707 476,223 440,239 11,958,654 SOURCE: 'Tanzania National Archives File 5/23/69, Tanzania E19713, 1971c], Lucas and Philippson [1973] and Egero and Henin 1973 . aEstimated by author. Official 1921 figures included Pare District. bIncludes Moshi Town. cExcluding non-natives. dIncludes 3,802 non-natives. All totals far the district from 1931 on include non-natives. eExcluding about 30,000 non-Africans. the traditional 1 production expani First Horld liar. was put in perma' evidence as perm. lover belts. By approaching 400 the first time 1 landless_ The gradua toward coffee pr Droduction to t'r tion of Killmdnj Changes in the C COifee Provided Population. At lowlands Proyi d1 16 the traditional kihamba belt at 3500-5500 feet.‘ Then, as coffee production expanded and population growth accelerated after the First Norld Nar, pasture land which used to surround the homestead was put in permanent crops. By 1945-50, mounting pressure was in evidence as permanent crops spread into the less favorable upper and lower belts. By 1967 population density had become a serious problem, approaching 400 per square kilometer in the kihamba belt proper. F0r the first time large numbers of male heirs faced the prospect of being landless. The gradual shift in land use from pasture and annual crops toward coffee production, and the accompanying shift in annual crop production to the drier lowlands, enabled the rapidly growing popula- tion of Kilimanjaro to be absorbed by means of successive marginal changes in the Chagga agricultural system. The steady expansion of coffee provided ever increasing cash income to the area's growing population. At the same time, expanding production of maize in the lowlands provided households with at least some of the calories necessary to replace the milk, bean, millet, banana, and meat produc- tion lost to coffee and not replaced by market purchases. On balance IChagga agricultural land is differentiated into two types: kihamba and shamba. Kihamba land is clan land to which the occupant has what amounts to permanent freehold rights. Traditionally located in the well watered middle-belt of the mountain, the kihamba is where an individual establishes his residence and plants permanent crops, almost always coffee and bananas. Shamba land is less securely held. It lies on the lower slopes of the mountain and is utilized mainly far maize and beans. Traditionally Shamba land was held on a year-to- year basis at the discretion of the chief. Today shamba tenure is growing increasingly secure and fathers desiring to acquire an inheri- tance for their sons often are obliged to purchase Shamba land. Although officially the sale of land in Tanzania is illegal, practi- cally it is quite common in Kilimanjaro with good coffee land selling for $600-$1000 per acre. For a more detailed discussion of Chagga land tenure customs see Johnston [1940] and Maro [1974]. the Cha§9a a although cle selling PC? By the prosperous 2 been aliena‘. area's grow health serv‘ no srall va. ting the po' attracted t the highest intern Cane were Proud 0f 1Ihat ca; Can bring. let u Extent and “Cry C956,“. 21 try tn 7‘ 17 the Chagga appeared to be adapting quite well to their swelling numbers although clearly, declining availability of well watered land and swelling population pointed toward a less promising future. 5. Kilimanjaro at Independence By the end of the colonial period Kilimanjaro was an apparently prosperous area, constrained only by the large land area which had been alienated to Europeans-—increasingly eyed as an answer to the area's growing population pressure. Consumption of education and health services were among the highest in the country, facilitated in no small way by the cash income generated by coffee, but also reflec- ting the political power of the Chagga and the pleasant climate which attracted the missionaries. Total consumption and cash incomes were the highest of any rural area of comparable size in Tanzania and modern cement houses and radios were visible everywhere. The Chagga were proud of their accomplishments and were held out as an example of what capitalist development and production of cash crops far export can bring. Yet underneath this prosperity is a society under pressure, the extent and nature of which has only recently become apparent. Far every cement block house, there is one with a banana leaf roof;1 for every two radios, one malnourished child.2 Traditions of c00perative 1The field study showed 15 percent of main dwellings have cement block walls and 12 percent have grass roofs. 2The study also showed that 35 percent of the households have a radio or one for every 14 people. Among children attending under five clinics in Kilimanjaro, Lindner [1972] faund 20 percent mal- nourished (less than 80 percent of Harvard Standard). Children in this age group account fOr 20 percent of the district's population. An earlier study, conducted in 1968, reported that 28 percent of 1100 distributior to protect 1 production 1 nllet-ail And the trac tion of hig' ceased to be thing was 111 appear to be In the tural develc laid down C: “landed 51'. 9TH an aye. SDite of a . its Export ' Continued t1 mm"; but ”Ewe. 18 distribution of f00d within the extended family or clan, which served to protect nutritionally vulnerable groups, had atrophied. As coffee production has increased, farm production of meat, milk, beans and millet-—all staple protein sources in traditional diets-—has declined. And the traditional irrigation system that was important f0r produc- tion of high protein food crops in earlier generations had long ago ceased to be adequate for the needs of annual cropping. Clearly some- thing was wrong and increasing cash incomes and education did not appear to be providing a very effective solution. D. Post Independence Political and Economic Development 1. Prior to the Arusha Declaration In the first years after independence the direction of agricul- tural development in Tanzania and Kilimanjaro fallowed the pattern laid down during the colonial period. Peasant and estate agriculture expanded side by side. The market value of agricultural sector GDP grew an average of 6.9 percent per year between 1961 and 1966, in spite of a 7 percent decline in the prices Tanzania was receiving fOr its export crops [Tanzania, 1964 and 1972]. Increases in production continued to arise primarily from expansion of the area under culti- vation; but in Kilimanjaro, evidence of intensification began to emerge. The commercialization of African agriculture was proceeding apace over this period. The proportion of agriculture sector GDP preschool children randomly selected in Kilimanjaro District had moderate to mild protein calorie malnutrition and 5 percent had severe Kwashiorkor or marasmus [Tanzania Nutrition Committee, 1970]. jeri vet percerl few if" ferenti in cert land ho tion of tart so society w'l‘l‘thou;j the con‘ Mastic 19 derived from the subsistence sector declined from 59 percent to 53 percent between 1963 and 1970 [Tanzania, 1964 and 1972]. Although few time series data exist it appears that social and economic dif- ferentiation within the peasantry was increasing as well. Gottlieb [1972] questions the overall magnitude of this differentiation based on an analysis of 1969 Household Budget Survey data. But Mbilinyi [1975], citing cross-sectional studies by several authors, shows that in certain areas, especially the most commercially developed, housing, land holding and labor use patterns leave no question as to the direc- tion of economic evolution. Public sector and service employment became an even more impor- tant source of social and economic differentiation within Tanzanian society especially between urban and rural areas at this time. Although salary levels were held remarkably constant over the period, the continuing replacement of expatriates with Tanzanians meant rapid promotion to higher paying positions. Since the fbrmer were paid largely by external sources this shift generated explosive growth in the public sector wage bill. Between 1961 and 1966, fOr example, the public sector wage bill increased 75 percent versus a 9 percent increase in public sector employment. This compares with a 19 percent increase in the retail price index of goods consumed by wage earners over the same period [Tanzania, 1964]. As a result, the average pur- chasing power of the urban employed increased at least 50 percent while rural per capita purchasing power increased by no more than 5 percent. This rapid differentiation between the employed tenth of the labor fOrce and a rural self-employed eight-tenths was causing increasing government concern [Green, 1974]. 20 Education continued to provide the principal medium of access to public sector and civil service employment and a visible class of edu- cated elite was emerging. This class had easy access to the better schools, especially English medium schools, for their own children. In spite of government efforts to provide more balance to educational opportunities, the Chagga still maintained a considerable edge over other tribal groups in Tanzania with respect to access to quality education. At the time of the 1967 population census Kilimanjaro District still had proportionately twice as many children enrolled in school as the mainland average and a slightly larger proportion enrolled in secondary school [Tanzania, 1971b]. In 1964, Tanzania launched its First Five Year Plan. The plan projected a growth rate of 14.8 percent for the industrial sector and 7.3 percent for agricultural output. However, while the surge in com- mercial agriculture and cash crop exports which occurred between 1964-66 pushed the share of agriculture in GDP from 57 percent to 59 percent, investment and production in the industrial sector fell short of expectations. Principally this resulted from the inability of ministries and parastatal organizations to implement projects as quickly as intended over the first plan period. But it also was due to a planned heavy reliance on private investment and foreign aid which did not materialize for political reasons. It was the latter which proved particularly sensitive and in no small way, led to the Arusha Declaration.1 1Rather than the 78 percent of the five year development budget expected to be covered by external loans and grants, only 41 percent of the actual development expenditures had been so financed by 1967. See Tanzania [1967] and Henick, et a1. [1968]. WEEK ll nial at others I ticna‘: cratic c allPtasiz in TanZa relying 1101‘. not culture I ) l‘. 9”: lea: owning S, rieS‘ch Sir directed APas-id De $550” an OCCuy-red . 21 2. The Arusha Declaration and Its Aftermath In 1967, concerned with the growing inequality which was engulf- ing Tanzania and a growing economic dependence which threatened its freedom of political action, Tanzania committed itself to a radical break with the pattern of development it had inherited from the Colo- nial administration. In the Arusha Declaration the government declared its intent to gain effective control over the economy and to implement policies of African socialism and self-reliance in Tanzania. Hitting hard at income disparities and persons who live on the work of others (landlords, persons who hire labor), the declaration called fOr nationalization of the major means of production and effective demo- cratic control over public institutions by workers and peasants. It emphasized that hard work and not money would bring about development in Tanzania and pointed out the danger to Tanzania's independence of relying on fbreign gifts and loans fOr its development. The declara- tion noted that Tanzania was predominately agricultural and that agri- culture must be the basis of its development. It laid down a strin- gent leadership code, applying to all TANU members, which forbade owning shares in companies, collecting rent and receiving two sala- ries-widespread practices at the time [TANU, 1967]. Since 1967, many of Tanzania's development policies have been directed toward implementation of the principles laid down in the Arusha Declaration. Nationalization of banks, insurance companies, import and export trade, and agricultural processing industries occurred immediately. Government acquired a controlling interest in other firms employing large numbers of Tanzanians. In responding to the call to ”put great emphasis on actions which will raise the 22 standard of living of the peasants and the rural community,“1 the Party adopted, in October of 1967, the policy of gjamaa Vijijini or, literally translated, "Brotherhood in the Villages.” Under this policy the government initiated a program of integrated rural develop- ment intended to lead in the direction of a peasant socialist society. Hith respect to rural areas it was not until the beginning of the Second Five Year Plan (1969) that Tanzania began seriously imple- menting the policies laid down in the Arusha Declaration. Construc- ted around the five principles of social equality, glam a (brother- hood), self-reliance, economic and social transfbrmation, and African economic integration, the plan sought to improve the basic material conditions of the mass of the population. Rural water supplies, fOOd production and adult education were not singled out for special emphasis, rather they were part of a broad based rural development strategy. It was primarily the drought of 1973-74 and a growing feeling that peasants could not be effectively mobilized to plan their own development unless they were at least somewhat literate that estab- lished the high priorities those areas currently receive. The reaction to the Arusha Declaration in Kilimanjaro has been one of mistrust. Intensely individualistic and distrustful even of each other, farmers there fear any suggestion of ngm a production and would, no doubt, actively resist any effort to make them partici- pate. Partly in recognition of this, government has let it be known that people in Kilimanjaro already live sufficiently close together that few if any economies in providing services could be obtained by 1TANU, 1957, p. 20. their WV join 1,099 taeir com tion has The ez;anding surroundi feeding 5, animals it the tradi which thi \ By 2“ , .. .& H the [31’0ng DEESdPtS ‘ 0War the . Em vi' We: was ' Wiper-a” 2 bits of i Stra{CPS hnnsmn aCCQ'JrIt 0 n “Counts 23 their moving into villages. Government still encourages farmers to join together in agricultural and small scale industrial pursuits for their common benefit. However, since early 1974, cooperative produc- tion has been played down throughout the country.1 E. The Impact of Historical Forces on the Kilimanjaro Farming System and Diets‘ The cattle enterprise was the first to come under pressure from expanding coffee production. Initially, coffee took the grasslands surrounding the kihamba belt and necessitated a 100 percent stall feeding system. This increased the work load of women and made the animals much more dependent on man for their food. It also eliminated the traditional morning grazing and the frequent contact with the bulls which this provided. Judging from the very long calving interval on 1By the end of 1974 an estimated 3,000,000 Tanzanians lived in Ujamaa villages. In the spring of that year, growing impatient with the progress of the shift to Ujamaa villages, government "ordered“ all peasants to group themselves into villages before the end of 1976. Over the course of 1974 the concept of planned villages rather than Ujamaa villages came to the fore and gradually became the f0cus of what was amounting to a massive bootstrap resettlement effort. Peas- ants living in planned villages are not required to participate in cooperative production activities. 2Most of the historical information on diets is drawn from small bits of infbrmation provided by early explorers and colonial admini- strators. See, fbr example, the accounts of Kraph [1968], New [1968], Johnston [1886] and Dundas [1968]. Stahl [1964] gives a more general account of Chagga history. In addition to these published accounts, I have relied on oral accounts of Chagga elders. These accounts tend to make a much stronger case for the negative impact of coffee on fecd production than that made in the text. However, there appears to be a distinct tendency on the part of elder Chagga to glorify the past more than it deserves to be. As a result, I have tended to discount much of what they have reported. tne mi“ of the a' because I it becamt This, in stack, at farther a led to a lengtneni that 6116! uch as E Populatio than one- 24 the mountain (28-29.months) this appears to have induced a lengthening of the average calving interval by six months or more. Furthermore, because coffee yields are severely depressed by dense banana cover, it became necessary to reduce the number of banana clumps per acre. This, in turn, reduced availability of banana feedstuffs for live- stock, and increased the work load of women still more as they went farther and farther afield for more and more grass. No doubt it also led to a decline in the standard of feeding. Coupled with the lengthening of the calving interval and farmer observations it appears that average annual milk production per cow may have declined by as much as 50 percent over this period. Allowing fbr the growth in human population, farm production of milk per capita today is probably less than one-quarter of what it was 50 years ago. Over this same period bean and millet production also declined due to expanding coffee production, though it is difficult to say by how much. Fermerly, both were intercropped with bananas, pulses being used to regenerate the soil after a millet crop. Dundas [1924] gives a detailed account of the importance of millet production to Chagga agriculture in the 1920's. Older accounts note the importance of both millet and beans. By 1973, only about 43 (:3) percent1 of households cultivated .1 hectare of millet or more. A much higher percentage, ' 87 (12) percent, still cultivated beans (.1 hectare or more) but everyone readily notes the marked decline in both production and use of beans over their lifetimes. Shifting bean production from a dry season irrigated system in the kihamba belt to a rainy season lowland 1Numbers in parentheses fellowing mean population estimates are their standard errors. system CT‘GP fa once as to inte all rai results it is t 25 system has exposed yields to the vagaries of weather. As a result crop failures are frequent when rains are heavy or bi-modal, about once every two or three years. More recently, there is a reluctance to intercrop beans with hybrid maize. Hhereas 30 years ago virtually all maize was intercropped with beans [Swynerton, 1945], unofficial results of the recent agricultural census show that now 75 percent of it is grown in pure stands.1 Paralleling the decline in milk production in Kilimanjaro has been a decline in per capita meat production. It is impossible to tell whether consumption has declined, however (according to Chagga elders it has), since commercial meat sales using animals purchased from outside the area have grown fairly steadily. What is clear is that over the past 50 years the cattle-man ratio in the district has fallen from about 0.8 or more to less than 0.3 with no evidence of an offSetting increase in the rate of offtake. Nor is there any evidence that production of meat from goats and sheep has increased.2 1The reader should be careful not to place too much faith in the results of the agricultural census. Fbr some crops census results differed by such a wide margin from previous estimates that the government was considering not releasing the data at all. It showed, for example, no millet production at all in Kilimanjaro Region; bean production less than half of that implied by the very reliable mini- mum estimates obtained in our study (proportion of households with .1 hectare or more of beans); and maize production less than half of what is implied by reasonably good consumption, marketing and average yield figures. 2Dundas [1924] gives an account of the German attack on Sina, the chief of Kibosho, in 1891 that suggests a cattle-man ratio of about 1.0 and a small ruminant-man ratio of about 2.0. Hill and Hoffett [1955] cite cattle estimates, and the British Naval Intelli- gence Division [circa 1920] cite population figures that suggest a ratio in 1910 of more than one, but over a larger geographical area. Finally, in 1924 Dundas [1924] the first British colonial administra- tor and longtime resident of Kilimanjaro, noted that Chagga cattle holdings numbered between 140,000-150,000. The Chagga population fien Cc? tether With the data on ‘39 basi lithin t \ at this EVErage l a S ”'5 tong T ‘aizania 2n - 5 1n51Ce‘ 13the ET. ‘5" in s L1379] :310r‘ies “OEr fl 26 While bean, millet, meat and milk production have been declining, maize production has increased dramatically. Indeed it has been the availability of maize which has permitted reduction in the density of bananas so necessary to improve coffee yields. Reasonably precise estimates of the amount of increase are not available but marketed production for what is now Kilimanjaro Region grew from about 800 tons per year in 1939-41 to 12,000 tons per year in 1965-67. Most of this appears to have been produced by African smallholders and is in addi- tion to sharply increased household consumption.1 Given the rather high level of malnutrition among preschool children in Kilimanjaro documented in other studies [Tanzania Nutri- tion Committee, 1970; Lindner, 1974], the question arises as to whether the declining production of high protein foodstuffs associated with the expansion of coffee is a contributing factor.2 No historical data on nutritional status exist, of course, but we can speculate on the basis of changing consumption patterns and distribution traditions within the family.3 at this time was about 130,000. As a rough approximation, then, average figures of .8 to 1.0 head of cattle and 1.5-2.0 head of sheep and goats per capita seem reasonable up to about 1920. 1Swynerton [1945] reports marketed production of 1390, 561, and 516 tons for 1939-41, respectively. Figures for 1965-67 are from Tanzania [1974]. 2Both Lindner [1974] and the Tanzania Nutrition Committee [1970] indicate that insufficient protein rather than insufficient calories is the major cause of malnutrition in Kilimanjaro. This is not uncom- mon in societies where bananas are a staple foodstuff. However Zalla [1979] provides quantitative data on diets in the area that indicate calories are much more limiting than protein fbr all but children under five. 3For a description of traditional diets and distribution tradi- tions see Freyhold et a1. [1973]. nct had ig;orte declini* heat tT'h Ere 8V1 cambina‘ increase 27 For adults the shift in production described above has probably not had a significant nutritional effect. Cash purchases of meat imported into the district probably offset most if not all of its declining production. Traditionally adults consume less milk and more meat than children, and consume millet only as local beer-one product more available than ever before. The main fecds they consume less .today are bananas, beans and milk. Increased maize consumption would go a long way toward offsetting any protein lost. The traditional fbods probably have a somewhat better amino-acid balance given the combinations in which they are prepared but if meat consumption has increased at all it would make up with crude intake what was lost in balance. Fer older children the ready availability of beans and milk in family fbods has always been important in Uchagga. Unlike meat dishes, those prepared with milk and beans are distributed within the family in proportion to total intake. Meat, on the other hand, is known as a man's food which often is maldistributed. Had the shift toward maize not been accompanied by a decline in beans and milk, older chil- dren would most definitely be better off. As it is, however, they may have fallen back some, though not as much as the younger ones. Children under five are the critical problem. Many parents are reluctant to feed weaning children meat or beans. Hithout milk added to the traditional porridge it is difficult to see how they can obtain sufficient protein intake for adequate development. To be sure, the variety of millet historically grown in Kilimanjaro has one of the lower concentrations of protein among millets (7.5 percent) and is considera when mixe and mille and 1133126 To of possit income mi by consic of food c remains a inmstc in full-1 Wket it tended t: Ufcash c increasec the growi has been Utich Illa]: (‘5 been Wei [C iér. HaDs, 28 considerably lower in protein content than maize (9.4 percent).1 But when mixed with milk in the traditional fashion, the resulting milk and millet porridge has about 75 percent more protein than the water and maize porridge in common use today. To a person not familiar with traditional societies the question of possible adverse changes in diet associated with increases in cash income might never arise-—especially where cash income has increased by considerably more than the market value of the loss in production of food crops. But production of food crops fbr household consumption remains a paramount concern at the farm level in Tanzania as it does in most of Africa. Few households without at least one member engaged in full-time off-farm employment rely more than marginally on the market for staple foodstuffs. Concessions to the exchange economy have tended to be in the form of land and labor resources for the production of cash crops, not so much with specialization and exchange leading to increased productivity and aggregate output as an object, but rather, the growing need for cash fbr non-food purchases. The result too often has been a decline in production of nutritionally superior foodstuffs which makes itself felt more severely at those times when cash income has been exhausted and farmers cannot purchase foodstuffs in the market [00115, Dema and Omololu, 1962; Lev, 1981]. Even more serious perhaps, is the tendency to make marginal shifts in consumption 1Red finger millet (Elusine coracana) of the type grown in Kilimanjaro has a crude protein content of 7.5 percent versus 10-11 percent for bulrush millet (Pennisetam typhoides). By comparison white maize, the traditional variety, has a protein content of 9.4 percent. Recently introduced hybrid varieties of maize range one- half to one percentage point lower than this. See FOod and Agricul- Eure Organization and Department of Health, Education and Welfare 968 . patterns much in tion. we nave had dren is a available iismissec The cussion i to text c ‘fiPfirtan' 29 patterns which appear insignificant or reasonable at the time but which in the long run take their toll in terms of health and nutri- tion. Hhether or not the shifts that have occurred in Kilimanjaro have had an adverse impact on the nutritional status of young chil- dren is a matter about which we can only speculate. The weight of available evidence, however, suggests that this consequence not be dismissed lightly. F. Summary and Implications The description of Kilimanjaro emerging from the previous dis- cussion is one of increasing individualism and differentiation in the context of rapid population growth and land pressure. Coffee is important both as a source of cash income fbr producers and foreign exchange for the country as a whole. Yet the spread and intensifica- tion of coffee production is not without problems. It has increased the pressure on land resources and may have led to a decline in the nutritional status of young children, though this conclusion must remain speculative. It has contributed in a very direct way to dif— ferentiation both within Kilimanjaro and between Kilimanjaro and much of the rest of Tanzania. It has made Tanzania very dependent on international markets fbr disposing of the products of the labor of Kilimanjaro peasants and for acquiring dairy products fbr their con- sumption. What this means from the perspective of modern market economic theory is the subject of the next chapter. 3. ference taintai cipal 0 ist lin care an Period 110115 0 33Droac issues CHAPTER II ECONOMIC PERSPECTIVES As a nation, Tanzania is attempting to reduce the marked dif- ferences in consumption among its citizens while at the same time maintaining its freedom of political and economic action. Its prin- cipal objective is to radically restructure its economy along social- ist lines so as to provide adequate diets, clothing, housing, health care and education for all of its population within a reasonable period of time. In this chapter, some of the strengths and limita- tions of a price-oriented, Pareto optimal neo—classical project approach to economic development are reviewed. Other development issues that relate to the study are raised but no attempt is made to analyze them. These issues are raised only to bring them to the attention of policy-makers who must grapple with development in a broader context than that provided by an analysis of the returns to resources used to produce coffee versus milk, meat and manure. A. Pareto Optimal Market Economics The essence of modern market economics is its reliance on individual preference as revealed in the marketplace for determining what to produce, and on the ownership and productivity of inputs fer determining who gets the income to purchase what is produced. Free market prices communicate both relative preferences and relative 30 scarcit the val by rewa of prod individ economi equil it the use off. S , . . e.fic1e Dossibi economi Dewey-5y] imp t Vested that 83 cerhin: Pc'ii'cie {0059 W Dr alte. “‘1 in" 31 scarcities and encourage an allocation of resources which maximizes the value of output for the existing pattern of resource ownership by rewarding the owners of each resource in proportion to the value of product their resource creates. Through the pricing mechanism, individuals seeking to maximize their welfare and income cause the economic system to move toward equilibrium. In a perfectly infbrmed equilibrium, no person in the system can be made better off through the use of non-market force without someone else being made worse off. Such an equilibrium is described as Pareto optimal or Pareto efficient. Hithout interpersonally valid measures of welfare, it is not possible to determine in the context of the theory of modern market economics that society as a whole would be better or worse off if the ownership of resources, rights or privileges were shifted from one group to another via non-market actions.1 This problem has not pre- vented many neo-classical economists and policy-makers from making what Baumol [1975] essentially calls "common sense" judgements con- cerning interpersonal utility and fbllowing these up with recommended policies. It has, however, made the theory vulnerable to misuse by those who would use Pareto efficiency alone as grounds for maintaining or altering the existing pattern of ownership of resources, rights and privileges. Even if Pareto optimal market economics cannot evaluate non- Pareto optimal policies and programs, it can nonetheless predict their consequences. The behavioral assumptions of modern market 1Indeed, it is only by theoretical assumption that we can say that total social welfare is what we want to maximize in the first place. economics peasant 1 coffee ar ers will product 1 of the an another L of income in estiee result 01 at] polii help p014 allocatil CORSUi‘ney-g all“ 90‘ [ESOUl‘Ceg 32 economics apply reasonably well to the cash cropping enterprises of peasant farmers and an analysis of returns to resources allocated to coffee and dairying should give a reasonably good idea of what farm- ers will do under various non-Pareto optimal changes in input and product prices or services. Harket economics can also give an idea of the amount of income that will be transferred from one group to another under programs and policies which redistribute the ownership of income-producing resources, rights and privileges. It also helps in estimating the amount of fbreign exchange gained or lost as a result of such changes. These are important kinds of knowledge in any political context. Moreover, Pareto optimal market economics can help policy-makers identify those areas where markets do well in allocating resources and distributing products (real income) among consumers as owners of rights and privileges. Such knowledge can allow governments to focus their scarce managerial and planning resources on those areas where markets do not do so well. These strengths of Pareto optimal, neo-classical theory should not blind us to some important problems related to it. On the ideo- logical level, neo-classical behavioral assumptions conform so closely to capitalist ethical propositions that many trained econo- mists fail to see the differences between the two. Indeed, Friedman [1966] finds himself compelled to point out to his students that the proposition that an individual deserves what is produced by the resources he owns is a capitalist ethic, not a marginal productivity ethic. In a similar way, the theory lends a moral character to charging what the market will bear since only if inputs are allo- cated to their most efficient or highest use is aggregate welfare, in the P resuect as ofter the U. S hotecte and com; to earn of thosi are oft JUVEY‘SE tat 50!" t (0 .ff’? 33 in the Pareto sense, maximized. The question of "efficient with respect to what pattern of resource ownership“ is not asked by some as often as it should be. Average annual earnings of physicians in the U. S. provides an excellent example of both these phenomenon. Protected by a wide range of policies which restrict the supply of and competition between physicians they have been consistently able to earn returns to their labor and capital investment well in excess of those available to other groups within the economy. These returns are often justified in terms of their being "worth" it as though the value of services provided is independent of the supply of such ser- vices.1 In a related vein, maximizing coffee production may be a particularly efficient way of acquiring foreign exchange and increasing incomes but a particularly inefficient way of improving nutritional status among coffee producers if cash income does not translate into increased consumption of food. Many of these kinds of contradictions relate to implicit or explicit changes in the ownership of resources, rights and privileges and the need fbr interpersonally valid welfare measures to evaluate them. The question that poses itself from a development perspective is whether aggregate welfare in a more meaningful sense than that provided by Pareto optimality might be higher if resource ownership patterns were changed, or lacking the ability to do that, if resources were “misallocated” in the existing context. Lacking universally accepted, interpersonally valid measures of utility, we 1From a “common sense" point of view there can be little doubt that total social welfare would be increased if some of the rights and privileges protecting physicians were removed. have to second- preted discuss ownerst it take abstrac nine, t sociali a whole a set i give r‘ Prices. b usher ing n31 this. 34 have too often shied away from such considerations. The "theory of second-best" as articulated by Lipsy and Lancaster [1956] and inter- preted by Baumol [1965] does not really address this issue since that discussion relates to optimization after reallocations of resource ownership instead of evaluating alternative ownership patterns. Hhen it takes initial ownership as given, Pareto optimal market economics abstracts from those political and property relations which deter- mine, by and large, what any economic system (either capitalist or socialist) will produce and far whom. The ownership of resources, rights and privileges gives rise to a whole set of resource scarcities and demand patterns which generate a set of equilibrium prices. Different patterns of ownership will give rise to different patterns of demand and different relative prices. One only has to reflect on current demand for electric tooth brushes in African capitals versus what that demand would be if exist- ing national income were divided equally per capita to appreciate this. The value of computer programmers and Ph.D.'s in history would probably be similarly affected. Since the ownership of resources,rights and privileges is par- tially politically determined, it follows that prices derived from any pattern of ownership are, in effect, partially politically deter- mined as well. It also follows that evaluating public investments by adjusting such prices to correct for market imperfections, taxes, subsidies, etc. in order to obtain real opportunity-cost shadow prices does not, in fact, break the link between the distribution of the ownership of income-producing resources and relative prices. Opportunity-cost shadow prices-—or, fer that matter, market prices- in a F benefl NSC-Ur optima prices not re way as five di leges i derivir fienand ship of before they'd l £61556]- fire to It prices t tions w, ’0' alio wation invfilye . firiyed ‘ terns. tich Car 3.‘ tr esOUr 35 in a Pareto optimal equilibrium would reflect real social costs and benefits only if the underlying distribution of income-producing resources and the pattern of demands arising therefrom are socially optimal, whatever this may be. Analyzing any proposed course of action requires the use of prices which reflect socially optimal objectives. Such prices may not reflect the interests of existing resource owners in the same way as free market or opportunity-cost shadow prices since alterna- tive distributions of the ownership of resources, rights and privi- leges will alter relative prices. The problem is not merely one of deriving prices which embody the desired ex-post redistribution. Demand patterns established under an unacceptable pattern of owner- ship of resources, rights and privileges may need to be constrained befbre a country can rely heavily on markets to make rapid progress toward new social goals. An unanswered question in this regard is whether policy-makers have sufficient knowledge of interpersonal wel- fare to enable them to go beyond Pareto optimality. If the meaning of prevailing prices, and opportunity-cost shadow prices based on them are partially dependent on the political rela- tions which underly them, alternative methods may need to be adopted fer allocating scarce resources, especially where those political relations cannot be altered directly. These methods may indeed involve the use of shadow prices but not shadow prices automatically derived from prevailing prices and present resource ownership pat- terns. They would have to take account of distributional objectives which cannot be met within the context of the existing distribution of resources, rights and privileges. These prices would be no more nor less i They vault desired 5 welfare h 081: altering income-pr policy tl tion of i should h Riket e 10 take milemeh ”d for Ems--1 Th. Tanzania fiairy Ca- Provided resources UHF-ink Of mu F happ‘éng t the ”Sire he 7 edbyl 36 nor less political in nature than opportunity-cost prices would be. They would, however, be more effective in allocating resources toward desired societal objectives provided they are based on a knowledge of welfare having sufficient interpersonal validity. 8. Implications for the Analysis of Milk Production in Kilimanjaro Development defines a dynamic process. Where radically altering the existing distribution of income and the ownership of income-producing resources is an important objective of public policy there may be a need for a more explicit political determina- tion of what to produce, how it should be produced and for whom it should be produced than is likely to occur in a free market. Modern market economics can be very useful in analyzing infbrmation needed to make such determinations. Free markets can certainly be useful in implementing them. But the ultimate decision about what to produce and for whom-essentia11y a choice among alternative ownership pat- terns-is substantially a political decision. The crucial question fbr the present study is the value to Tanzania of using its land, labor and capital resources to produce dairy cattle and their products versus coffee. The answer cannot be provided without considering the distribution of the ownership of resources, rights and privileges both within and outside of Kilimanjaro. Information is required about nutritional dimensions of milk production and consumption as well as infbrmation on what happens to foreign exchange earned from coffee. One thing is clear: the answer calls fer a broader framework of analysis than that pro- vided by Pareto optimal market economics alone. in for lini analysis the inc: cattie I pare re Tnese a dairyin facilit [A 37 Increasingly, dairying and coffee can be expected to compete fer limited land, labor and capital resources in Kilimanjaro. The analysis in later chapters uses modern market economics to measure the income gains that appear to be available from various dairy cattle enterprises on the mountain. A similar analysis could com- pare returns from coffee production to returns from dairy cattle. These are important considerations for deciding the proper role of dairying in Kilimanjaro and what policies should be adopted to facilitate that role. Such analyses yield infbrmation on costs and returns that governments can use to identify policies which can make maximum use of free markets to pursue desirable social, political and economic objectives. This kind of analysis can also suggest areas where free markets are not likely to produce the intended result, suggesting the need for changes, more direct intervention or possible revision of intended objectives. In using modern market economics in this way decision-makers need to keep in mind the partially political nature of opportunity costs; the difference between capitalist ethical propositions and the essentials of marginal productivity theory; and the need to incorporate many other factors in the policy decision. He move now to a closer look at Kilimanjaro agriculture. Much o ters ls deriI sample of ap and a second some of vhic Willie selec dix A. This relates the The a Present on third; of a thOugh the hold heads had my 0n head of ho“ LEVel Stain.“ d1 CHAPTER III KILIMANJARO AND ITS DAIRY INDUSTRY Much of the information contained in this and subsequent chap- ters is derived from the results of a two-stage weighted random sample of approximately 450 households in Moshi and Rombo Districts and a second, more purposive sample of about 260 grade cattle owners, ‘ some of which were also included in the random sample. Details of sample selection, data collection and analysis are included in Appen- dix A. This chapter discusses some of the more general findings and relates them to other studies done in the area. A. Demographic Characteristics The average household in the random sample had 5.8 members present on the day preceding the interview. Slightly less than two- thirds of all households contained at least one child under five, though the ages of the household heads averaged 47 years. Most house- hold heads were male (95 percent), lived at home (93 percent), and had only one wife. Thus the normal situation is a monogamous male head of household who lives at home with his wife and family. Levels of education vary widely within Kilimanjaro, with sub- stantial differences between generations. Heads of households 38 averaged 2.' (s=2.3l- ll in the hous 6.8 years l high esteem rent for ru returns in better jobs Farmii area. Only Employment l as evangeli heads are tl nessnen and Hage ' glance, app Musehold. days of lab the lAlert/i deceiving, hiring Out . Suggest tha 39 averaged 2.9 years (s=2.8)l of education and their spouses 2.0 (582.3). However, the highest level of education attained by anyone in the household-inc1uding children who had moved away-averaged 6.8 years ($83.4). Traditionally education has been held in very high esteem in Kilimanjaro and has represented an important invest- ment for rural households. This investment has had substantial returns in the form of remittances from household members who find better jobs in urban areas as a result of their education. Farming provides the principal source of employment in the area. Only 20 percent of the heads of household hold permanent employment off the farm. Most of these were lower skilled jobs such as evangelist, clerk, cook and driver. Only 1 percent of household heads are teachers while nearly 4 percent consider themselves busi- nessmen and traders. Few women have outside employment. Hage labor within the agricultural system does not, at first glance, appear to be an important source of income for the average household. On the average, households reported hiring out 19 man days of labor and hiring in 18 man days during the 12 months prior to the interview. The similarity of these two independent estimates is deceiving, however. A breakdown in the numbers hiring in and those hiring out as well as the observations of the enumerators themselves suggest that both hiring in and hiring out of labor were seriously underreported. The Tanzanian government opposes the hiring of labor 1The letter “s” refers to the sample standard deviation. The standard error of the estimated mean is designated by (1;) and is included where an appreciation of precision is more important than a measure of dispersion. Both are expressed in the same units as the mean value. on the 9‘- deprives labor. relating out voul than nor therefor hired la 1h laborers (:5) of (:3) of th0ugh l YEt onlg labor. ratiOna' iErent 3 the mon! and the \ 1l tilally ‘ nearby i Pmduct abor h ably hi than eq 40 on the grounds that it expropriates the wage earner's surplus and deprives a person of participation in the management of his/her labor. Because of this, many farmers hesitate to report information relating to the use of hired labor. At the same time, labor hired out would tend to be underreported because of the area's greater than normal anxiety over income-related questions. The field survey therefore, probably yields only minimum estimates of the use of hired labor on farms in Kilimanjaro.1 There is some evidence of the existence of a class of wage laborers in Kilimanjaro. At the time of the survey, over 70 percent (15) of hired labor was provided by a neighbor. About 90 percent (30) of all hired labor originated from within the survey area, though perhaps not a neighbor, and had its own family coffee plot. Yet only 16 percent of households hiring out labor also hired in labor. Thus the practice of hiring labor is not simply a more rational allocation of available labor within the context of the dif- ferent seasonal demands of coffee picking at different altitudes on the mountain. Rather it probably reflects different access to land and the need to supplement limited incomes from small family plots. 1Other evidence suggests these numbers are not reliable. Vir- tually all coffee estates in Kilimanjaro rely on hired labor from nearby small holdings for picking coffee. In terms of the magnitude of output these large farms account for about 20 percent of coffee production on the mountain. Accurate population estimates for casual labor hired out from small holdings should, therefbre, be consider- ably higher than estimates fbr labor hired into small holdings rather than equal as the data suggest. Tana envi fee. cond vice area cent have rura ingl the 1 cent Seep! respe Heall tant: drea' cent VEPSL 41 B. Reflections of Health Kilimanjaro is one of the richest agricultural areas in all of Tanzania. Its altitude and rainfall patterns create a very favorable environment for agricultural production, especially for Arabica cof- fee. The mountain's wealth is reflected in everything from housing conditions to the distribution of water, health and education ser- vices. Over 18 percent (12) of the principal dwelling units in the area are constructed of cement, stone or bricks as opposed to 3 per- cent fbr the rural mainland as a whole. Moreover, 87 percent (12) have tin, tile or concrete roofs as opposed to 14 percent for the rural mainland; and 36 percent (:2) of all households get their drink- ing water from a tap or sealed well versus 7 percent for the rest of the country.1 In education the statistics are equally lopsided with 15 per- cent of the population enrolled in primary school and 1.1 percent in secondary school. This compares with 7.7 percent and .3 percent respectively for the rural mainland as a whole [Tanzania, 1971b]. Health services are less skewed with one hospital bed to 623 inhabi- tants in Kilimanjaro District versus 1:745 nationally. But the area's dense population creates easier physical access with 40 per- cent of the population living within 5 kilometers of a hospital versus 13 percent nationally [Freyhold et a1., 1973]. C. The Agricultural Economy, Kilimanjaro agriculture is quite advanced by African standards. The entire mountain is laced with a gravity flow irrigation system 1Housing tatistics for Mainland Tanzania are taken from Tanzania [197 d]. fed b snow shat; by th with sarke outsi Dette stapl fen v salve 42 fed by ground water retained by the rain forest as well as by melting snow from the mountain's peak. Rainfall and snowfall patterns vary sharply from one side of the mountain to the other, dominated largely ’by the influence of the peak on prevailing winds as these interact with temperature changes at different altitudes.. This results in marked differences in water available for agriculture both during the rainy season and the dry season, depending on altitude and aspect. Coffee does best in the 3000-5000 feet range, though most farmers outside this belt also grow coffee. Maize, on the other hand, does better in the drier, hotter lowlands. Bananas, the traditional staple in the Chagga diet, seem to do better under the same altitude, temperature and resulting rainfall conditions as coffee. Almost 95 percent of households in the survey area grow at least some coffee and nearly 90 percent grow at least .25 hectares of maize and .l hectares of beans. Farms in Kilimanjaro are general- ly small, estimates ranging around 1.2 hectares of mountain (kihamba) land and .8 hectares of lower lying (shamba) land [Beck, 1961; Marc, 1974; Sykes, 1959; Wallace, 1968]. Of this, about .5 hectares are planted in coffee and bananas with coffee yields averaging between 320 and 380 kilos of parchment per hectare [Coulson, 1972; Mphuru, 1965]. Cash income from coffee accounted for roughly 70-75 percent of total cash income from agriculture for small holders in 1967 [Ministry of Economic Affairs and Development Planning, 1968]. The proportion would have been at least this great in 1973 when coffee prices were considerably higher. Apart from coffee, the agricultural economy of Kilimanjaro is based on bananas, maize, beans and livestock. Some bananas, between Sandi banana establ banana cropoe ping a take 1 in ID! homes valk great the 1 313111) as pa Show: Stocl Klllg MEat goats for 1 data QCOHO “39d fiftieth geDEr probe tgmmo: 43 5 and 10 percent, are grown in pure stands. Generally, however, bananas are intercropped, first with maize and millet and then, once established, with coffee. At any point in time, about 90 percent of bananas are intercropped with coffee. Virtually all coffee is inter- cropped with bananas. Together coffee and bananas are the main crop- ping activities on the mountain kihamba homestead. Maize and beans are important to Kilimanjaro agriculture but take up very little kihamba land. Rather, both tend to be planted in the drier lowlands on plots of land (shamba) detached from the homestead. In general, maize and bean shiny; are one to two hours walk from the homestead. In some cases, however, the distances are Agreater. Some maize and beans are grown on the upper slopes near the fbrest reserve (6000 feet) but both do poorly there. More com- monly, land near the forest reserve not in coffee or bananas is left as pasture and forage fer livestock in nearby homesteads. Table 3.1 shows the minimum proportion of farms having specific crops and live- stock enterprises as indicated by the random sample.1 Table 3.1 gives some idea of the importance of livestock to Kilimanjaro agriculture. Cattle, goats and sheep contribute most to meat production but swine and poultry are also important. Cattle and goats are important sources of milk while all livestock are valued fer their manure. Slightly more than 10 percent (19.8) of households lBecause of the very sensitive nature of acreage and income data in Kilimanjaro most questions alluding to important crops and economic status were general or indirect. For coffee, for example, farmers were not asked whether they grew coffee but whether they used certain cultural practices. Maize and bean farmers were asked whether they grew .25 or .1 hectares or more respectively. In general, such minimum estimates should not be far from the actual proportions since cultural practices and field sizes known to be common to most farms were used. ans 44 TABLE 3.1 PERCENTAGE OF HOUSEHOLDS IN KILIMANJARO WITH SELECTED AGRICULTURAL ENTERPRISES OR GROWING SPECIFIC CROPS Producing Households as Enterprise Percent of Standard or Crop Total Errors Bananas 99a N.E.b Coffee 94c 1.4 Maize 89 1.9 Beans 87 2.0 Cattle 67 2.8 Goats 57 2.9 Millet 43 2.9 Cabbage 32 2.7 Tomatoes 25 2.5 Squash 22 2.4 Spinach 11 1.8 SOURCE: Random sample estimates. aQuestion not actually asked because of its obvious answer. bNot estimated. cActually the proportion of farms which prune their coffee. have one or more grade dairy animal. None of the households in the random sample had exotic goats, though one in the grade cattle sample (n826l) did. The 67 percent of households which have cattle have an average of 2.3 animals each, most of which are stall fed. Carrying forage to feed livestock is one of the principal tasks of women. The average household with cattle spends 20 hours (39) per week gathering grass, banana leaves and banana stems fbr feeding livestock. Manure from livestock is highly valued and is the principal source of ferti of Ki the a; as fertilizer fer the mixed coffee-banana groves that form the mainstay of Kilimanjaro agriculture. Cultural practices among farmers in Kilimanjaro are quite advanced fer smallholder agriculture in Africa. Very few farmers with coffee do not prune at least once a year and more than 85 per- cent spray fungicides to control coffee berry disease (080) and leafrust. This proportion is down from previous years because of the emergence of a strain of C80 that is resistant to conventional copper oxychloride compounds. More than 95 percent of coffee produ- cing households use insecticides to control stem borer and leafminer insect parasites on their coffee. Only about 6 percent of farmers use chemical fertilizers on their coffee. Soils on the mountain are very fertile and are main- tained by applications of manure and by mulching with banana leaves. Coffee benefits from heavy applications of manure on the bananas which are intercropped with coffee. Nevertheless, coffee yields on small holdings continue to run at about one-third to one-half the level of the large estates. Better pruning techniques, more frequent and timely spraying, more frequent irrigation, fertilizer applications and a further reduction in the density of banana stems could raise yields on small holdings. However available evidence indicates that reducing the density of banana stems by itself would not stimulate sufficient yield increases in coffee to offset the lost value of the bananas, much less the banana leaves and stems which are important to the animal populations of the area [Nallace, 1968]. A relatively small percentage (16 percent) of farmers have planted hybrid maize at least once but this was just becoming gener PTOPC tinue terns decis lands past . farni Chagg; Yield Milk y of the val be same 1 the nu Sense lng as I”CFEa lllS- . r E} .J 46 generally available during 1973, the year of the survey. Not a small proportion (22 percent) of farmers trying hybrid maize fail to con- tinue using it. This is related primarily to marginal rainfall pat- terns in the lowlands and appears to reflect a rational economic decision on the part of the farmers. Erratic rainfall in the low- lands may also be responsible fbr the large proportion of farmers growing hybrid maize who do not use fertilizer (77 percent 17). More than one-third of those trying fertilizer on maize discontinued its use. Table 3.2 summarizes data on cultural practices in the area. 0. Alternative Milk Production Systems In spite of the tremendous inroads made by coffee during the past 40 years, cattle are still a very important component of the farming system in Kilimanjaro. Cattle manure is the mainstay of the Chagga banana groves which, with bananas valued at market prices, yield more income per acre than coffee [Sykes, 1959; Wallace, 1968]. Milk continues to be highly prized and its relative scarcity is one of the most frequent of complaints. In addition, it's nutritional value, especially fOr young children, is widely appreciated. At the same time meat production continues to be important, attested to by the number of households that have only one male animal. In this sense at least, it is not appropriate to speak of cattle and dairy- ing as though the two are synonymous, especially fbr zebu cattle. Increasingly, zebu cows are used as breeding stock for an artificial insemination program in the area. CMFLU ..uU CZHUZ UCudecam msu =9 mecm>wca xppoaucp> my mcpcaca .mmwmou gymsu voczca Aug» mcpucoamc mcmscmm mo amass: seem umaaEFHmm my mmrcacmucm mcv>ug mupogwmao; mo cowaconocmn .ma mama :o p ouocuooe mama .mmumsrpmm manom soucum "mumsom mm o_ mpuuau co wuruwcauo um: new mm 3.: 8 co mmu'uwpummcv a_aa< gem mm mmmmou co mvvu.mcae cannou zucnm gem coop mmmmou manta new no mmmmou co cm~_~_pcmm PeUVEmsu apaa< now we mNVae co Lonwpvacom —eu*Em;u apaa< mom mp emom m~rms teens; meeneepa mm «m macaw unexpez mvpogmmao: commune; umuauvvcm on: muwuumcm peace to was» emceaenpem new: “cmucoq m< mupogmmaoz mo acoucmm voccmucou mmecncmucm mcv>ez mupogmmsoz mnmp .mwo~hu<¢m 4Fm pm» we: w>oc squz mmFmEmF FFm mew memmezu .mwmmu umpcmsz mcFmemmee cF meoeem mcFuczoL on use mFauop puaxm op eve yo: Fazn .mmm mF mFaEch mo Lunsac umpcmszca mgFm .mFapOF umpcmFoz mFQEam Eoucem ”momaom mmm mm FF a F w m mF wF aw ea we ea mm anaqu cstou «N m o o o o o F a e e N o o mFan mgaumz mF F o o o o o o o m m mF om Fm mFan mgzumssF F o o o o o o o o o o o o o mgmmum va mo FF o F F m vF a F o o o o mzou eeF oF o o o F o m e mF mm mm mm mm umLoFFw: czocx Lm>o oF m w F o m e m N F FA anmqu -c: use FF Angaaaz 30¢ Fazxmm menu» :F mm< Fm4Fm Fox uo: m>mc gquz mmFaEmF FFo wee mcmmszu n .mcm we: mFmEch Fo Logan: uwucmszc: ocFo .anuou wFaEam mFuuau mecca muFmonEou ”mumoom Nom mm N m F F oF NF m FN Fm oe Fm anmqu gazFou N F o o o o o o F N v o o mFFan weave: Fm N o o F o o o o m N «F mN mFFB menunEEF mFF Fm N m o F oF FF F oF m o o mzou «NF NF 0 o o o o o N m 0N mN mm umLmFFm: czocx cm>o oF m N F m e m N F FA smquoF .:3 van FF quczuaz ; you Fozxmm meaoF :F mm< mFm4eonno yo goons: ugh .aou.e_uuo apnea“ woe—aeou uougo.ozc=~ .mucouom co.u—>—o asap—=u'ou< yo asunwcwz "wouaomu .33 .932. "wueaomu .x—ve smog» apnonoon Agu>u .noo— .awcuncap ”uuuaoma .o—na—.->u an: no: ooeuou we uu_ca curmsmp mg» 29.23 eon nua— vac .uuov ou—cn oueeou as» ou especou cunt ou.cn x—_e ago save: Lou ouroomfi coy uaouxo .eovuuuaa e. such on» ova—oc. ;u_;3 avovcoa as“ use so» ooocosn one can «to; wean." «has» .m_uun ecuxrvfia o» eemauuvs a co wanna—nan a». muu¢ea .zuzx xn vouauoca unassucun mo ecu can A—_ne co ugm.oeuv .u.o.~ cuuoaa use moo—cg .=uzx ”oueaomo A—numvmv- Laom omooo>< oug< comm mum. Aem.mvma_ guaea amaeasa a.aeam anew. nwpimm ghee“ ouago>< oesom .mn _~P Pe_-~m_ euaea uaacu>< .guoz och-aoo. an e_. ... uezcexea amoea>< .euo: unam— . mo.-o~ . ea=_a . wm~ mo Nap-mm uezoexcz ueoem Sena: ammo, Ace» moaem>< cocoa x—wz camomm uocosou gao> can my ea oaxw uuvganvo , “to“.a to; maeeuv numaaou xu~= no.1a nsinoo— .o¢0 WCKWI U.L.L.<1 p.o mamQL 103 Eoucoe ecu seem umueewumm mew mammpcmucmn m—uueu ano~-_P< .m_qsem m_upeu muecm muwmoaeou ecu soc» empee_pmm ace meow; mueem-_pa ecu umxwe com mamaucmucmm .o—aEam "mumzom q.m m.m N.N mmwm com: .m>< . _ _ +N v m N m NF om m e up om cp m we cm —v N mp 0 mm P wumcw-FP< use aan :amN-—F< m~wm new: mwmwcqcmucm mo ucmucma mnmp .omh mmHmmmmhzm >m momm: m4hhHozu mo mNHm F.m m4m mzh mo :pzoz >m omoz .uuo .amm .m:< .Faa .caa he: .ca< .cmz .nma .cec Amy Amv Amy Amy on m o |\\\ N H. o m Ao_v Ao_v A__v N_ m_ m_ mm mp mZou mecca u weep vwfiom mzou samN u secmoumw: O ,— m SHJJLB 30 JaqwnN I catl 110111 and C331 is i cal 11111 C10: int. 106 consistent with the age at first breeding, though perhaps a bit too close given that 25 percent of the zebu animals bred do not conceive on the first service. The figure of 49 months compares with 3-1/2 to 5 years reported by Pullan [1979] fbr Nigeria, 52 months for zebu cattle in Uganda reported by Sacker and Trail [1966] and 41 to 44 months found by Stobbs [1967] in a Uganda research station. Wilson and Clarke [1976] provide data from the Sudan showing that 29 percent of cows in sedentary herds and 65 percent of those in migratory herds were in calf befbre they were four years of age. e. Calving Interval. Data on both the most recent as well as the previous parturition indicate that the average calving interval for zebu cows is between 27 (:2) and 29 months (11.5) respectively. The difference between the two reflects the fact that the survey took place shortly after the peak calving season with the result that average time since last calving was shorter than would have been the case had the study been evenly spread over a year. The higher figure is based on the calving interval between the last and the previous calf and is, therefbre, a more reliable measure of the average calving interval, though sample size is smaller. The mean and median are very close to this estimate and the distribution of observations rather well spread. Figure 6.2 gives an idea of the distribution of calving intervals between the most recent and previous parturitions. Using a deductive approach wilson and Clarke [1976] feund calving intervals of 18 months for migratory herds and 30 months for sedentary herds studied in Sudan. Pullan [1979] found 27 months in Nigeria and cites other studies done there indicating a range of I5-2‘4 months. Sacker and Trail [1966] and Stobbs [1967] found 107 mnm— .ommmm mzb oz< hm<4 mzh zmmzhmm m4<>mmwzH uHmHummm p< uzummauuo mzhmHm mo mmmzzz ~.o mmzwuu .mmpeswumm mpasmm soucom ”mumaom mguc_m :mmzpmm mcpcoz me Lm>o melee omivm mmiwm NNINN .Pmqu mp Lmvc: PP T N mp T_op 591418 30 JaqwnN 108 calving intervals of 11 to 13 months but their data refer to research station and stocking farm herds respectively. Hilliamson and Payne [1978] confirm the lower figures, citing a range of 11 to 14 months as being typical of the East African zebu. Thus the figure of 29 months fbund in this study, though on the upper range of estimates found under farm conditions elsewhere, is not unreasonable for a stall feeding system facing a scarcity of bulls. The long calving interval for zebu cows arises partly from farmers' belief that zebu cows tend not to come into heat while they are nursing and if they do, breeding them will make a cow go dry as will withdrawing the calf from nursing. Whatever the truth of these beliefs, and they should certainly be researched, the mere fact that they are held suggests that a major benefit of upgrading the zebu herd would arise from breaking an actual or perceived genetic bottle- neck to shorter calving intervals. As we shall see in the next sec- tion, farmers have no trouble breeding grade cattle while they are still in milk. Upgrading through an effective artificial insemina- tion program should also overcome the second major factor accounting for the long calving interval, namely the dearth of mature bulls relative to the needs of a stall feeding system. 2. Grade Cattle a. Age at First Breeding. For grade cattle the average age at first breeding is 27 months (11.8), more than one full year earlier than zebu cattle. The mode and median are both 26 months and only 2 percent (12.9) of the females were bred fer the first time after they were 4 years old as compared with 22 percent (17.9) zebu 109 females. Moreover, 17 percent (17.7) of the grade females were first bred when they were 18 months or younger, indicating that many farmers have no trouble adopting separate breeding practices fbr grade and zebu cattle. b. Type of Breeding. Hhereas only 1 percent of zebu females were bred by means of artificial insemination, 22 percent (14.6) of grade females were serviced artificially for their first insemination since their last parturition. Slightly more than 6 percent (12.7) were serviced by zebu bulls with the remainder being bred to grade bulls. The percent of second and subsequent inseminations with A.I. is higher, 42 percent (17.1), reflecting the poor pregnancy rate (and consequent higher return rate) for artificial semen in Kilimanjaro. In spite of the high proportion of grade cows bred by grade bulls, very few farmers on the mountain have their own grade bull. Most borrow a bull in return fer caring for it until it breeds the borrower's cow. However, there is evidence that the quality of grade bulls used by farmers is downgrading the progeny of the cows served [Tanzania, 1963]. The Ministry of Agriculture would be better ‘advised to make higher quality bulls available to farmers and to encourage rather than discourage their use, at least until the arti- ficial insemination program is functioning properly. c. Seasonality of Births. Not surprising, the distribution of births over the year for grade cows is similar to that for zebu cows (Figure 6.1). The magnitude of the variation between months is only slightly more subdued. One would expect less seasonal variation in the generally better managed grade cattle population. Many of the current owners of grade cattle possess sufficient resources to pur the rel BYE bet sug tha abl far 191 1111 20. the W 1111 1111 81 br 50. 1r 110 purchase feedstuffs or transport deder over long distances to carry their animals through the dry season. This ought to reduce stress- related clustering of estrus periods. d. Age at First Calving. In line with the earlier age at first service, grade heifers give birth for the first time at an average age of 34 months (11.3), though the eight month difference between average age at first breeding and average age at first birth suggests that these mean estimates give orders of magnitude rather than precise measures. Data on grade heifers is especially unreli- able in that a large proportion were purchased off the farm and farmers do a lot more guessing in reporting these kinds of events. As with zebu cows, reported values fbr age at first calving are clus- tered on 24, 36 and 48 months exactly. The modal value is 36 months, though the median is similar to the mean. e. Calving Interval. The calving interval fbr grade cattle is 20.5 months (11.9 and 11.1) both according to the length of time since the cows' last birth and the length of time between her last and pre- vious calf. The latter estimate is the more efficient of the two, though the histogram in Figure 6.3 indicates wide variation within the grade cow population. The large number of calving intervals under 16 months suggests that many farmers see no problem in breeding grade cows while they are still in milk. Interestingly, these same farmers follow traditional breeding practices for their zebu cows. This suggests there may be something more than mere belief and tradition which prevents farmers from adopting similar management practices for their zebu cows. 111 mm cm>o .mwpeswumm m—aEwm opuumu mecca mppmoasoo mmimm mumr .ommmm oz< Hm<4 mxk zmmzhmm mzhzoz no mmmzaz m.o mmaon mcucwm :mozuom mzpcoz Fm-mm nmivm mwuom mpuwp ”mumsom mpump Np smug: P N N OF N— o T Q j N 1 mp 591418 30 JaqwnN 112 3. Summary of Reproduction Coefficients Table 6.2 summarizes the average reproduction coefficients fbr zebu and grade cows as well as those prevailing fbr the top 25 per- cent with respect to milk production in the six months following parturition. These two sets of figures represent average and superior management and/or genetic potential respectively, for both types of cattle.1 The data in Table 6.2 show that differences in parturition- related variables do not accompany whatever it is that causes high milk production fer zebu cows, though there is some association fer grade cattle. The lack of more dramatic differences arises from the use of milk production for the six-month period fbllowing parturition rather than annual milk production. The latter would have been a better measure since it combines milk production with reproduction management factors. However the large proportion of observations for which the length of the last calving interval was unknown pre- vented use of this measure. Six month milk production does give a good indication of relative milk production due to factors other than drawn out calving intervals, however. B. Calf ReariggPractices Zalla [1974] has discussed calf rearing practices in Kilimanjaro based on a tabulation of unweighted sample responses. He present only a summary of that work here, with appropriate revisions where weight- ing of responses has changed the magnitude of some of the estimates. 1Using the top 25 percent gave a weighted sample in excess of 20 for each type of cow. 113 .mpsewmmm msm mo mcoccm ucmusmum ms» mcm mmmmsusmcsa sp mcmsE:zs .msws coumcmsssm co mswucoqmccmus: com umumawum me: use umucoamc ms sovpuuuocs x_wzm .wmpsE—umm mFaEum m_mumu mumcm mmwmoasou use mFaEsm Eousmm "momzom cums Ammc cmcc sc__v acmcmm__v socmuzuocs c_cs mms_ mom omm omp smsoe xcm mmecm>< Amsusosv mp om om mm _m>cmusw msw>~su Amspsosv mm en cm as mscscau umccc me mus nmm nob mmmcm>< “mm sob mmmcm>< usmwuwmcmou mumcw :nmN mpuuem co mash mnmp .om< mom mhzmmunmmuou onhusooxamm ~.m m4mcumao Lo consaz III C... A_~o.v Am“... » Roe.fiv Aa_._v .~._v Am_.~V A.o._v » .=.u oe.v - m.m~ . z o.o “.m. m.- m.a_ m.v~ .. zv Longs ..uo» Amo.. A~_.v “0.. .~_.V an..v v. a. a. Q. ~.. o_.uo»-n=m c o o o o Lasso ac. oc.n~=ucaa .mc_umocm m. n. m. n. m. aces—.2 o. o. P. _. o. auwuee.u. ac.»,aa< ~. m. o. m. o. Luau: newc.ouno . Loa.s cozuo CC. .CC “mom.v A_~._V . Ace..v “....v Am—._. Ao_._v Axm.v .2.9 .o.. . ...N . z e.¢ ,.~_ m.o~ ..~_ o.F~ o_.uo»-a=m ~.~ o.p 0., m._ ~.~ aceuto: ¢.o ~._. m.m_ «.mp ".m— uu.co acetogu.u Auzv coaxed ucwx—qaam CC. CC. . .MNmoHVAmm_ . a Ao_. “o... .m_.v .op.v eaao.v u a . u - om — . z m. w. m._ ~.. m.. A 2v uaem oc.=~o_u - m ~ . ao.tas< .ooz L.a A. + m + a n A a. n a as: o: xcoaou-u con-4 bau~e_umu co_mmuc a .o~.m ecu: >3 .=.u to; .aa: can mesa: co Longs: MNo— .o« no mum—ammuPZu maph< ace: soc ouM“_“»w uoflumouwoa A a. a «use: snooouau son-4 u .o~.m ago: an .=.u so; ...: 5.; neaox .o cones: aha. .oa¢az¢x_4~u :. «UN—n m:o~¢<> no mum—c‘xuhzu ughho vcn nu.ea m.N can on.~ a» m._ .on._ co m. No anoene »_—n:uugonno No Lucas: CC. CC. Amo._v ....Nv h Amm.v Amo._v .m~.mv A—_._. ...... k .:.u mo.v - oo.N - z N._P v.n_ N.wN m.mp N.om 0A zv Lcan; .nuo» A_~ v Ann V Ann V A~_ v Ann v o N m _ o N m P a n npnuohinam N. o. F. p. n. gonna ucn mcvmngugaa .ocwvoocm N._ a. _. m. ..N on.._,: _. N. N. . N. c. ouNUNLnun oc_x_aa< c. v. o.— o. F.P Loan: ocvcnnuno . Loans cacao CC. CC. . . Ago ..AmmHNV a A_m . “an V Ago my Aoou_. ANN Fe 2 u me n - N MN . 2 [mum . _. o «N o n. o «N o_nuop-nam m. m. o o. m. uc.neox o.m o.o_ o.nN o.N. _.nN mango newcogunu A zv conga; ocvapaaam C. QC. Ammo..Aoo~.. a aa__.v uflmo.v nflmm.. uA._.v n.._.v .:.u ON. - Nm.. . z N. n. m.~ ..P a.. non» u¢.¢..Fu . . .n N — noncns< so»: so; mume_umw nowunou as m a a ncao: . agoonunu Loan; 6 non.m new: an .=.u so; gun: can «use: No cones: Mmap .oa no mum—amampzu ugph a... 38:. .p .2 29.2.... 5...: -3. .2 .8... ...8 .e 2.35 .3. ...-...; .25. 323.23. .3555... 3:22:25 3:33: 3n 3533 33:5; 25.25.32 5 mSu :9: z- 20.5—x; ..n was: 3.93 32.538; 1...: 5::— ._a~w vim—0:38 33352. no >¢§m 156 The sign for aspect is not what is hypothesized for the zebu model and raises the question of whether it measures what we think it does. Available grazing becomes more constrained as aspect increases but population density, rainfall and coffee production (and therefbre incomes) increase. The excluded variable Gf, the frequency of grazing, had a positive sign but at a very low level of confidence. If ASp were measuring what was expected in the zebu model Gf should be more significant. The fact that it is not suggests that little policy impact can be assigned to this variable. The three equations representing various components of the zebu model are summarized in Table 8.1. C. Grade Cows The model does a much better job of explaining variation in 1 Equation (4) grade cow milk yields especially fer Moshi District. presents the results of the technical set fer Moshi. The subscript g on ng refers to six month milk production far grade cows. (4) ng 8 515 + 19.4Hf + 23.5f + 123Gr + 1.25Cg + 7.98Lg (226) (13.1) (8.72) (52.6) (.375) (7.28) * ** * *** - 694Pb - 525Pp - 24.0Gf + 64.88n - 41.3Aw - 152En (344) (448) (19.0) (51.9) (22.5) (112) * * 1The recent introduction of grade cattle into Rombo District generated a high percentage of don't know responses for six month milk production, the dependent variable. As a result the all-Kili- manjaro sample on which the grade cattle regressions were to be based retained only feur grade cows from Rombo District out of a total of 72 cows. This, coupled with ecological differences between large parts of both districts suggested that it would be more appropriate to separate the two sets of data. Those observations fer Moshi District retained their original weights and give, therefbre, a rep- resentative picture of the grade cow population in Moshi District. 157 where: R2 = .59 F = 7.3*** *** a significant at the .OOl level ** = significant at the .01 level * significant at the .1 level. The variables fbr the frequency of feeding salt (Sf) and cash expenditures on grass per cattle unit (Cg) are significant at the .Ol level with the signs as expected. Kilograms of grain fed (6“) is significant at the .02 level, and the proportion of the ration coming from bananas, at the .05 level. The sign of the fOrmer is as expected but its magnitude is lower than it should be since it 1 are associated with an increase implies that 180 kilograms of grain in milk production by only 123 liters of miik.2 This is explained partly by the fact that about half of the grain fed to dairy cows is, in fact, bran, with a lower TDN and digestible protein content than concentrates. Recording grain normally fed to cows, as the one visit study did, introduced another source of error since many farm- ers would not feed this normal quantity continuously, especially when their stores were temporarily depleted or when they were short of cash. Thus the actual quantities of grain consumed were probably much lower than those recorded.3 1One kilo per day for six months. 2Hilliamson and Payne [1965] indicate that fOur pounds of grain (1.8 kgs.) are required to produce one gallon of milk (4.5 kgs.) in tropical breeds, with a higher ratio required at lower production levels. Five hundred grams of grain per liter of milk per day is a good rule of thumb. 3Data on cash expenditures confirm that farmers did not pur- chase as much grain as they reported feeding. Though the two mea- sures do not refer to exactly the same things the average household reported feeding 275 kilograms of feed supplements to its herd over 158 The sign of the variable representing the proportion of the forage ration coming from bananas is plausible but its magnitude is surprising. It is possible that a high proportion of bananas in the ration indicates an unwillingness to go off the farm fbr supplemental fOrage supplies. A more likely explanation is that the ratio of stems to leaves is higher in these rations containing a higher pro- portion of bananas.1 Unfbrtunately these two feedstuffs were not distinguished on the recording instrument so the actual composition of the banana portion of the ration is unknown. The sign and magnitude of Pp, the proportion of the ration coming from planted grasses is hard to accept, even allowing for the level of significance of the coefficient (.25). It is not clear why a farmer who would take the time and resources to plant grasses would not also ensure that the rest of a cow's forage ration was ade- quate. Perhaps this suggests, more than anything, that counting bun- dles is not a sufficiently precise measure of relative fOrage intake for estimating differences in the quality of the forage ration. The age of the calf at weaning (Aw) is significant at the .07 level and negative as expected. The frequency of feeding water (Hf) and the number of times given birth (8") carry the expected signs but the previous year (.29 kilograms per day per animal) and purchasing about 145 kilograms. Some byproducts are acquired from local flour mills and household milling activities and some from beer making residue, some of which may be purchased. This could not account for all of the difference, however. As a rough guess per cow grain con- sumption is probably overstated by about 50 percent. There is no reason to believe this overstatement is a selective phenomenon. 1Banana stems have a much lower TDN and crude protein content than banana leaves. 159 at .15 and .21 respectively, they are not particularly significant. The same is true of the amount of labor used to gather grass (Lg) which is significant at the .28 level. The coefficient for the enumerator dummy (En) is large, though in this case, negative and less significant than in the zebu model. The frequency of grazing carries a negative sign which was unexpected. This may indicate that grade cows in extensively grazed herds are lower grade fi's and fz's rather than more poorly fed animals, and points to the need to carefully interpret all of these variables since the data contain many crosscurrents. The dummy fbr pure bred animals, the only technical variable not retained in the final model, had a positive sign but its level of significance was only .50. This is not surprising since this variable suffers from the fact that few farmers know the genetic history of their grade cows.1 Introducing the management set of variables into equation (4) fOr Moshi District increases the R2 and has a substantial impact both on the magnitude of many of the technical variables as well as their levels of significance. This can be seen from equation (5). (5) ng = 5l.3 + ll.4wf + 22.7Sf + 1206, + .664Cg + 10.3L (259) (12.5) (8.14) (49.1) (.396) (6.83) ** 'k * (337) (430) (17.9) (49.0) (22.3) (107) * +34.» “7.65,, (15.9) (15.3) * 1The lack of knowledge about the genetic history of grade cows introduced some enumerator error as well. One of the better 160 significant at the .01 level significant at the .1 level. where: R2 = .66 F = 7.9 ** = Obviously there is collinearity between the management related variables and the level of use of some technical inputs. The increase of .07 in the R2 is only a minimum estimate of the effect of management on the production relationships. Some of the effects attributed to technical variables, such as the frequency of feeding water and cash expenditures on grass are probably due to the influ- ence of management on the selection of grade cows with above average milk producing potential. Progressiveness (P9) is positive and significant at the .03 level. This variable explains the bulk of the variation explained by the management set. The education of the head of the household, again picking up some of the effect of wealth, is positive as expected, but is significant only at the .26 level, barely qualifying it fOr inclusion in the final estimating equation. It's high col- linearity with P causes its addition to explain relatively little 9 of the variation not already explained by progressiveness. Adding the ecological/institutional set of variables brings about major changes in the magnitude, and in the case of banana ferage, the sign, of almost all previously included variables. The one variable in this set which is retained, aspect, drives the R2 up to .76, a .10 increase over the equation including only the technical enumerators, knowing full well that many farmers did not know the genetic history of their animals and realizing that truly pure bred animals were rare, recorded all-grade animals as crossbreeds unless the proprietor could convince him otherwise. 161 and management variables. Part of this increase in explained varia- tion arises from including the age of the head of household (Ah)’ a variable which does not meet the inclusion criterion when aspect (Asp) is not in the model. Both coefficients are positive with aspect significant at the .001 level and age of the head of household (Ah) at the .06 level. The positive value of the coefficient fbr aspect indicates that milk production per cow increases as the location of residence moves from Marangu to Machame and Mashati. This reflects perhaps partly the greater and more evenly distributed rainfall which falls on the southwestern slope of the mountain. But more certainly, this vari- able measures the historical distribution of veterinary and extension services and the much longer history of caring for grade cattle fOr farmers on the southwestern slope. Some internal selection has taken place and the existence of a Kilimo extension center and artificial insemination program since the mid 1960's has resulted in substantial- ly greater upgrading than has been possible in areas where farmers rely only on available low-grade bulls. Using six month milk produc- tion as a dependent variable removes the effect of the one negative aspect of the A.I. program to date, namely, the low conception rate. It seems likely that using a variable which included the effect of the longer calving interval arising from a low A.I. conception rate, such as annual milk production, would show considerably less effect for this institutional/ecological variable. The values of the coef- ficients for all three equations are summarized in Table 8.2. Table 8.2 also summarizes coefficients fOr all three versions of the model fer Rombo District. The data for Rombo are not 162 \ C .C Am.m~V Romev figmev A¢.flpv Am_o_v m.em- mmN- _.o. m.- eeem n ..eoceaueene~\~.u_oo_0uu .0 C Am.m~v A_eev .emev Ac...v Aoemv ~.M.- eke- .m. m.~. ~we m beeeaoeee: .m C. «a. Am.e~v Reamv ANeev A..m.v Amm_v w.fle- m_m- mpm- e.e_ oem ~— eOee. eeeeeesmeee ace xoo_oczuou .uuaae_ .5 uu2cum.o oacom I. I C C C CO. Am.ne. A_.~,v Amen. Amen. Aam.mv .eem.v Am.ee. A~m.~v .m.__. Amen. ..oe ..Nm- owe- “.5m ~._, mme. c.- m ~. o- “Ne.- n ..eeeusu.one~\_.u,oo_08u .e I C C C. “c.0ev “0.2.2 Acmev 25mm. Amm.ov Aeom.v A..oev Ae_.mv Am.~_. “ommv 0.0e ..~_- .ma. Nem- n.o_ «we. ON. ~.- ..PF n._m m geeseneeez .m O CC. C C. C Am._mv Ao.o_v Ameev Aeenv Am~.~v Amkm.v Ae.~mv A-.mv A_.m_v AONNV coca. beeeeeameee m.ee o.e~- mwm- «so- we.“ 0N.. n~. ..mN ..o. m.m ~_ we. soc—eeeueu .mcsae_ .e uu.eum,o _emon e a em .8 a ea 4 my 28 em 2: a ence.» “zuc_m cvuaeu vogue—a «accuse “was” «aqua c.~ca u—om Lou»: “noun nopaovca> . Logan: .coeu .aocn .noLa gonad onscugza mop—x .aucm .aogu -cou we uoeoucu “om o_ao.e~> cones: nuco.u,.eoou vow-.uoum< v:— uoc.~uo¢ u:¢.eo> ou uzuzu.dmxu mhzu~u_m «.0 u4nm._ _oo. .23 ». .e.u.e.eo_m... .pese. .0. on» u. be.u.e,eo,m.. ...... —. ecu ». ue.e.e.ea2m. ..ouos as» c. nova—uc. a, en< cog: spec noe_~uu¢~ C O CO AN~..V .en.ev Ae.n.v Ame—v .em.mv Nm o.m ov. nm~.- «0.0- o.Pm woe ”a..- paco.u=u.uncmxpau_mo_oou .o O C. l Am.n_v Aomfiv Amm.nv em o.e em. e.a~ owe oo.o- beeeeoeees .m . Loeco uc~5oeamooe Km o.~ m_. we. zoo—ocnuou .muaac~ .~ uuweumwo carom CC. . I . Ae~._v ane.~v ~.., Am.npv A..~o. Aa.a_. .m o.—_ on. -.m om.v _.¢n P.o~ a.—- o.o pacovuau_unc_\_oo_oo_09u .o C an.m_v Ao.m~v Akoflv an.-v . um o.~ we. a 0.“, ~.qn Pm_- o.~_. acesooacoz .m I A~__v Am.-v Locus unusacamowe em m.~ em. ~m_- n.2e. ue. A8.253 .muaee~ .e uuwcum_o _zmQL i.iii a» o m .e.e o.sea pp< e a u g< em a em xe .enu . a a N . “no“ “coexo.qem no»: new: 20 unoco>wu Lounges osprey: ooau—up< uuoam< gu_ou3 pacaupau.co< yo oo< co.uau=uu -uogoocn vacw ea mo< vogoucu “on o—noveo> i. mu_um,poum AcoEEJm uuco_u_.eoou vou~.uomm< we. voc.~uo¢ n:o*L~> voacvu:09ii~.w uam<~ 164 representative, however, in contrast to those for Moshi District. These results need special care in interpretation.1 As can be seen from Table 8.2, the technical variables explain only 15 percent of the variation in six month milk production. The coefficient for water is positive but significant only at the .29 level. Those fOr the proportion of the fbrage ration coming from bananas and planted grasses are negative, as in the Moshi model, but significant only at the .25 and .11 levels respectively. The frequen- cy of grazing is again negative but unlike Moshi District, highly sig- nificant (.01). The frequency of feeding salt (Sf), kilograms of grain fed (6“), cash expenditures on grass (C9) and number of births (8") all had positive coefficients but did not increase adjusted R2 when added to a model containing the retained variables. Age at weaning (Aw) had a negative sign, as in the Moshi model but it too did not increase adjusted R2. The signs for the enumerator variable (En) and labor fer gathering grass were reversed from the Moshi model but the low level of confidence associated with the coefficients for 1Because of the high incidence of recent arrivals in the grade cow population of Rombo District we had fewer six month lactation histories. In order to increase the number of observations all house- holds in the Rombo regression models were given an equal weight of one instead of being weighting down to overall population probabilities. This had the effect of increasing sample size from 4 to 62. Since the Rombo sample is a two stage sample drawing ten households in or near each of ten first stage sample units, assigning each observation a weight of one does not correct for the lower efficiency of a two stage sample relative to a simple random sample of the same size. Conse- quently standard errors are underestimated and significance levels based on them are artificially low. Moreover the reader is reminded that more than three-quarters of all-grade cattle in the Rombo sample were selected purposively rather than randomly. All these factors suggest the results fbr Rombo be treated with caution and taken as indicative of orders of magnitude only, with an additional need fOr considerable interpretation. l65 these variables indicate there is nothing contradictory here from a statistical point of view. The poor explanatory power of the technical set of variables fOr Rombo District reflects the very recent nature of the grade cat- tle industry there. A much higher proportion of farmers have animals acquired from outside the area for which perfbrmance variables could not be objectively verified by purchasing farmers as well as they could from a local animal. As a result, large amounts of money have been spent fbr infertile and low producing animals-obviously not intentionally. These results point out that wealthy progressive farmers and early adopters expose themselves to substantial risk of financial loss. Thus part of the higher returns realized by those who are successful represents a necessary risk premium rather than monopoly profit. Not surprisingly, the management set of variables explains much more of the variation in milk yields in Rombo District than in Moshi District. Their addition to the technical set raises R2 to .34 an absolute increase almost three times as large as fOr Moshi District. The agricultural employment dummy is positive and highly significant (.01), indicating either a better ability to select or more likely, an inside track to infbrmation on which animals are better producers. Since animals financed by KNCU usually arrive in groups, persons familiar with veterinarians and others involved in the actual purchase and distribution of the animals are in a good position to get the better animals. Indeed such favoritism is one of the most common com- plaints of farmers in Kilimanjaro. In Rombo, unlike in Moshi District, l66 the small number of locally born grade cows in relation to the total allows the data to reveal this favoritism. The influence of wealth (H) is positive and significant at the .04 level in Rombo District while the coefficient for the age of the head of the household (Ah) is negative and significant at the .09 level. Recalling that (Ah) was positive in Moshi District the likely explanation fOr the switch in signs arises from the inclusion of wealth in the Rombo model. In Moshi the age of the head of the household is probably picking up some of the effect of wealth. The major change in the management model is in the sign of the coefficient for the proportion of the ration coming from bananas, from negative to positive. This variable became considerably less negative in the Moshi model when the management set of variables were added and positive as well when the age of the head of household was indluded. This reflects the fact that traditionally bananas have been an important source of livestock roughage and the lower yields associated with feeding bananas reflect lower yields associated with older farmers. Whether these lower yields arise from excessive depen- dence on bananas pgr_§g_or simply less human energy available fOr gathering more distant sources of roughage is a question the data can- not address. Addition of aspect and altitude, the two ecological/institution- al variables, increases R2 proportionately about as much as in the Moshi model but still leaves overall explained variation at about one-half the level for Moshi. The negative value fbr aspect and alti- tude reflects the high concentration of grade cattle in Usseri and Tarakiya and the presence there of an extension agent who is active 167 and well respected by farmers. He has been very successful in helping farmers acquire good breeding stock from Kenya and West Kilimanjaro. There is no A.I. program in Rombo to speak of so this set of variables gives an indication of the potential benefits of a well-organized and well-staffed extension program. D. Implied Marginal Productivity of Selected Inputs Only the data relating to grade cows in Moshi District reflect enough stability and explanatory power to be used to estimate the marginal productivity of selected inputs into the grade dairy enter- prise. Table 8.3 compares the average value of the dependent vari- able and each of the independent technical variables far all-grade 1 Based on the cows and the 25 percent top producing cows only. values of the estimated coefficients the increase projected for superior cows is 485 liters versus 680 liters actually found fbr these animals. Thus the technical variables explain slightly over 70 percent of the variation in milk yields between the two groups when calculated in this way versus the 60 percent found for the model itself when estimated on a much more limited set of data. The implied marginal productivities of the five technical vari- ables which involve a quantifiable expenditure are also given in Table 8.3. Since linear fbrms were used for each of these variables the theoretical characteristics of these marginal productivities are not particularly sound but they nonetheless shed some light on the profitability of selected inputs. 1Includes all-grade cows in the composite grade cattle sample for both Moshi and Rombo Districts. 683.... a... . ...¢ 65...... so. .2...- ¢.c .52... .0 ...:325- .33 .8 3:3 3 .- .....) on .8 ts... ad. .48.. ... :3... £9.23: 3.... .0... ..I 333 3 a. :3 on. so. ...-too... 3.. £3. cl ...-5:... ..3 a m...” .02—.... n... no: 3 9.3.1.9.. c. 33.5... 3:... 1» ..s3 .3 :3 8.... .2. ..3 3 33.2.8... .022: .3... .... 5...... 3. .- 2.3.. : .... ...... 2...: fl. 8. 952 2.. ...... .... .838... ......IECI s: :3 .39-Ia... so. .82.?!) 33.8.... a. 533...: .o 2......- n.. ...... 2...... :8... . , .5: .... 8.. .e 2...: .... ......o .33.... It»... .o 3.2.. .o 3.31 3.5.3. 3.3-...... 3. no... 33.»... 2.03.23... 39.. 3.3.2. .29.: 2.. ... 2.3 =- 83.8.. .230 ......uua .51 50. sauce-=26. 168 ...: 3......- .3: .... .... .... .... e .2. N... .351... .. .... .... .... . .... 8.» 3.. ...? .33! .. .2 2 .... .... .... ..8 3.. .... .... 2:... .. .35.. .. .... .... .... ..- 3... 3.. .... as... 2...... .. .... ...e a... ..z. 8.. 8.. .3- . .52... .. .... J... .... ...: ... :. ...... ......3 .. —.— on. .9... ed? ... u... 2.. .3...- so. s3... 0.. ... a: a. ...: 8. 3 .... ....z. 2.2:... .u 9. ... ... 18 8.. 3. 2. ...... to .... 8. ... ..8 ...~. .... .... ...: x 9... 2 ... .... ...: ... .... .5... .. ...: .8 ......33: .... to tunnmua «2.3.... .1... ...”..de H... ”gown“. .3.......:.. .35.: :3. o .5... .2835... among“... ......r: .e S... 2.3:. 3...... 3...... .. 8......» I i i 3H”... i asp—anaeuu 3 :5 cu. «nan—Zuflag 45.5.5. 335: a... a: 5:8 9,83 2.» no 3.?» E 35.: 33¢ at» .3 ~80 3:33.? 2.» :5: 5:53 3!! .2 § 83882; 5.: u..— 5 «3.3: 3.2.0: .- Q’Suu 3 3 83“ s “an: an)»: 3:; ...—x 3. ~33..— n.. g 169 The marginal productivity of expenditures on water and salt are too high to be credible given the rather low rates of utilization of each in the average herd. No doubt coefficients for both variables pick up some non-causation correlation between the frequency of these practices and high milk yields. The figures for grain and cash and labor expenditures on grass are much more believable. Cash expenditures on grain and grass yields around two shillings fOr each shilling of expenditure. No doubt capital constraints and risk aversion account for the failure to use each at optimal levels, though farmers appear to be operating on a line of least cost combination with respect to the two. Labor for gathering grass, a relatively abundant input in Kilimanjaro. appears to yield returns only slightly above its oppor- tunity cost. Though the significance level of the coefficient is weak (.28) the implied return is quite believable since milk is only one of three principal outputs of the dairy enterprise, the other two- being meat and manure. This input has the greatest impact on produc- tion of manure by far and inclusion of the value of manure in the dependent variable would raise the MVP of this input relative to the others. The purpose of this exercise is not to estimate a production function as much as it is to explain variation in milk yields. Returns to labor and capital fOr the entire enterprise are computed in a different way in a later section. E. Summary of Determinants of Milk Yield Very little of the variation in milk yields among zebu cattle can be explained by either technical or management variables. The 170 two technical variables which were significantly different from zero at the .01 level explain only 3 percent of the difference in average milk production between average and superior (top 25 percent) zebu cows. For grade cows the results were more as expected, especially for Moshi District where the grade dairy industry is more established. Technical variables kept in the model, exclusive of enumerator bias, explain almost 60 percent of the variation in average milk yields. The coefficients for these variables embody at least some increase more properly attributable to management, especially as management relates to selecting cows capable of superior perfbrmance. But in general, the data suggest that feeding water, salt, grain and pur- chased grass all have a substantial economically positive impact on milk production from grade cows. The return per unit of expenditure on labor for gathering grass and for purchased grass, coupled with the negative coefficient fOr the proportion of the ration from bananas, suggest that the quantity of forage given to animals is constraining milk production in Kilimanjaro. Thus, efforts aimed at reducing forage constraints will have a positive impact on milk production. As expected milk production by grade cows increases after the first lactation but the relationship is not particularly strong in the data used in this analysis. The regression analysis also confirms the importance of insti- tutional variables related to the distribution of extension and arti- ficial insemination services and management variables related both to progressiveness and wealth as well as special interest politics. 171 The grade cattle data confirm that variables over which a well- organized extension and artificial insemination program can have some influence explain at least 70 percent of the variation in six month milk production for grade cows in Kilimanjaro. No doubt much of the unexplained variation arises from genetic differences in grade animals that the A.I. services could address as well. Just how much of this potential can be realized will depend on organizational and implementation issues beyond the scope of this study. We turn now to an examination of the costs and returns of cattle/dairy enterprises in Kilimanjaro. CHAPTER IX ENTERPRISE ACCOUNTS AND INCENTIVES FOR UPGRADING THE DAIRY ENTERPRISE A. Budgeting Techniques In this chapter simulated capital budgets are prepared and analyzed for eight different types and sizes of cattle enterprises over two different planning horizons. The inputs and outputs used fbr each of the capital budgets are derived from average values fOr relevant coefficients or from values which themselves are derived from or otherwise based on those average values. These budgets then serve as the basis for a discussion of the profitability of the vari- ous alternatives. The use of simulated results to analyze the dairy industry in Kilimanjaro was necessitated by the fact that many of the herd per- formance parameters necessary for calculating the profitability of the cattle/dairy enterprise could not be obtained on an individual herd basis. Age specific mortality rates, calving interval, age at weaning and lactation milk production are only a few of the variables that had to be calculated on a herd basis. This is partly because far an individual household or cow the last occurrence of the event is not a particularly good predictor of the most recent as yet uncom- pleted occurrence, and partly because many animals, especially grade animals, were recently acquired and had not yet completed a complete reproductive cycle with the farmer now owning it. At the same time 172 173 the flow of benefits of a cattle enterprise cover many years and depend on a number of chance events such as the sex of the calf and the genetic potential of grade animals purchased in an imperfect market. Further problems in calculating profitability on an individ- ual herd basis arise from the long calving interval and the low per- centage of cows actually in milk at the time of the survey, as well as the fact that 1973 was a period of rapidly rising zebu cattle 1 and grade cattle prices in Kenya. prices in Northern Tanzania Finally, the single visit method of data collection, with interviews of various households distributed over time, yields population esti- mates that are more independent of seasonal variation than the indi- vidual household observations. The enterprise budgets are thus synthetic budgets. Using syn- thetic budgets of this kind helps overcome unknowns and anomalies in the individual herd data and provides more representative figures for the entire year. "here the distribution of the values of particular coefficients in the sample population is fairly continuous with a clear central tendency this approach poses no problems in interpreta- tion. However where distributions are more dispersed and no central tendency is clear, such as with lactation milk production fbr grade cows, a question always arises as to the usefulness of the results far predicting farmer behavior. 1Large numbers of cattle were being purchased by Asians and smuggled into Kenya as'a way of exporting capital. In 1973 the Tanzanian Government nationaliZed all rental properties with a value in excess of 100,000 shillings ($14,000) with no compensation to be paid on buildings more than ten years old. This was the last straw fer many Asians who sought to export as much of their capital as pos- sible befbre leaving themselves. 174 8. Summary of Procedures and Coefficients Two principles guided the selection of the specific enterprise size/type combinations which are simulated and analyzed in this chap- ter: the enterprise must represent central tendencies fOr large groups of farmers or it must represent a reasonably accessible target for existing producers, given the quantity of labor required. There are five basic enterprise size/types. In addition, three of these basic combinations are constrained by the requirement of a minimum proportion of males in the herd in order to assure breeding services. Overall, the eight enterprise size/types which are analyzed include six that are 1.6 cattle units in size and two which are 2.2 units. The 1.6 unit enterprises require roughly the same amount of labor that is currently expended on the average all-zebu herd in the same area, the size of which also averages l.6 cattle units. The 2.2 unit enter- prises reflect the average size of all-grade cattle herds in Kiliman- jaro in 1973 and are included to get an idea of the returns to existing grade cattle owners, even though the amount of labor and capital they utilize is considerably greater than that which could be brought to bear by the average household without offsetting changes in the farming system and in expenditure patterns. All of the capital budgets for the various enterprises assume the purchase in year 0 of a heifer twelve months younger than the average age at first calving for the particular type of heifer in question. Subsequent calves are born at the appropriate average calving interval fbr each type of cow until such time as the cow is removed from the herd. The age specific mortality rates as detailed in Tables 4.5 and 4.6 and summarized in Table 9.1 are used to 175 TABLE 9.1 COEFFICIENTS ASSUMED FOR THE ENTERPRISE BUDGETS Enterprise and Type of Cow Coefficient Aézbgge Zebu-Grade All-Grade F1 F2 Average Superior a. Average size of enterprise in cattle units 1.6 1.6 1.6 1.6 2.2 1.6 5. Age in months of heifer at a purchase 37 na na 22 22 22 c Age at first calving in months 49 42 34 34 34 34 d Calving interval in months 29 25 20 20 20 17 e. Lactation length (months) 8 9 10 10 10 11 f Production period (months) 11 12 13 13 13 14 9. Milk production per lactation (liters) Overa11 380 750_ 1100 1470 1470 2220 First lactation 380 750 1060b 1350‘ 1350‘ 2040‘ Second lactation 380 750 1100‘ 1470‘ 1470‘ 2220‘ Third and subsequent lactation 380 750 1140b 1590‘ 1590‘ 2400‘ h. Number of cattle units per full grown adult animal 1.0 1.15 l 3 1.3 1.3 1.3 1. Labor inputs (hours per month per cattle unit Nonmal maintenance and growth 67.1 69.3 70.6 70.6 58.9 70.6 Production period 14.3 17.3 19.5 22.5 19.9 29.9 Mortality rates Male under 1 year .24 .22 .20 .20 .20 .20 1122 years .21 16 .10 .10 .10 .10 2-<3 years .10 .10 .10 .10 .10 .10 3 years and older .10 .10 .10 .10 .10 .10 Female 1 under 1 year .13 .10 .07 .07 .07 .07 l-<2 years .10 .09 .07 .07 .07 .07 2-<3 years .10 .08 .05 .05 .05 .06 3 years and older .10 .08 .06 .06 .0 .06 aMot applicable. bAssuming that increases in milk production for second and third lactations for f2 zebu- grade crosses are one half as great, for grade cows in Table 8.3. in percentage terms, as the increases indicated cAssuming increases in milk production for second and third lactations are proportional to those found for six month milk production and reported in Table 8.3. 176 compute interdependent survival probabilities for the initial heifer and all subsequent offspring, according to age and sex. With the exception of a zebu enterprise in which the first calf is assumed to be a male, internal reproduction is the only way the herd is allowed to grow. The number of animals (or fraction thereof) in the herd rises with new births while both the probability of a birth and the survival probability of individual animals in the herd declines according to mortality probabilities. By year 10, the last year for which herd performance is simulated, some of the herds will have contained as many as 12 animals with survival probabilities of 40 percent or less. In no year is the size of the herd-—adjusted f0r survival proba- bilities-—allowed to grow beyond a specified maximum permissible size. This limit ensures that the average size of the enterprise between years three and nine-essentially years of full capacity-—does not greatly exceed its average size in the population. It also causes substantial fluctuations in herd size from year to year as animals are removed from the herd. Hhen cattle need to be removed from the herd they are generally sold in the middle of the first year during which the herd size limit would otherwise be exceeded. Animals selected f0r sale are those adding the lowest increment to total value over the next 12 months. In general this causes male animals to be removed first. In three of the enterprises, however, a minimum proportion of males is required even if retaining a female would add a greater increment to total output. 177 If any zebu or f1 cows are removed while still in milk their calves are sold with them since the regression analysis suggests that these animals stop letting down their milk once the nursing stimulus is removed. In those cases where the reduction in value added would be less by selling a nursing calf instead of an adult animal, the calf is valued as a male, regardless of its sex. This presumes such animals are sold for slaughter and reflects the very high mortality rate for calves away from their mothers. All male animals are sold within six months of reaching their maximum value whether or not the maximum permissible herd size con- straint has been exceeded. This gives farmers the service of sexual- ly mature males for about two years while limiting their presence in the herd to the period of increasing weight and value. The herd structure data indicate that this is the practice fellowed by most farmers. Animals which are sold are valued using the age-sex specific average prices included in Appendix E. In addition, females are assumed to capitalize one-quarter of the value of a newborn calf if sold when three to six months pregnant and one-half of the value of the calf if sold over six months pregnant. Cows sold within three months of parturition are assumed to capitalize 25 percent of the value of milk production expected fer the fellowing lactation, in addition to one-half of the value of the unborn calf. Animals which die are assumed to generate income equal to their 1 mortality probability times 50 percent of their meat value at the 1Meat value is defined as the sale value of the male animal or if a female animal, the lower of the value of the female animal or a male animal of equal age. 178 mid-point of the period covered (much of this meat is consumed by the household or retailed at a discount to neighbors). At the same time, variable costs and variable returns such as milk and manure are counted only to the extent indicated by the survival probabilities. Labor inputs per cattle unit are those detailed in Table 7.4 for normal growth and maintenance and production, converted to a monthly basis. Production inputs are included only over the actual production period for lactating cows, defined as the last three months of pregnancy and the entire lactation period. Each hour of labor is valued at .43 shillings, its average sex specific opportu- nity cost over the entire year as described in Appendix H. All reproduction coefficients and labor hours assumed for the various enterprise types are summarized in Table 9.1. Variable cash costs, like labor expenses, are divided into those fer maintenance and normal growth and those associated with milk production. Since most grain is fed either to cows or calves, all grain and milk replacer cash expenses as reported in Table 7.5 are assumed to be milk production related expenses and are charged entirely to the production period. The breakdown of variable labor and cash inputs are summarized in Table 9.2 with the details of their derivation explained in the f00tnotes to the table. Table 9.2 includes also the value of manure from maintenance and growth converted to a monthly basis on the assumption that pro- duction is spread evenly over the relevant production period. Manure has been valued at 75/8 per ton with annual manure production amounting to 1200 kilograms per 100 kilograms of body weight. Cattle unit values are used to approximate weights and one cattle unit 179 TADLE 9.2 sumanv OF THE c051 0F VARIABLE iNPUlS AND THE VALUE or VARIABLE ouwu15 PER CATTLE un11 P19. HONTH FOR vmous 512E5 or ALL-ZEBU. ZEBU-GRADE AND ALL-GRADE CATTLE NERDS IN KILIMANJARO UNDER ALTERNATIVE LEVELS or MANAGEMENT (11 SHILLINGS) “ ‘ Level by Enterprise . lebu Zebu-(rade A 1 Grade Input/Variable Average F1 F2 Average Superior Average number of cattle units in enterprise 1.6 1.6 l.6 1.6 2.2 1.6 Cost of variable inputs (shillings) for maintenance and normal growth 1 Labor‘ b 29 30 30 30 25 30 Variable cash inputs 1 2‘ 2‘ 6 6 28 For pregnancy and milk production Labor‘ 6 7 a 10 9 13 Variable cash inputsd o 1 1 10 1o 22 Value of variable outputs From maintenance and growth: mime 21 21 21 21 21 21 From production: 111111'. First lactation 71 125 159 203 203 278 Second lactation 125 165 221 221 303 Third lactation 71 125 171 239 239 327 .Values in Table 9.1 multiplied by 11.43 to get mothly values in shillings. I’Ilata taken from Table 7.5 and reduced to lllO'lthly basis. Variable cash inputs for maintenance and growth do not include grain and milk replacer since these are fed to cows and calves only. USually while the cow is producing milk. ConseQuently they are treated as a production period expense. cAssuming all increase in cattle unit cash inputs over the levels for zebu cattle are expended on sgrade cattle in these herds and assuming further that grade cattle make up 50 percen: of the her nc udes grain and milk replacer costs from Table 7. 5, reduced to per month of production period basis for cows. Thisi nvolves multiplying the annual expenditure for these items fmro Table 7. S by the following term calving interval 12 x , period it proportion of cattle units which are cows This reduces these annual expenditures to a monthly production period basis for cows only. Valies used for each of these variables are listed in Appendix Table 0-1. suming one cattle unit weighs 275 kilos. all cattle produce 1.2 tons of manure per ye." per 100 kilograms of body weight and manure has an economic value of 75 shillings per ton (see App-man fAverage monthly milk production 9 1/50 per liter. 180 produces 3.3 tons of manure having a total value in crop production of 248/= per year. Appendix I explains how the quantities and value of manure production were derived. Lactation milk production is based on average production fer each enterprise type as reported in Table 6.5 and summarized in Table 9.1. Milk yields fer the 1.6 unit superior managed herd reflect average grade cow lactation milk production plus the increase in production explained by technical variables in the milk production model fer grade cows as reported in Table 8.3. All milk production is assumed to be spread evenly over the entire lactation period. The monthly value of milk production for the different types of cows is also given in Table 9.2. Milk has been valued at 1/50 per liter, slightly below its weighted average sale value of 1/57 based on 1/65 per liter for fresh milk and 1/43 per liter for sour milk quantities actually sold. The 1/50 figure reflects the fact that most milk consumed on the farm is consumed in the cheaper fermented milk f0rm (69 percent).1 Investment costs for the five groups of enterprises are detailed in Table 9.3. Each herd begins with sufficient housing to cover the maximum size the herd will attain before being forced to liquidate an animal in order to stay within its respective labor constraint. 1The actual average price weighted by the 1973 quantities of sweet and fermented sales and on-farm consumption comes to 1/51 per liter. Given the amount of unmet demand for milk at current prices there is little justification for budgeting a decline in the relative value of milk over the ten year period fer which the enterprises are compared. However, problems could arise in such places as Machame and Marangu if government policies continue to favor concentration of pro- duction for accumulating surpluses rather than dispersal of production to maximize on-farm consumption as well as the value of production. 18] TABLE 9.3 SUMMARY OF INITIAL CAPITAL INVESTMENT REQUIRED TO ESTABLISH ONE CON ZEBU AND GRADE CATTLE ENTERPRISES IN KILIMANJARO (TZ SHILLINGS) Enterprise and Management Level Investment Zebu- All-Grade Zebu Grade Average Superior Herd size in equilibrium (C.U.) 1.6 1.6 1.6 2.2 1.6 Maximum permissible herdsize (C.U.) in enterprise sinxflations 1.9 1.9 1.9 2.5 1.9 Initial purchase of cattlea 37 month old heifer 360 360 --- --- --- 22 month old heifer 1100 1100 1210 Housi1 9b 450 510 570 1010 67’.) Other purchased itemsc 14 34 51 100 Total initial capital investment 810 984 1804 2161 1980 aTaken from Appendix E. bTaken from Table 7.9 a55uming two C.U. capacity for 1.6 unit enterprises and three C.U. capacity for 2.2 unit enterprises. cTaken from Table 7.10 assuming as 'n b. Average figures are used on all management levels except superior. A full complement of sprayer and utensils has been assumed for the latter. 182 Housing values per cattle unit are those summarized in Table 7.9 and are fully depreciated over 10 years. The cost of zebu and average grade heifers are taken from Appendix Table E.l. The cost of grade heifers in superior herds has been increased by 10 percent in line with the higher average prices farmers reported paying fer breeding females of this type. Table 9.3 details the actual costs assumed for the initial purchase of heifers. The cost of other capital inputs is taken from Table 7.10 using the maximum size of each of the enterprises as a basis fer establish- ing the respective costs and assuming a life of five years with zero salvage value at the end of that period. Fer superior herds, a full complement of sprayer and utensils has been assumed while those for lower levels of management reflect average values as obtained from the respective samples. Though no farm will have one-half of a sprayer, and in this sense the costs assumed are unrealistic, the effect of spraying as measured in mortality rates and milk production are average effects across users and nonusers. It is appropriate, therefore, to reduce costs to a similar basis. Since the profitability of the cattle enterprise-—especially the grade cattle enterprises-—is greatly influenced by the sex of the calf, separate simulations were done according to the sex of the first calf. Subsequent calves then alternate sex and the sex of the first calf of the first female offspring is the opposite of the sex of the first calf of its mother. These two simulations are then averaged to come up with the com- putational parameters used in the enterprise capital budgets. The procedure gives an expected outcome across all farms on the assumption 183 that the probability of getting a male and female calf is equal. Appendix J contains details on the simulations fer one of the enter- prises to give the reader a clearer idea of the steps involved and the procedures used. C. Enterprise Types and Results The following eight enterprise size/types are analyzed: 1. An all-zebu enterprise which grows through internal reproduction only. All males are grown out to 54 months befbre sale and cows are sold after the end of their last lactation preceding their twelfth birthday. No other animals are sold. Because of the rather high mortality rates and long calving interval fbr zebu cows this enterprise type is unable to attain the average herd size of 1.6 cattle units and averages only 1.39 units between years 3 and 9. This enterprise type represents subsistence farm- ers who have relatively little involvement with the cash economy apart from coffee sales. 2. A 1.6 unit all-zebu enterprise which adds one additional female heifer in year 4 for the case where the first calf of the initial heifer is a male. This forces sale of a two-month old male calf in year 6 and reduces the proportion of males in the herd from .23 feund under the first alternative to .16. This compares to .20 fer the entire Kilimanjaro zebu herd. This alternative still represents a primarily subsistence-oriented management system that relies on internal reproduction with the occasional addition of a young heifer to raise herd size to a level that more fully uti- lizes available labor resources. Data on transactions and herd 184 composition indicate that both management systems are commonly feund among zebu cattle producers. A 1.6 unit mixed zebu-grade enterprise that begins with a zebu heifer and crosses all subsequent offspring to a grade bull. Alternative three requires that a male over two years of age be present in the herd at least half of the time between years 3 and 9. Though this reduces profitability it ensures the availability of males fer breeding purposes. Reproduction rates fer the f1 and f2 are sufficiently high and mortality rates sufficiently low as to f0rce the sale of animals in order to stay within the maxi- mum herd size constraint. At the same time the herd is upgraded such that there is an f3-by Kilimanjaro standards a high grade animal-in the herd by year 10. Another l.6 unit mixed zebu-grade enterprise but one that retains males only to the extent to which they are increasing in value and are necessary to bring the herd to its maximum permissible size. Thus the ratio of females is as high as it can be under conditions of natural reproduction only. This enterprise pre- sumes the availability of breeding services from bulls held by other farmers or from artificial insemination. The proportion of male cattle units in the herd between years 3 and 9 is .15 as compared to .16 for the entire Kilimanjaro grade herd, .20 fer the zebu herd, and .24 for the previous mixed zebu-grade alterna- tive. Both mixed zebu-grade enterprise types represent systems which are common on the mountain, though at the present time the average size of mixed herds is 2.8 cattle units of which 1.3 units are zebu cattle and 1.5 are grade cattle. The 1.6 unit sizes 185 analyzed here represent what existing zebu farmers, representing about 57 percent of all farm households on the mountain, can aspire to without major changes in their labor and expenditure patterns. A 1.6 unit all-grade herd under average management. Because of the larger size of adult grade cows it will not be possible fer average zebu cattle owners to shift to grade cattle unless they can be assured of breeding services from elsewhere. One cow and one calf under six months represent l.6 cattle units. Thus except for raising one replacement heifer-—perhaps taxing avail- able labor resources fer one year and doing without a cow for another-—l.6 unit grade herds will be essentially milk producers. Feeder calves and very young heifers will be important by- products. A superior managed l.6 grade cattle enterprise. This alternative uses non-labor input-output relationships fOund fer the 25 percent top producing cows with respect to six month milk production. Labor inputs are those used for the 1.6 unit average grade herd with adjustments to reflect the increased need fer forage and labor for milking during the production period of these higher producing animals. It represents the best that a zebu cattle household can aspire to over the next ten years and then only with a very substantial increase in cash inputs and more s0phisticated management than it uses at present. A 2.2 unit all-grade cattle herd under average management with a two-year old male in the herd at least half of the time between years three and nine. Maintaining a male in the herd is easier 186 and less costly in this larger herd but it does reduce returns over what would be possible if reliable bull services were avail- able elsewhere. The 24 percent of the three to nine year period cattle units in this herd which are male compare with 16 percent in the entire Kilimanjaro herd. Thus this represents what we might call a conservative management strategy fbr existing grade cattle owners. 8. A 2.2 unit grade cattle enterprise under average management with no requirements for males in the herd. This alternative results in only 10 percent of the cattle units in the herd between years three and nine being male, clearly an aggressive management strategy in relation to the population average and one that is only possible in the context of a smoothly functioning artificial insemination program or a friendly, more risk averse neighbor. Each of these eight alternatives are manually simulated fer ten years. The results of the simulations, when inputs and outputs are appropriately priced, summed and discounted, gives a measure of the returns to new investments in each of the selected enterprise types. These measures are necessarily rough because of uncontrollable fluc- tuations in herd size from year to year and actual average herd size over the period, which though similar, are not exactly the same. Because of the time it takes to grow from a single heifer to a 1.6 unit herd the ten year results underestimate the returns to on- going enterprises. To correct fer this, a second set of computations involving only years three through ten was made. For this exercise all animals in the herd in the middle of year three were assumed to have been purchased and their cost treated as a capital cost in the 187 first year of a seven year budget. Cattle unit months of life, cattle unit months of production period and months of milk production were then recalculated for year zero and all other inputs and outputs-— with the exception of the salvage value of undepreciated capital inputs-—remained the same. The results of these shorter capital budgets then reflect the returns to a fully functioning, ongoing enterprise fer each respective size/type examined. Additional details and the ten year summary capital budgets for each of the eight enterprises are given in Appendix K. D. Incentive for Upgrading the Dairy Enterprise Table 9.4 summarizes the results of the simulated capital budgets and gives the gross benefit cost ratios and the internal rates of return for each one. It also shows the distribution of undiscounted returns between milk, manure and meat, the latter. including sales of live animals, salvage of dead animals and the increase in the value of livestock holdings over the period con- sidered.‘ Turning first to the all-zebu enterprises we note gross benefit] cost ratio's of below .9 f0r both types and both time periods. The similarity of results between the two types and the two time periods reflects the fact that returns from meat and manure over the growth phase are almost as high as the returns from milk as long as male animals are still in their growth phase. Shifting the composition of the herd toward fewer males does not appreciably alter the overall 1The increase in the value of livestock holdings includes the future value of milk production capitalized into the value of females sold during their production period. 188 .33... ...... ...... ...... 3.39.8. . 3 ...... 0.2.. move... ...»...E.... .... .... ...-...... 3.3 ......»e. ‘e. 8... 03...... .83... ...... .... ...)... ....2..... ... ...... 4.2.3.3.... ...... 8 3...... $3.... .... .... ...-.... 3...... 3...... . .. .... 3... .5 ...... 2...... .... ...... 33...... ...... .... ...... .2... ......3- a. ......I .. vs... .... .. ...... ...... .. ...-.... .... ...... ...» .... 2. 9:3... £2.33... ...... e. 3...... $3.... .... ... ...... 3...... e... .... e. 2...... ... .....- 33.x. .. .22.. 2.2.... e. .... .... 8 3. .2 ...... c ... 0...... .....u 8. .....6 a... 8.3... :9... ......6 o... 8.. .... .... ... .... 2.3:... . ...: 0.0.... zuczozoaa ....) 3...... ......a: €3.33 0.! 9. .3..o.ao.~ “...! . . . .. ... .... .c..._a.. . o. on u v .n v n n ...: :9... .3533...» ... ... 3...... 95...... 62.3... ...a .8255... ...... ... S. .... .... z. 2. ... .... .... ... 2. ... 9. 9. 8:7... .- 2. z, .... .... ... ... ... .... 3.. ... ... S. ... a. 82...: .. ...g..eu ...5 ~.~ 3. 0., «.3 2.. ... 3. 2. a... .n.. 2. 2. 2. 0.. 8. 3.3.3.3..» .o 3. 8, ..on 3.. n.. 2. 3. ..~. 3.. ... nu. co. 9. 8 3.3.24 .m on. .... ..- 3.. an. a... .... a... 3. an. .o. .... 9. 1 3.5.3... «...... .. on. 2 9o. 3.. e... on. on. ~.: «0. 3. ... an. .0» at 8.5.5... 9...... .n on. ... ... an. .... .... ~n. ... on. n.. 3. an. ..9. ... 3.....4 .~ on. 2 0.. on. ... 3. ...: 0.. ... 2. a... 2.. u... 3 ...-....c .. 3:93... 2:8... .3. 3.1.1 s: :- ...: :25. ...: ...: 2. .2: 225. ...: a. m. ... a... .... 3...... ......3... .....o: 31.0.; 00.. 9......4 5.8. .902... mm..... 9.3.: W N Wm. .. ...... ... e. 3...: ...... .9 3.3.0.. 2...“.- u..§3..v§ .o .33.! 2......- 9..§3:.:.: u u. ”m 3...: ......u .o .o co...ogo& .o «3:80.. .o c0239... 9 a n .35.. ......4 .3233... 2:88 3.35:»... ...: 1 3:3 gaze «3:553. 22.5.... .53.... o... .33 3.33:. 32.3.... ...... :o: .2 :5... .... 5:... 22.5.... 9:8... .35.»... .o 39.5.... 9... "3.5.5.... :3: a... .35.. \ .. 2:5 3:3 3 3:38.. .... o... 2...... .5... to: 2.2.. 2.5... 8.52:. 3.58293 .. 8:88.. v.0 “...: 189 returns per unit of expenditure. Increasing the size of the herd does, however, increase total returns to the enterprise. The low rate of return explains why the zebu herd has been contracting in recent years, though one might expect a more rapid rate of contraction than that which has been occurring if the numbers are correct. It is possible that the value of manure on cereals underestimates its value on bananas. It is also possible that the opportunity cost of labor during the dry season is somewhat overesti- mated. Though this presents a problem in comparing returns to resources invested in cattle to returns to those resources invested in coffee and other crops competing for farm resources, it does not present serious problems for the analysis of cattle enterprises since all enterprises use very similar amounts of labor and produce similar amounts of manure. It is also possible that actual labor expenditures fbr gathering grass are overstated since much of the time spent carry- ing grass during the maize cultivating season doubles with time spent walking to and from the maize fields. A second noteworthy aspect of the all-zebu herd is the importance of manure in the total benefit stream, accounting fOr nearly 55 percent of total benefits. Though valuing this output is necessarily diffi- cult, the order of magnitude suggests that stall feeding zebu cattle probably makes most sense where land pressures fOrce intensification causing measures to maintain soil fertility to become important. The two mixed zebu-grade enterprises come reasonably close to yielding a return equal to the estimated opportunity cost of the resources employed, ranging between 17 and 22 percent on invested capital. For an ongoing herd, the usual case in Kilimanjaro, the 190 upgrading enterprise can even yield a positive net present value if farmers move to eliminate males from the herd and rely on others fbr breeding services. Currently, mixed herds are much larger than 1.6 vunits and many of these farmers do have a sexually mature male. But before smaller farmers can begin upgrading with confidence, a more reliable supply of grade cattle semen will have to be assured. If one assumes that unmeasured returns or utilities account for the large number of all-zebu herds in the population in spite of the relatively low rates of return indicated by the budgets, then up- grading the zebu enterprise becomes a very attractive alternative indeed. In terms of changes in the flow of costs and returns between the mixed and the all-zebu enterprises the additional costs and returns associated with the mixed enterprises though small in abso- lute amount, yield internal rates of return in excess of 34 percent on the additional investment. More importantly, the benefit cost ratios are well over two and the highest of all the enterprises studied} One would expect, therefore, a fairly keen interest in upgrading. This interest is readily apparent in Kilimanjaro where the most fre- quent complaint voiced by non-grade cattle owning farmers against the artificial insemination service is its failure to respond to requests for service from farmers with only zebu cows. Moreover, as the budgets in Appendix K indicate, the rather small increase in opera- ting costs required fbr the upgrading enterprise is more than offset 1The gross benefit cost ratio is probably a better measure of investment worth for the average farmer in Kilimanjaro since labor is the principal constraint on expanding the stall feeding enterprise and labor represents over 80 percent of total costs associated with the all-zebu and the mixed zebu-grade alternatives. l9] by an increase in the value of livestock holdings. Thus farmers investing in this alternative have, in effect, an insurance reserve that grows along with their operating investment. The 1.6 unit all-grade alternatives, effectively possible fOr the average farmer only if he can rely on an artificial insemination service or another farmer's bull fOr breeding his cow, are, from the point of view of returns to capital, the most profitable of all cattle enterprises l.6 units in size. On the basis of the benefit cost ratios, however. the return to total resources is much lower than upgrading an existing zebu enterprise. In fact, given the impact of the non-continuous nature of the data on the actual results attained, there is probably very little difference in the return per unit of capital between these two sets of alternatives. Investment for the all-grade alternatives is considerably higher when beginning from scratch and the proportion of returns coming from manure is much smaller than for the mixed enterprises. Grade cattle are, as a result, a much more suitable investment fbr school teachers and other households with important non-agricultural sources of income who may not be able to realize the full value of the manure. Fbr such households capital would most likely be the most important constraint, especially for purchasing grass and operating inputs, and the internal rate of return no doubt gives a good measure of investment worth. On that basis a l.6 unit grade dairy enterprise, with an internal rate of return of between 40 and 55 percent, is attractive indeed. Most surprising of all the findings of the survey is the fact that superior managed herds with 44-56 percent internal rates of return do not yield returns on capital that are measurably higher 192 than those available from average management. In terms of total resources, in fact, returns per unit are lower as evidenced by the lower benefit cost ratios. However the fact that the superior herds use many more capital resources and earn a very high return on them is no doubt a very powerful incentive fbr using improved management practices among households with large cash reserves. As would be expected, the 2.2 unit grade enterprise, with no male required, yields returns noticeably higher than the l.6 unit herd, especially using benefit cost ratios as a measure of investment worth. This is because labor costs per unit are lower in the larger herd. Interestingly, the 2.2 unit herd with a male required, with an internal rate of return of 42-45 percent gives returns very simi- lar to those of the l.6 unit enterprise where no males were required. This suggests that an efficient artificial insemination service available to all farmers may have a substantial positive equity impact in addition to raising aggregate incomes. E. Potential of the Various Alternatives for Increasing Milk Production It is clear from the very weak linkages between management prac- tices and milk production among the zebu cattle herd found in Chapter VIII, coupled with the relatively low returns to the resources employed in such enterprises, that zebu cattle alone do not provide a viable alternative for increasing milk production in Kilimanjaro. It is also quite obvious that the increasing difficulty of obtaining grade cattle from Kenya and the very high prices and associated high risk which they present to farmers will continue to constrain increases in milk production from pure grade animals. Moreover, even l93 if such animals were available in larger numbers and credit was forthcoming, past history suggests that it would be only a few wealthy farmers who would benefit. There is little doubt that upgrading the zebu herd presents the most viable alternative for rapidly expanding milk production in Kilimanjaro. Over 60 percent of all households in the survey have zebu cattle at a given point in time. Current levels of management are not adequate to assure the survival and efficient exploitation of high grade cattle. But farmers upgrading their herds do show evi- dence of changing their management practices. Moreover, the return to the additional resources required for upgrading is very high and exceeds that of the all-grade enterprise on a total resource basis. Upgrading also eliminates the economic justification for excluding small farmers from grade dairying because of their limited access to additional resources. The principal constraint on expanding mixed zebu-grade enter- prises is the lack of sufficient numbers of grade bulls or a viable artificial insemination service that meets the needs of the mass of small zebu-owning households. Farmers are using the grade bulls that are available but these are limited in number and frequently are low in quality. Yet, every year large numbers of high grade bulls die or are slaughtered to conserve milk for household use and/or sale. Clearly there is much to work with in formulating a dairy development strategy. What has been missing has been a realistic assessment of government capabilities and structures. CHAPTER X EXPANDING SMALLHOLDER DAIRYING IN KILIMANJARO A. The Political Economy Context Tanzania has committed itself to promoting self-reliance in rural areas, reducing disparities in opportunities and incomes both within and between urban and rural areas, and disengaging from inter- national linkages which restrain a socialist transformation of its economy. Though such aspirations find many sympathetic ears both within and outside of Tanzania, translating them into practice involves concrete actions, the political consequences of which cannot be ignored. These consequences constrain alternative strategies for developing Kilimanjaro's dairy industry. l. Returns to Milk Versus Coffee Production This study has not fbcused directly on the question of the returns to coffee production on the mountain. Until recently milk production was not seen as a viable alternative to coffee. With the introduction of grade dairy cattle, however, this is beginning to change. The neo-classical analysis of the returns to alternative milk production systems in Chapter IX shows that upgrading zebu cattle yields net returns to the additional resources employed that are con- siderably above their opportunity cost. Returns from coffee 194 195 production constitute a portion of that opportunity cost, but so do returns from bananas, maize, beans and other livestock enterprises as well as the availability of household labor. Only a whole farm analysis that simultaneously considers the entire set of resources available to the farm household and the available production oppor- tunities can determine the impact of substituting forage production for coffee on net farm income. At the time of the survey very few farmers were producing fOrage on mountain land and the state of the art was crude indeed. Consequently it would have been difficult to draw strong conclusions from a study of such farms. Still, ll of T40 weighted households in the composite grade cattle sample did plant a quarter acre or more of their mountain land in grasses as a source of livestock feed. This represents about 10 percent of the average kihamba holding and suggests that some farmers, at least, found forage production more profitable than coffee. Of all the enterprises on the farm, coffee is the one most likely to be replaced by fbrage. Bananas are an important fOod staple and source of livestock feed and thus an unlikely candidate for replace- ment. The maize and bean fields in the lowlands cannot produce forage during the dry season when feed supplies are most constraining. It is possible to produce forage in these fields during the rainy season but storage is a problem given the damp, cloudy climate at that time of the year. Grass grown on the mountain homestead, however, can be cut fresh throughout the year thereby maximizing the quality of feed as well as TDN per acre. To date, direct competition between coffee and dairying for resources has been avoided because of the limited supply of grade 196 cattle and the availability of labor to cut and carry grass from the forest and plains. As the quality and numbers of grade cattle increase, however, and milk production becomes more responsive to improved feeding and management, we can expect more direct competi- tion between the two enterprises for available land and labor. The magnitude of the benefit-cost ratios obtained fbr the grade cattle enterprises suggest that under current relative prices between coffee and milk grade dairying can successfully compete against cof- fee for available resources when these are valued at their average annual opportunity cost. The reader does need to exercise caution in drawing such conclusions from the analysis in Chapter IX alone. Because of taxes and union levies on coffee, farmers receive only 70 percent of the F08 value of their coffee whereas milk produced on small holdings-—essentially a nontraded good-—is not taxed. About 15 percent of the total FOB value of coffee reflects cooperative union levies for various services, including processing the parchment, and thus represent legitimate costs of production. However another l5 percent consists of development levies, export taxes and District Council cesses [Mhaville, 1966]. These transfers effectively reduce private returns relative to social returns and cause the social returns to enterprises using the same resources to be overstated when prevailing market prices are used to value inputs and output as we have done in Chapter IX. The magnitude of the distortion caused by using prevailing market prices is not as great as appears at first glance. In the semi-subsistence economy of Kilimanjaro any increase in cash income quickly translates into higher prices fer nontraded locally consumed 197 goods. Milk and banana prices rise and fall with coffee prices, the former because it is clearly a superior good and the latter because it is a principal ingredient of the local banana/millet beer.1 This means that manure also increases in value. Meat prices rise as well though the extent of such a rise is greatly tempered by the ready supply of cattle from the non-coffee producing areas of the country. Thus the benefit-cost ratios of the grade dairy alternatives, though overstated, are probably not so overstated as to give a misleading picture of underlying economic relationships. A third factor favoring dairying over coffee is the recent wide- spread outbreak of Coffee Berry Disease on the mountain. By the 1972-73 crop year KILIMO estimated that 25 percent of the coffee in Moshi and Rombo Districts was being lost due to the disease and the infected area was increasing [Regional Agricultural Development Office, 1973]. Chemicals for controlling the disease are very expensive, ranging from 770/: to over 2000/= per hectare [Bujulu, l973] versus average gross income from coffee on small holdings of about l400/= at the farm level and 1900/= at the national level, though yields on plots which are properly sprayed are well above average. The heavy spraying in 1972-73 greatly reduced the extent of infestation and the area covered by the special CBD control program was expanded consider- ably the fOllowing year. However costs of this magnitude greatly reduce the attractiveness of coffee unless intensification of produc- tion succeeds in increasing yields enough to offset the increased costs. He can only speculate on the extent to which this is likely 1There is little doubt that consumption of beer is highly elastic with respect to income for the average peasant in Kilimanjaro. l98 to occur but it seems unlikely that farmers will end up with higher net incomes from coffee than before the outbreak unless these inputs are subsidized. On balance then, the evidence suggests that as opportunities fer producing milk from grade cattle expand, competition with coffee for land resources will intensify, at least until such time as milk prices begin to fall. As we pointed out in Chapter V, this may be quite some time in the future. 2. Macro-Economic Effects of Increasing Milk Production The macro-economic effects of increasing smallholder milk produc- tion in Kilimanjaro will depend on how such an increase is brought about and what efforts are devoted to those crops which compete with dairying for resources. Smallholder coffee yields in Kilimanjaro, fer example, average only about one-third those on estates, suggesting substantial room for intensification. There may also be some poten- tial fer introducing sources of forage that do not compete so directly with coffee. In the absence of intensification of coffee production or research on new sources of ferage, increasing milk production will probably induce a decline in coffee production. This will lower foreign exchange earnings from coffee dollar fer dollar. What it does to milk imports depends on the kind of strategy that is adopted fer increasing milk production. Maximum savings in foreign exchange expended on milk imports would be obtained if grade cattle were con- centrated on the mountain so surplus milk could be collected, and distributed to urban centers where consumption of imported dairy l99 products is highest. Net fereign exchange savings would be reduced if such milk were first processed since most of the plant, equipment and operating inputs are also imported. A second approach to expanding milk production would be to pre- vent the build up of pockets of milk production that is surplus to local consumption needs by promoting a more even distribution of grade cattle over the mountain. This would increase agricultural income from milk production over the concentrated alternative by retarding the build up of supplies of milk that could depress local market prices. At the same time, however, much of this increased production would be consumed by households not now purchasing milk and much higher levels of aggregate milk production would be needed to save equal increments of foreign exchange. Thus the choice is between increased farm income, increased consumption of milk and expanded production/consumption linkages within the national economy on the one hand, versus saving fbreign exchange on the other. Here available foreign exchange being used to develop an integrated industrial sector capable of bringing about a true structural transfbrmation of the economy this would be a difficult choice. Of course nothing suggests that if Tanzania needs foreign exchange fer industrialization it shouldn't stop imports of dairy products completely and allow consumers to bid up the price of local supplies. This would shift income away from urban consumers which, the government admits, are favored by current policies, and toward milk producers and capital goods industries. Since the fermer are mostly small holders and the latter are critical fer economic trans- formation the choice would seem to be obvious. That such a policy 200 has not been taken confirms that development policies are more easily articulated than implemented. Thus the social value to Tanzania of producing milk versus coffee, should the choice come to that, depends on the array of policies that Tanzania envisions as acceptable to it and which the country is willing to implement. Such policies can only be decided in the political arena and are, by and large, beyond the scope of this paper. 3. The Local Context The structure of political and economic institutions at the local level also have an important bearing on what the effects of particular development strategies are likely to be. In Kilimanjaro three factors stand out fer consideration: the cooperative movement, organization of agricultural and veterinary extension services, and the patron-client relationship between government leaders and peas- ants. All three are interrelated. a. The Cooperative Movement. Since its early days the coopera- tive movement in Kilimanjaro has been dominated by a class of larger, more successful, commercial farmers. Membership control is weakened by the patron-client relationship encouraged by cooperative leaders vis a vis the less-educated membership. Fears of witchcraft and the knowfledge that some of these leaders will extend a helping hand to a grateful client in time of need lead to a great deal of membership acquiescence to policies that favor these elite farmers. Allocation of cooperative society loans for the purchase of grade cattle (which available evidence suggests go unrepaid) and subsidies of the trans- port of concentrates are but two examples of the way in which this class of farmers enriches itself at the expense of the broader 20l membership. As long as grade cattle and grade cattle services remain scarce, using the cooperatives as a mechanism fer allocating them would leave the mass of middle and lower level peasants as residual claimants. The elite would continue to extract economic rents from their investments in grade cattle while peasants have to be content with zebu cattle until rents on grade cattle are reduced to more normal levels. b. Organization of Agricultural and Veterinary Extension Services. The organization of Ministry of Agriculture extension services also favors elite farmers and discriminates against peasants. The chronic shortage of budget allocations fer petrol, transportation and medicines means that farmers who can transport agents or entertain them with beer and other niceties are more likely to obtain medicines or have their calls fer service answered. The essentially unscheduled program of activities for veterinary assistants leaves them free to ignore requests from peasants fer service with the excuse that they are too busy. Indeed in such a system much time is consumed simply going from one farm to another in rather helter-skelter fashion. It leaves little scope for peasant supervision and control and any effort to create some would be actively resisted by the agents. c. The Patron-Client Relationship Between Government and Peasants. In spite of all the talk about self-reliance in Tanzania many government leaders in Kilimanjaro still "bring" development to the people and expect them to be grateful for it. Rather than self- reliance, induced dependency is more common. An alliance between cooperative leaders, elite farmers, local government officials and extension agents is very much in evidence. Green [l974], Mbilinyi 202 [1975] and others have commented on this and have noted the difficulty of overcoming it. Only a program which places a large amount of con- trol over the work programs of extension agents in the hands of peas~ ants is likely to neutralize this alliance and give peasants broader access to public sector resources. B. The Technical Context The analysis of the determinants of milk yields for grade cattle in Chapter VIII confirms that variables over which a well-organized extension service can have some control explain about 70 percent of the variation in milk yields between average and high producing grade cows. Moreover, there is a clear tendency for farmers with both zebu and grade cattle to manage the animals differently. This difference suggests that genetic improvement by itself may generate substantial increases in milk production by appreciably reducing calving intervals for the average farmers' cows. A reliable and predictable supply of high quality semen is an essential ingredient for any program aimed at upgrading zebu cattle. Zebu cows confined in stalls may not exhibit their oestrus periods 'Iopenly enough or long enough for farmers to catch them in heat in time to get them bred with artificial insemination. However, the fact that conception rates for zebu cows bred to zebu and grade bulls are high [Zalla, l974] suggests that the quality of the A.I. semen and the way in which the delivery service is organized may be more of a prob- lem than farmers' ability to identify heat periods in time. There is the additional problem-—reported by farmers and con- firmed by some extension agents-—of artificial inseminators not responding to calls for service from farmers who do not already have 203 grade cattle. This practice is justified on the grounds that the number of inseminators is insufficient to cover zebu cows and any attention directed to them would cause farmers already having grade cattle to down breed their herds in an effert to get their cows in milk as quickly as possible. There is some truth in this contention given the way the system has been operating, but there is also con- siderable scope for improving operating efficiency for overcoming such problems. In addition to breeding, feeding is another problem. This is confirmed by the analysis of yield determinants which show marginal value products in excess of marginal factor costs for grain, cash expenditures on grass, frequency of feeding water and salt to grade cows, as well as for the quantity of labor expended carrying forage- presumably also a measure of the quantity of roughage fed. Grain and salt will have to come from off-farm sources and the cooperative soci~ eties provide an excellent structure for obtaining and distributing these items. Additional forage will be more difficult to come by and will probably not be ferthcoming in adequate quantities for the average farmer until greater forage production can be incorporated into the existing farming system or researchers demonstrate and farmers per- ceive a greater return to resources employed in fbrage production than in coffee production. This will probably be quite some time in the future. In addition to increasing the availability of inputs, farmers need extension advice on how to care for grade cattle, including spraying, feeding, and care of calves. Given the high price of milk it will probably be necessary to offer financial incentives to farmers 204 with good quality male grade calves in order to bring these animals 1 So far, veterinary officials have to maturity for breeding purposes. encouraged the use of local grade bulls only in remote areas that are not accessible to A.I. However, the time has come to acknowledge the problems with the A.I. program by shifting to a more flexible approach until A.I. can be made to function effectively. C. A Strategy for Developingthe Dairy Industry in Kilimanjaro The nature of Tanzania's stated political and economic objec- tives suggests that dairy development effbrts in Kilimanjaro center around a broad based effort to increase the number of grade cattle by upgrading zebu cattle across the entire coffee/banana belt of the mountain. Such an approach would maximize the income, nutrition and equity effects of increasing milk production, promote self-reliance in rural areas and promote integrated economic development within the national economy as a whole. It should maximize local value added and create a vast reserve of grade cattle that can eventually be trans- ferred to other suitable areas of Tanzania, such as Hest Lake and Mbeya regions. Hhen milk surpluses do begin accumulating on the moun- tain serious consideration needs to be given to legalizing direct sale of fresh and fermented milk in Moshi Town and other urban markets while keeping milk processing plants at a size justified by the spon- taneous demand for pasteurized milk. A potential strategy might consist of the following components: 1High milk prices raise the opportunity cost of milk fed to calves and lead to the slaughter of bull calves in an effort to release more milk for sale. 205 Cash payments to farmers with relatively good quality grade animals to encourage them to grow out male animals fer breeding purposes. Guaranteed payments of l400/8 per mature bull would be adequate incentive given current returns to grade dairying. Distribution of these high quality males to selected farmers. throughout the mountain who demonstrate good husbandry techniques and a willingness to make the bull available to local farmers for .breeding purposes. Such farmers would charge a minimal fee which they could keep, with government guaranteeing some minimum level of income. The enterprise budgets indicate that an income of 900/- shillings per year would assure a return on resources equal to that of a grade cow if these farmers were required to purchase the bulls. This would come to around 25/- per pregnancy if the bull serviced one cow per week. Farmers would be quite wiiling to pay this if they are reasonably assured of a grade pregnancy. .' Restructuring the A.I. service around a system of roadside crushes that can be partially serviced on feat if necessary. This will reduce the area that can be covered by one inseminator and will concentrate services more. But it will eliminate the excuse fer doing nothing when transportation budgets are exhausted. Conse- quently, it will increase inseminator time in the field and increase the number of inseminations per month. Reduce the training time for inseminators by concentrating on the technique of insemination and handling semen. Leave diagnosis to veterinary assistants. Allow cooperative societies to send their own candidates for training on the condition that KILIMO will provide material support if the society pays the candidate's salary. 206 5. Increase the ratio of inseminators to veterinary agents so that the latter can service a larger area than the former. In addi- tion veterinary agents should be required to make regular rounds to central points in the morning with afternoons left fer farmer requested house calls. As with the inseminators this should be along routes which can be abbreviated should transportation not be available. 6. Allow cooperative societies and selected rural stores to stock veterinary medicines fer cash sale to farmers. This will provide a source of supply when ministry.supplies are exhausted. 7. Put each veterinary assistant and artificial inseminator under the supervision of a cooperative society. The society should provide a perfOrmance bonus in addition to any regular salary the agents might receive from government. 8. Carry out serious research on native and exotic grasses and legumes in order to identify those with the best production capa- bilities per hectare of land at various altitudes and those which can be intercropped with coffee or bananas or otherwise be inte- grated into the farming system. Something like Kudzu, a prolific legume, which can grow along fence rows, on houses and trees, perhaps even on coffee trees and banana pseudostems is an example of what needs to be tested.1 9. Conduct research on intercropping nitrogen-fixing leguminous forages with bananas and coffee to identify economic combinations 1The prolific growth of Kudzu that has been such a disaster in the southeastern U.S. would be a major asset in Kilimanjaro because of the scarcity of land. 207 that can increase farm incomes at various altitudes and levels of rainfall. l0. Study the feasibility of using hungry season tuber reserves normally planted along irrigation furrows as cattle feed in years when crop harvests are adequate. Eventually a dairy expansion strategy will have to incorporate marketing and pricing considerations. The immediate emphasis, how- ever, should be on maximizing on-farm consumption of milk, inter- farm sales of raw milk and, as a result, farm incomes. The strategy outlined here addresses the principal problems of the supply of semen and forage and gives farmers more control over veterinary services and the supply of other inputs. It maximizes access to bull services and other inputs and should, as a result, have more favorable equity effects than the current limited access program. D. Further Research The study uncovered several areas relating to the cattle enter- prise and dairying in Kilimanjaro that should be examined more closely. In addition to the forage research already discussed, con- trolled studies of the relationship between intake of grain, water, salt and other products and milk production are necessary to evolve recommendations to farmers on economic feeding regimes. Researchers also need to explore more fully the reasons for high male calf mor- tality. High slaughter rates would not be surprising but just why the animals are allowed to die is not clear. Much more attention needs to be given to goat milk production in Rombo District. At this point breeding research is more important than economic research on existing local goats. But as soon as a 208 sizable number of improved goats become available their potential role in the farming system should be examined. It is quite possible that a goat dairy enterprise makes better economic sense than a cattle enterprise in Rombo District where dry season feed supplies are of very poor quality. E. Epilogue This analysis of the smallholder dairy industry in Kilimanjaro has taken a broader approach than one commonly observes in dealing with farm level agricultural production problems. It recognizes that agricultural production decisions can have important macro-economic and political economy implications. It uses neo-classical market economics to analyze some of those implications but emphasizes the importance of going beyond neo-classical theory into social, political and other institutions which are not static in a real world context. Understanding how these institutions operate is an important step in ensuring that the intended results of agricultural production projects will, indeed, be ferthcoming. BIBLIOGRAPHY BIBLIOGRAPHY Abdullahi, A. and G. Lombin. 1978. ”Long Term Fertility Studies at Samaru-Nigeria: III. Comparative Effectiveness of Separate and Combined Applications of Mineral Fertilizers and Farmyard Manure in Maintaining Soil Productivity Under Continuous Culti- vation in the Savannah." Samaru Miscellaneous Paper No. 75. Zaria: Institute fer Agricultural Research, Ahmadu Bello University. Acland, J.D. l97l. East African Crops. London: Longman Ltd. Baumol, Hilliam J. l965. ”Infermed Judgement, Rigorous Theory and Public Policy.” Southern Economic Journal. vol. XXXII, No. 2, pp. l37-l45. Baumol, H.J. l974. "The Transformation of Values: Hhat Marx 'Really' Meant (An Interpretation).“ The Journal of Economic Literature. Vol. XII, No. l, pp. 54-60. Beck, Rey S. l96l. “An Economic Study of Coffee-Banana Farms in the Machame Central Area.” Tanganyika: Agency fer International Development. Bene, E.J., C.R. Hogland, E.D. Longnecker and R.L. Cook. l96l. “Animal Manures: What Are They North Teday.“ East Lansing: Michigan State University, Agricultural Experiment Station Bulletin 23l. Bornstein, Set. l97l. "Artificial Insemination Development in Kilimanjaro Region.“ Paper presented to the Veterinary and Agricultural Animal Conference, Moshi, Tanzania, August 23-26. Brevin, D.R. l965. “Kilimanjaro Agriculture." Tanzania Notes and Records. No. 64, pp. llS-ll7. British Naval Intelligence Division. cl920. A Handbook of German- East Africa. London: Her Majesty's Stationery Office. Bujulu, J. l973. “The Evaluation of Fungicides in C.B.D. Control.” Lyamungu Newsletter. No. XLI. Moshi: Ministry of Agriculture and Cooperatives . Calvert, Albert. l970. German East Africa. New York: Negro Universities Press. 209 210 Cochran, William G. l963. Sampling Techniques. New York: John Hiley 5 Sons. Collis, H.R.F., J. Oema and A. Omololu. l962. "On the Ecology of Child Health and Nutrition in Nigerian Villages I B II." Tropical and Geographical Medicine. No. l4, pp. ldO-l63, 201-229. Coulson, Andrew. l972. "A Dairy Development Project for the Northern Regions of Tanzania." Dar es Salaam: Ministry of Agriculture and Cooperatives. Coulson, A. and S. Botros. cl97l. ”Report on the State of Planning of the Northern Region's Dairy Development Project.” Dar es Salaam: Ministry of Agriculture and Cooperatives. Dahl, Gundran and Anders Hjort. l976. Having Herds: Pastoral Herd Growth and Household Economy. Stockholm: Department of Social Anthropology, University of Stockholm. Dundas, Charles. l968. Kilimanjaro and Its People. London: Frank Cass and Co. Ltd. First published in 1924. East African Customs and Excise Department. l973. Annual Trade Report of Tanzania,_Uganda and Kenya. Mombassa: East African Customs andiExcise Department. Egero, Bertil and Roushdi Henin. l973. “Pepulation Growth. “ In The Pepulation of Tanzania. Ed. by Bertil Egero and Roushdi Henin. Dar es Salaam: The Bureau of Resource Assessment and Land Use Planning and the Bureau of Statistics. Emmanuel, Arghiri. l972. Unequal Exchange: A Study of the Imperi- alism of Trade. New York: Monthly Review Press. FOod and Agriculture Organization and Department of Health Education and Helfare.1968. FOod Composition Tables for Use in Africa. Rome: FOod and Agriculture Organization and Department of Health, Education and Welfare. Freyhold, M. von, K. Sawaki and M. Zalla. l973. "Moshi District.” In The Young Child in Tanzania. Ed. by the Tanzania National Scientific ResearCh Council. Dar es Salaam: United Nations Information, Children and Educational Fund (UNICEF). Friedman, Milton. l966. Price Theory, A Provisional Text. Chicago: Aldine. I Fuggles-Couchman, N.R. l964. Agricultural Changgin Tanganyika: l945-l960. Stanford: Food Research Institute. le Gottlieb, Manuel. l972. “The Extent and Character of Differentiation in Tanzanian Agriculture and Rural Society, l967-l969.” Paper delivered at the Conference of the Social Science Council of the Universities of Eastern Africa, Nairobi, December. Green, Regionald Herbold. 1974. "Toward Ujamaa and Kujitegemea: Income Distribution and Absolute Poverty Eradication Aspects of the Tanzania Transition to Socialism." Dar es Salaam. (Unpublished Mimeograph.) Gwassa, G.C.K. 1969. "The German Intervention and African Resistance in Tanzania.“ In A History of Tanzania. Ed. by I.N. Kimambo and A.J. Temu. Nairobi: East African Publishing House. Hathaway, Dale E. 1963. Government and Agriculture. New York: MacMillan. ‘Henick, Allison Butler et al. 1968. Area Handbook fer Tanzania. Hashington, D.C.: U.S. Government Printing Office. Hill, J.F.R. and J.P. Moffett. l975. Tanganyika: A Review of Its Resources and Their Development. Norwick: Government of Tanganyika. . Iliffe, John. l969. “The Age of Improvement and Differentiation, (1907-45).“ In A History of Tanzania. Ed. by I.N. Kimambo and A.J. Temu. Nairobi: East African Publishing House. . l97l. Agricultural Change in Modern Tanganyika. Historical Association of Tanzania Paper No.10. Nairobi: East African Publishing House. Institut des Recherches Agricoles et Ferestieres (IRAF). l976. Rapport Analytique l976 de Centre Des Cultures Vivrieres et Textiles de Maroua. Tome II (Troisieme Partie). Maroua: Office Nationale de la Recherche Scientifique et Technique (ONAREST). International Bank for Reconstruction and Development. l96l. The Economic Development of Tanganyika. Baltimore: Johns Hopkins University Press . . l974. Tanzania, Agricultural and Rural Development Sector Study, Vol. l. Hashington, D.C.: International Bank fer Reconstruction and Development. Jabara, Cathy L. and Robert L. Thompson. l9BO. "Agricultural Com- parative Advantage Under International Price Uncertainty: The Case of Senegal.“ American Journal of Agricultural Economics. No. 2. pp. 188-198. 212 Johnson, Glenn L. 1972. "Characteristics of U.S. Agricultural Economy.” In The Overproduction Trapyin U.S. Agriculture. Ed. by Glenn L. Johnson and C. Leroy Quance. Baltimore: Johns Hopkins University Press. Johnston, H.H. 1886. The Kilimanjaro Expedition. London: Kegan Paul, Trench a C0. Johnston, P.H. 1940. ”Some Notes on Land Tenure on Kilimanjaro and the Kihamba of the Hachagga." Tanganyika Notes and Records. No. 21, pp. 1-20. Kilimanjaro Native C00perative Union. 1963. "Chagga Coffee." Pamphlet. . Kosikowski, F.V. 1970. Cheese and Fermented Milk FOods. Ann Arbor: Edwards Brothers Inc. Kraph, J. Lewis. 1968. Trials, Researches agd Missionary Labours .Quringan Eighteen Year's_Residence in Eastern Africa. London: FrankPCass and Co. Ltd. First publishedin 1860. Laak, R. 1970. “Cattle, Swine and Chicken Manure Challenges Haste Disposal Methods." Hater and Sewage Horks. April, pp. 134-138. Lev, Larry. 1981. “The Effect of Cash Cropping on Fbod Consumption Adequacy Among the Meru of Northern Tanzania." MSU Rural Development Horking Paper No. 21. East Lansing: Department of Agricultural Economics, Michigan State University. Lindner, Irmgard. 1972. ”A Baseline Study of Under Five Clinics in Kilimanjaro.‘ Moshi: Kilimanjaro Christian Medical Center. Mimeo. Lipsey, Richard and Kevin Lancaster. 1956. “The General Theory of Second Best.” Review of Economic Studies. Vol. 24, pp. 11-32. Lombin. G. and A. Abdullahi. 1977. “Long Term Fertility Studies at Samaru-Nigeria: Effect of Farm Yard Manure on Monocropped Cotton, Sorghum and Groundnuts and a Rotation of the Three Crops Under Continuous Cultivation.” Samaru Miscellaneous Paper No. 72. Zaria: Institute fer Agricultural Research, Ahmadu Bello University. Lucas, Stephen and Gerald Philippson. 1973. ”Ethnic Characteristics.“ In The ngulation of Tanzania. Ed. by Bertil Egero and Roushdi Henin. Dar es Salaam: Bureau of Resource Assessment and Land Use Planning and the Bureau of Statistics. Marealle, T.L.M. 1952. ”The Hachagga of Kilimanjaro." Tanganyika Notes and Records. No. 32, pp. 57-64. 213 Maro, Paul S. 1974. “Population and Land Resources in Northern Tanzania: The Dynamics of Change, 1920-1970." Unpublished Ph.D. dissertation. St. Paul: Department of Geography, University of Minnesota. Marth, E.H. 1968. ”Salmonellae and Salmonellosis Associated with Milk and Milk Products: A Review.” Journal of Dairy Science. No. 3, pp. 283-315. Mbaga, Fadhili Daudi. 1968. ”Some Economic and Technical Aspects of Milk Production on Small Holdings in Kilimanjaro and Meru Districts of Tanzania." MISER Rural Development Research Paper No. 58. Kampala: Makerere University College. Mbilingi, Marjorie J. 1975. The Transition to Capitalism in Rural Tanzania. Economic Research Bureau Paper. Dar es Salaam: University of Dar es Salaam. Meadows, S.J. and J.M. Hhite. 1979. "Structure of the Herd and Determinants of Offtake Rates in Kajiado District in Kenya, 1962-1977." Overseas Development Institute Pastoral Network Paper 7d. London: Overseas Development Institute. Mhaville, J.A. 1966. “Report of the Presential Special Committee of Inquiry into the Cooperative Movement and Marketing Boards." Dar es Salaam: Ministry of Agriculture and Cooperatives. Miles, Vernon. c1968. Unpublished draft of a report written for the Kaputaie, Kenya, Masailand Area Planning Commission. Mimeo. Ministry of Agriculture. 1959. Northern Province File No. N/465. Arusha: Ministry of Agriculture. Mimeo. Mishra, 6.5., E. Camus, J. Belot and A.E. N'Depo. 1979. ”Enquete sur le parasitisme et la mortalite des veaux dans le Nord de la Cote-d'Ivoire: Observations preliminaires.“ Review d'Elevage et de Medicine Veterinaire des Pays Tropicaux. No. 42, pp. 353-359. Morrison, Frank B. 1956. Feeds and Feedigg, Ithaca: The Morrison Publishing Co. Mphuru, N.E. 1965. The Effect of Fuggicide and Fertilizer Treatments gn_Yields of Coffee on Peasant Farms. Research Report, Coffee Research Station, Lyamungu. Moshi: Ministry of Agriculture and Cooperatives . New. Charles. 1968. Life,Handeriggs and Labours in Eastern Africa. London: Frank Cass and Co. Ltd. Originally published in 1885. Pratt, D.J. and M.D. Gwynne. 1977. Rangeland, Management and Ecolggy_in East Africa. London: Hodder and Stoughton. 214 Pullan, N.B. 1979. ”Productivity of Hhite Fulani Cattle on the Jos Plateau, Nigeria: Vol. I. Herd Structures and Reproductive Perfbrmance.” Tropical Animal Health and Production. Vol. 11, pp. 231-238. . 1980. “Productivity of Hhite Fulani Cattle on the Jos Plateau, Nigeria: Vol. III. Disease and Management Factors." Tropical Animal Health and Production. Vol. 12, pp. 77-84. Ranger, Terence. 1969. ”The Movement of Ideas, 1850-1939.“ In.A History of Tanzania. Ed. by I.N. Kimambo and A.J. Temu. Nairobi: East African Publishing House. Rose, Hilliam T., T.L. Loudon and B.A. Stout. 1979. ”Anaerobic Digestion of Livestock Hastes into Methane Gas.” Agricultural Engineering Information Series, No. 403. East Lansing: Depart- ment of Agricultural Engineering, Michigan State University. Rweyamamu, Justinian. 1973. Underdevelopment and Industrialization in Tanzania: A Study of PerverseCapjtalist Industrial Development. Nairobi: Oxford University Press. Sacker, 6.0. and J.C.M. Trail. 1966. ”A Note on Milk Production of Ankole Cattle in Uganda.“ Tropical Agriculture, Trinidad. No. 3. Pp. 247-250. Shann, G.N. 1956. “The Early Development of Education Among the Chagga." Tanzania Notes and Records. No. 45, pp. 25-29. Shulman, Robert. 1979. "Strategy fer the Advancement of Animal Traction in Mali." Bamako: United States Agency for Inter- national Development. Mimeo. Simoons, Frederick J. 1978. ”Lactose Malabsorption in Africa." African Economic History. No. 5, pp. 16-34. Snedecor, George H. 1948. Statistical Methods. Ames: Iowa State College Press. Soulides, D.A. 1949. Better Utilizatiqggof Milk. FAO Agriculture Studies No. 7. Hashington, D.C.: Food and Agriculture Organization. Stahl, Kathleen. 1964. History of the Chagga People of Kilimanjaro. The Hague: Mouton and Co. Stobbs, T.H. 1967. "Management of Small East African Zebu in Rela- tion to Milk Yield, Calf Growth and Mortality.” East African Agricultural and Forestry Journal. Vol. XXXII, No. 3, pp. 250-255. Stout, B.A. and T.L. Loudon. 1976. "Energy from Organic Residues.” East Lansing: Agricultural Engineering Department, Michigan State University. Mimeo. 215 Sukhatme, Pandurang. 1954. Sampling Theory of Surveys with Applica- tions. Ames: The Iowa State College Press. Swantz, Maria-Liisa, Ulla-Stine Henricson and Mary Zalla. 1975. Socioeconomic Causes of Malnutrition in Moshi‘District.- Bureau of Resource Assessment and’Land Use Planning Research Paper No. 38. Dar es Salaam: University of Dar es Salaam. Swynerton, R.J.M. 1945. Native Agriculture-ibshi District. Moshi: Ministry of Agriculture and Cooperatives Files. Mimeo. Sykes, A. 1959. ”Coffee/Banana Survey-Moshi District, 1959.” Moshi: Ministry of Agriculture and Cooperatives Files. Mimeo. Tairo. 1965. “Report of a Survey of Six Grade Cattle Owners in Moshi District.“ Moshi: Ministry of Agriculture and Coopera- tives Files. Typescript. Tanganyika African National Union. 1967. The Arusha Declaration and Tanu's Policy on Socialism and Self Reliance. Dar es Salaam: Tanganyika African National Union. Tanganyika Territory Department of Agriculture. 1952. Annual Report. Dar es Salaam: Department of Agriculture. Tanzania, United Republic of. 1964. Statistical Abstract, 1963. Dar es Salaam: Central Statistical Bureau. . 1967. A Mid-Term Appraisal of thg Achievements Under the Five Year Plan, July, l964-June,fi1969. Dar es Salaam: Ministry of Economic Affairs and Development Planning. 1969. 1967 ngulation Census,Vb1. 1, Statistics for Enumeration Areas. Dar es Salaam: Bureau of Statistics. . 1971a. 1967 Population Census, Vol. 3, Demographic Statistics. Dar es Salaam: Bureau of Statistics. 1971b. 1967 ngulation Census, Vbl. 4, Economic Statistics. Dar es Salaam: *Bureau of Statistics. . 1971c. 1967 Population_Census, Vol. 5, Census Methodology. Dar es Salaam: Bureau oflStatistics. . 1971d. Hggsing,Conditions;_ 1969 Household Budget Survey, Vol. 2. Dar es Salaam: Bureau of Statistics. . 1972. Statistical Abstract, 1970. Dar es Salaam: Bureau of Statistics. 1974. Crop Production Statistics in Tanzania, 1964-72. Dar es Salaam: Ministry of Agriculture and Cooperatives. 216 Tanzania, United Republic of, Ministry of Agriculture and Cooperatives. 1963. “Development of African Owned Grade Dairy Cattle on Kilimanjaro.” Anonymous report found in Dairy Industry, Tanganyika, File No. 1714, Moshi Station. Moshi: Ministry of Agriculture Files. Files PU93/Agric/2506 and Arusha/VY/1654. Arusha: Ministry of Agriculture and Cooperatives. Tanzania, United Republic of, Ministry of Economic Affairs and Development Planning. 1968. District Data 1967. Dar es Salaam: Ministry of Economic Affairs and Development Planning. Tanzania, United Republic of, Regional Agricultural Development Office. 1973. “Coffee Berry Disease Control: Kilimanjaro Region; Review of the 1972/73 Programme and Proposals for the 1973/74 Programme.” Moshi. Unisgned Mimeo. Tanzania Nutrition Committee. 1970. "Protein Problem in Tanzania.“ Paper presented to the 17th Protein Advisory Group Meeting, 1970, New York. - Thomas, Clive Y. 1972. "The Transition to Socialism: Issues of Economic Strategy in Tanzania." Economic Research Bureau Seminar paper. Dar es Salaam: University of Dar es Salaam. Turk, L.M. and A.G. Heidemann. 1945. ”Farm Manure." Michigan State College Agricultural Experiment Station Bulletin 196. East Lansing: Michigan State University. Hallace, Ian Richard. 1968. ”Peasant Production of Arabica Coffee in East Africa: Technical and Economic Studies in Bugisu, Meru and Kilimanjaro.” Unpublished M.S. Thesis, Makerere University College. Kampala: University of East Africa. Hilliamson, G. and H.J.A. Payne. 1978. An Introduction to Animal Husbandry in the Tropics. Longman: London. Hilson, R.T. and S.E. Clarke. 1976. ”Studies on the Livestock of Southern Darfur, Sudan: vol. II. Production Traits in Cattle.“ Tropical Animal Health and Production. No. 8, pp. 47-57. Yates, Frank. 1960. Sampling Methods for Censuses and Surveys. New York: Hafner Publishing Company. lalla, TxM. 1972. "Dairy Development in Tanzania: Implications fer Nutrition Improvement, Product Locations and Import Substitu- tion, Vol. 1.“ Provisional Report to the Ministry of Agricul- ture, Food and Cooperatives. Dar es Salaam: University of Dar es Salaam. 217 Zalla, T.M. 1974. "Herd Composition and Management Practice Data on Small Holder Milk Producers in Moshi and Rombo Districts: Some Preliminary Results." Economic Bureau Research Paper 74.8. Dar es Salaam: University of Dar es Salaam. Zalla, Tem. 1979. ”The Relative Importance of Money and Subsistence Incomes in Explaining Dietary Intake in Kilimanjaro: Prelimi- nary Results.” Paper presented at the Midwest Conference on Economic Development held at Ann Arbor, November 9-10. APPENDICES APPENDIX A APPENDIX A DATA COLLECTION AND ESTIMATING PROCEDURES A. Sampling,Methodology The survey area covers the smallholder coffee-banana zone of Rombo and Moshi Districts from about 3000 to 6000 feet elevation on Mount Kilimanjaro. Topography, rainfall, and agricultural productivi- ty in the area are dominated by the mountain peak with distinct pat- terns of variation according to altitude and direction of the slope. Rainfall is heaviest on the southern slope, gradually diminishing in a northern direction. The northern slope is dry and uninhabited. However, rainfall varies with altitude on all slopes of the mountain as does temperature. Ridges and hills interrupt these general pat- terns here and there, giving rise to pockets of more or less rainfall, but the overall pattern is remarkably continuous. A two stage probability sample was employed for the study. The 1967 census enumeration areas served as first stage sample units, being systematically selected with probability proportional to their 1967 household populations. From each selected first stage unit with- in each of the districts an independent constant size sub-sample of individual households was drawn systematically. From a practical standpoint the systematic sampling procedure adopted can be taken as yielding a representative random sample to which conventional two stage estimating procedures apply. A-l A-2 The systematic sampling procedure used for selecting first stage sample units took account of the continuous variation in rain- fall and altitude in the survey area. As a first step the survey area was divided into 12 zones of roughly equal household population (Nz . 6364 1 19). Each zone consisted of a vertical slice of the mountain running from the lower boundary of the survey area to the upper boundary (see map on next page). Then beginning at the top of the first zone and moving toward the bottom, all census enumeration areas were listed and a cumulative household listing constructed. After listing all first stage units in zone 1 the listing process continued into zone 2, working from the bottom of the zone toward the top. Zones 1, 3, 5, 7, 9 and 11 were listed from top to bottom and zones 2, 4, 6, 8, 10 and 12 from bottom to top. The listing procedure ensured that a systematic sample with a systematic sampling interval equal to 1/36 of the household popula- tion would include high, medium and low altitude variations within each of the 12 zones. Listing being opposite for contiguous zones, the expected value of the k systematic samples was the mean of the respective systematic sampling interval regardless of where on the l to k systematic sampling interval the sampling process began. Actual selection of the first stage units was done in normal systematic fashion. A random number between 1 and k a 2121 was taken and the first stage sample units were those which included the it“, 1 + kt", 1 + Zkth. . . . 1 + 36kth households as indicated from the cumulative household population listing. Once identified by enumera- tion area number, maps of the selected first stage units were obtained from census records and the areas delineated on the ground. m>~5m ...—o as; _.1< maze: —u_¢~a.o .843‘4133. .- . no.3 33:53 3 :63 .383: ocoN x. 3.3 5:22.?» oEEmm % TIIII'N'l I. 03190 .ucoN 3.320 1...: ..I. ...... .323”. :11 3:32:85 .1 . . -Hmotmucaom A-4 All households within each of the 36 selected areas were then random- 1 An independent systematic subsample was r: I“, 1y listed by ten-cell unit. drawn from each with the systematic sampling interval given by where Mi is the household population listed in the selected enumera- tion area and m is the desired subsample size, constant within a given district. Actually there were as many as three separate samples taken from or near each first stage unit. The first, sample A, was a ran- dom sample picked systematically as described above. Excluding alternates, it included 8 households in each of the 26 first stage sample units in Moshi District, and 20 households in each of the 10 first stage units of Rombo District-—giving roughly 200 households per district.2 A second sample, sample 8, consisted only of those households in the enumeration area having grade cattle but not drawn in the ran- dom sample. The frame for this sample was obtained from ten-cell leaders at public meetings held in each of the selected first stage units. A variable number of households per first stage unit was selected systematically from this list so that a total of at least five grade cattle owners per first stage unit in Moshi District, and 1A local political subdivision including every self-sustaining household. They range in size from 3-30 households but have an expected value of about 10. Each unit has a leader who is the link between the people and the local level administrative structure. 2The larger number sampled per first stage unit in Rombo District was in response to a request fer district level estimates made by government officials after the listing process had been com- pleted. At that time Moshi and Rombo Districts were created out of the fermer Kilimanjaro District. It was not possible to increase the number of first stage units because of prior commitments. A-5 10 per first stage unit in Rombo District, would constitute a com- posite AB sample of grade cattle owners. A third sample of grade cattle owners, sample C, was drawn in all first stage units where fewer than the 5-10 grade cattle-owning households desired for the grade cattle survey were found. Ten-cell leaders were asked to identify those owners of grade cattle living nearest to the sample area until the desired 5-10 households were obtained. Only the first of the three samples, sample A, was truly ran- dom. This sample is used to provide population estimates and descriptive statistics for local cattle and the area in general. Sample 8, the sample of grade cattle owners living in the first stage unit but not selected in the random sample, is random but the sam- pling frame from which it was drawn appears to have been biased. The third sample, sample C, is clearly biased toward better known, wealthier, presumably more progressive farmers. Unlike for sample 8, this was expected. It was intended to provide households with superior management practices and to identify problems faced by inno- vators who receive little governmental support. As it turned out, samples A, B and C were combined in order to derive estimates for grade cattle. The reasons for this are explained later. B. Data Collection Field data were gathered simultaneously from group discussions and individual farm interviews. In each of the selected first stage units one or more group meetings were held with ten-cell leaders, the selected farmers, and any others wishing to attend. The local A-6 T.A.N.U. secretary or ward chairman was usually present at the meeting but there was little evidence of a feeling of intimidation on the part of farmers. The meetings consisted of a brief presenta- tion of the research objectives followed by a question and answer dialogue in which special problems, concerns or constraints relating to the dairy industry in the area were discussed. At the end of the meeting farmers were asked fer their approval to conduct the inter- views. In Moshi and Rombo Districts no first stage units refused to participate.1 The farm survey consisted of a single visit to the sampled households with a fellow-up visit to take blood samples from the cattle. Enumerators gathered data on household membership, income indicators, food consumption, cattle management practices, and detailed individual cattle records. The interview lasted 30-45 minutes for households having no cattle and about 1-1/2 hours fer households with cattle. Infermation was provided almost entirely on a recall basis with the recall period varying from one to two days fer household consumption and milk production data to as long as several years fer cattle records. In general, the length of the recall period required was offset by the significance of the informa- tion asked so that reliable answers could be expected. Only for cash inputs did the recall period present serious problems. These could not always be related to a well-defined time period or change in work 1In a companion study in Arusha District about 20 percent of the first stage units refused to participate. These were mostly Haarusha areas where farmers were more reluctant to divulge inferma- tion on livestock. ‘ A-7 schedule. Often the person answering the questions and handling the cattle was not the one who made the purchase. The survey team consisted of fbur to six enumerators and one supervisor, apart from the author. The entire team interviewed with- in each first stage unit befOre moving on to the next, leaving one or two enumerators to follow up persons not at home and non-respondents. Care was taken to ensure that each enumerator interviewed a cross- section of the combined sample so that subsequent analysis of vari- ance could evaluate enumerator bias. This was a rather costly exer- cise but absolutely necessary because of continuing problems with supervision in other farm surveys in Tanzania. The survey was carried out over the period February-June, 1973, with a scattering of repeat and follow-up visits throughout July and August. This corresponded with the period immediately preceding, during and immediately following the long rains. During about half the period ground cover was relatively scarce and during the other half, relatively plentiful. For seasonally-sensitive items measured for the day or week preceding the survey, sample averages should roughly approximate annual averages. Fer the regression analyses, "however, variables for the time of the year during which the survey took place were included. C. Data Qualiry About 15 percent of all farms sampled were checked for enumera- tor arrival, thoroughness and accuracy. Another 10 percent were reinterviewed by another enumerator either as a fellow up of a previ- ous omission or as an intentional check on enumerator error. In A-B Moshi and Rombo Districts no cases of falsification of questionnaires were fbund. There was only one case of completing a questionnaire after the fact and one of falsification of itinerary. However, frequent discrepancies (:20 percent or more) in numbers of animals reported, calving and breeding months occurred. A good deal more discrepancy occurred in milk production on the day preceding the survey and at key points during the lactation period (at one, three, and six months after parturition and at the end of lactation). There was considerable evidence that many farmers were not being honest in reporting milk yields. Hhere the analysis of the data indicates serious inconsistencies this has been noted in the text and attempts at adjustment have been made. To check for enumerator bias a one-way analysis of variance was performed on 15 important variables. The results are presented in Table A-1. Six of the variables show statistically significant differences at the 95 percent confidence level. Eleven of the 15 are significantly different at or near the 80 percent level. As a general pattern enumerators one and four consistently fall on the extremes. For enumerator one this is not surprising since this is a composite of several enumerators who worked for brief periods only. Normally they would have been assigned to the more literate and wealthier grade cattle owners until they established their inter- viewing skills. Their estimates merely reflect this tendency and do not suggest bias. Enumerator four is generally an outlier. In general, mean esti- mates fer this enumerator reflect his rather timid, unassuming per- sonality. In contrast, enumerator three had a very aggressive, A-9 .-.. ..n.. Aa~.. * .n~.. — .oo.. .n_.v . “a. ”a... u~.~ m¢.~ u4.~ _ .n.~ _ . m..n an.~ .axs..u as.au to so. .o..o,‘ ... . ...”. ._~.. .n~.. _ ..~.. ..u.. .o... .a...». 82 as 8 N 3.» «..n a." ...n 3.” :33 ... ... .2. 82.3 .2 ...a..t.> .puuao . ....».u g».: .o_o‘oa:o= an . .m. . .N... .._.. .c... .mo.. an... area ... got. ..atosx.» ~_m can. .,._ ac. ca. .o.. 08,— a... uac...oo .... .9 «Lob_. .a o_uu~u au.3 «upozouaog >a on.sc . A“... A“... .~._. A“... .m.o. .o.o. -Lasc. x...v 0. o.u.ao._. ..m coco ...n ..n~ ..an ...N “.5” ~._n ...: L.a to... co .Laox .o 2.2 3.2 2.3 2.3 8.3 3.2 329.33.. 2 “no --. ..n~ .... ~.~n ~..~ o.~. o.n~ soc—g Loa.. .0 a».u =.: .5 . ...m. .n... An.~. A..~. ...o_. ...A. ”No .ooo an. an. N._ mm. an. o._ oso.u .0 .Loaa ..a..o=_a< .. . .o_. .m_. .m_. .m.. .om. A”. a~o m~oo man o.” ~.. .c. can can cuaasacou =..aaea co aeaeu .m A.no. A.” . .oqo. .ano. .~oo.. ._~o.. aucasoe.aa¢. .0 .caaeoa m~o coco. o_. ~m.~ ~.. -. -.. ~~.. . . a. uussncoa .o.ao..u .- ._~.. .on.. .AN. .a~.. Aao.. .m... «on ~oo—. no.0 oo.o o~.o nn.¢ oo.o ...o manoeo>.uuucoocn yo nove— .n Ac... An... A.... .m—. A-.. Aso.. Lou-uvvc. usage. no» new.. o.. —.m ... ... a.. a.. ego....a.oa u_og.a=o= .~ ._~.. .-.. ..~.. ..~.. Ac... 6..... . atouaa.o¢. Koo ease. as.“ oo.¢ am.“ so.“ on.» .~.m anode. .ase o_ogaa=ox .— ...na_c-> o_ogon:ox n v n ~ -— can: 0.9..nm m . n u . .3.aoacoo a m. m. m. m. mugs. countesscu vougo.oxc: Iovoocu an.» m «oc.vcooo¢ Lou-coeacu an «outage so oueau_~_cu_m yo «conga venueaum 9:. «coat you-c.9uu oougo_o3e= a.—na.eo> mnu<¢uz=zu z nzxuwp_oru oc.vouoca as» :.g .ns.oa oe.ua ...»couuo ago—o .gu Lam v.5.9coc u-uco.——.e cos. vo._a¢su asaca._—.e mo no.u~c as» ye “nomads \ 52:3... 05 3 Lara a6 a5 .3 332:6» .a ..eeoaa< ..m .n~-o not. a._.om. .a 322.2 .3 61. :2. 3:8... .nu—aaoc n.couccuuaeo aceu co sou-a a».a_uno oz» yo coca. vcaucauu 0:» a. nanosecocna c. coaeazo .. x.ue.aa< a.” .0.-. not. u...am a ..I.» ma avo.coa mo.ca so. vogue: on: acouocolaeo pages.» .0 ou.aoomou - a. «co Lou-Louaeu- .5 on on as a .oN ..u..u as... g..a n._ as .n. ... .N a.“ ..u».a g..: .o. no as. no. «a .m. .u_og..=og _.< .~.m .— .m «a»... 2.2.3 «- o_o. .n_o. Am_o.. ...o. ..«o.. .eoo.. ..as.u .9.29 as a». m-m o-_o. nn~ mop. «mm. o-. m».. .nu. .u.gaaoc co co_uaoaoea .m. .333: 52231.3 ..n.. Avn.v A“... .an.. “up... .o~.v -auoa «gages note» a. ”zoo on. 35.”. n~.. -.n -.. a... .o.. mn.. oo.tu .9 a...» .... s...a ... ..op. .n..v .on.. .9... A.G.. ..ev .A.L..... .zou .uaeo ... omo~. ~m.. -~ “on ~_~_ «an. _o__ .0 toga .... eo.u.ao.s .n. :5 32 $2 82 .E a: .23.: :3 :3 ..~ .~K_. a- a." -~ ~.~ N.— ~n~ .a was; .... ¢o_u.ua.4 .~. ...coo. ...a..... .....u m c n ~ . c.ax ..asan 1 - 3.39.8 nx-vxanxnuunpnus‘. scuoso-acu vougo.oxca 839... 1 1 amend 1 a 5.833 c8335! a. ...Lo.a co .ae.a...co.m .o ates.“ v...¢.um ac. .¢.oz u.u.....u u.uga...¢= ...a..a.> vague-6.1 —1< mac: A-ll perhaps too quick, manner. Both these characteristics come out most strongly on questions relating to food consumption and sensitive areas such as use of hired labor. However, using the composite mean as a standard, these two biases offset each other somewhat. Using the same standard, other enumerators offset the extreme tendencies of enumerator four fer other variables. Partly for this reason, partly because it is not always clear which, if any, of the enumera- tor subsample means are unbiased, and partly because of the imprac- ticality of adjusting different variables for the various enumerator 'biases reflected in the estimates, all observations have been kept. Dummy variables adjust fer enumerator bias in the regression analysis. For population estimates no adjustments fer enumerator bias have been made other than fer lactation milk production. 0. SamplingrBias The sampling procedure adopted fOr the farm survey introduced two sources of bias which are of practical concern for statistical inference: the weights used in selecting first stage sample units; and the quantitative and qualitative difference between grade cattle- owning households included in samples A, B and C. These are discussed in turn. Two stage estimates from samples in which the first stage sam- ple units are selected with probability proportional to an estimate of size require a correction fer the difference between the actual size of the selected first stage unit (1973 household population) and the measure of size used in sampling (1967 household population). In order to avoid introducing a bias it is imperative that the area be A-12 precisely defined at both points in time. This was not possible with the census enumeration areas. Enumeration areas were mapped using footpaths, roads, hollows, forests, the name of household heads, etc. as boundaries. In trying to relocate those boundaries from the census maps we sometimes had considerable difficulty. In a number of cases we were not able to identify the boundaries at all and had to select one ourselves. In others we chose the wrong boundary, a fact which usually became evident only after the field survey team became very familiar with the area in the course of their interviews. As a result of this the areas defined in'l973 are not sufficiently precise to give meaning to the correction factor,-%%, where Mi = the number of elementary units actually included in the ith first stage unit, and Zi = the selection probability associated with the ith first stage unit. A less biased, more efficient and, in fact, more appropriate estima- tor given the stratification effect of the systematic sampling pro- cedure would be the probability proportional to actual size model, even though 1967 sizes rather than 1973 sizes were used. Clearly the 1973 population of the sample area is greater than the 1967 population. There also have been internal population shifts as families continue to move from the mountain toward the lower slopes in search of land. But how these shifts affect given Mi is not possible to determine except in a very broad sense. It would be possible to use the systematic sampling procedure to stratify the sample after the fact so that estimates could be calculated and A-13 weighted separately by high, medium and low altitude strata with 12 first stage sample units each. Normal two stage probability propor- tional to size estimates would then be no more biased fer household estimates than the particular strata weights employed. This has not been done, however. The added computational complexity, the small bias which overlooking this population shift introduces1 and the spurious accuracy such an adjustment would imply given the substan- tial non-sampling errors reflected in the data all support this simplifying oversight. A second source of bias is introduced by aggregating grade cattle households from samples A, B and C to derive population esti- mates relating to grade cattle. Several factors suggest that these samples do not reflect the same underlying populations. 0n the basis of the random sample, sample A, the estimated proportion of house- holds having grade cattle is .10 (1.02). Yet according to the name lists provided by the ten-cell leaders only 6.6 percent (11.2 per- cent) of the households in the survey area have grade cattle. Although this difference is significant only at the 16 percent level, the very small proportions involved coupled with differences in other variables suggest that the two groups do not reflect the same under- lying populations. Table A-2 demonstrates this quite clearly. 1A realistic assumption would be that one-half of the popula- tion increase occurring in the overcrowded medium and high altitude strata moved to the less densely settled, lower stratum over the six-year period 1967-73. As a result population distribution would shift from the 33-1/3 - 33-1/3 - 33-1/3 percent fer the high, medium and low strata implied by the sampling frame to about a 32 - 32 - 36 percent distribution in 1973. It would take very large differences in means and totals fer such a shift to have much impact on population estimates. A-14 TABLE A-Z ANALYSIS OF THE MEANS AND VARIANCES OF SEPARATE GRADE CATTLE SAMPLE POPULATIDNS Variable Estimated Means and Standard Errors of the Significance Degrees Hear of Grade Cattle Samples of F Test of (Pr‘X .i .i ) Freedoail Composite A B C ' a b c Sanole aeb+c Random Listed Listed (Heighted) Hithin Outside FSSU FSSU (Heighted) (Unweighted) Household Variables 1. Household type 7.59 6.71 7.75 7.67 .1424 231 income 11111111.. 1.17? 1.49) (.25) (.26) . 2. Household possession 5.94 5.73 5.69 6.30 .0119 229 income indicator (.10) (.25) (-15) 1-15) 3. Index of progressiwp. 10.5 8.4 9.5 12.3 .0000 230 3.53 (.24) (.45) (.31) (.36) 4. Calories consumed as 1.26 1.10 1.27 1.29 .1789 216 percentage of re- (-03) (~09) (~05) 1.05) quirement 5. Grams of protein 44B 37B 393 532 .0000 216 consumed (14) (30) (16) (25) 6. Amino acid score 179 137 201 166 .0003 217 of diet (7.3) (15.0) (12.2) (9.4) 7. Man days of labor 43.0 16.0 45.1 48.8 .1953 231 hired by household (5.7) (10.7 (7.5) (10.4) 8. Hours of labor/week 34.6 32.2 31.5 38.8 .0175 231 allocated to dairy (1.2) (3.7) (1-8) (1.8) enterprise 9. Litgrs of milk obtained 2.06 1.57 1.55 2.72 .1217 225 yesterday from all cows ( 28) (.48) (-20) (~50) Cattle Variables 10. Average age of all 2.85 3.13 2.78 2.81 .8410 506 cattle (years) (.13) (.35) (.19) (120) 11. Average age of grade 2.40 2.53 2.40 2.35 .9110 367 cattle (years) , (.13) 1-4?) (.17) (-20) 12. Lactation .111 produc- i 233 255 255 190 .2325 54 tion of Zebu cows (21) (61‘ (33) (29) 13. Lactation milk produc- 1106 1391 1129 1050 .6572 116 tion of grade cows (56) (349) (105) (33) (liters) 14. Daily milk yield of .36 4.5’ 4.44 4.27 .8991 179 grade cows at three (.21) i-6 I (.36) (.30) months post parturition 15. Kilos of grain fed to .60 .83 .41 .71 .0612 206 grade cow per day (-058) (~Z7‘) (-0811 1-031) .0. ...... .0... .............. d ........... J ........... J ........... J- ........... d ............... d ........... A-15 TABLE 1-2—9,12,31,13 Variable Estimated Means and (Standard Errors of the Significance Degrees Mean) of Grade Cattle Samples of F Test F of a ,".' _' reedom Composite A I B C (Pr'xa xb xc’ Sample a+b+c Random Listed Listed (weighted) Hithin Outside F550 F550 (Heighted) (Unweightefli Cattle Variables cont. 16. Age of grade cow at 2.85 2.77 2.96 2.77 .8340 86 first parturition (.10) (.18) (.19) (.12) (item) . 17. Number of months since 11.9 10.9 13.1 .6428 203 grade cow last gave (1.0) (2 3) (l O) (1 B) birth 1 111. Length of last 11m- 10.1 11.5 ' 10.2 9.9 .8336 118 tion (months) for grade (.4) (2.1) I (.7) (.5) cows ' 19. Age of last calf at 5.52 5.86 I 5.14 5.81 .7601 136 weaning (months) for (.31) (1.05) ( 42) (.48) grade cows 20. "with: between last par- 8.5 11.0 1 7.7 8.9 .3647 91 turition and first in- ( 54) (3.37) (.86) (.92) semination for grade cows 1 5 Household Simple Size _ Actual 261 38 ! 126 97 weighted“ 234 29 I 108 97 ’Heighted samples are used in the above analysis. Differences between weighted household sample size aid degrees of freedom are primarily due to the number of cattle per household, the number of cattle having the characteristic in question or don't know responses. bNumbers in parentheses are the standard error of the estimated mean. A-16 Subsample A is weighted in proportion to the occurrence of grade cattle households in the random sample and subsample B in proportion to the number of cattle-owning households indicated by the ten-cell leader listings. Subsample C is not weighted. Fer each of the nine household variables listed in Table A-2 the differences between the subsample means are significant at the 20 percent level; five of them at the 5 percent level. Fer five of them the means indicate little difference between subsamples A and B. The means of the other fbur suggest greater similarity between subsamples B and C, both being quite different from subsample A. In general the data indicate that the level of wealth and sophistication of the subsample populations increase as they become less random_-as would be expected. Interestingly enough, the socioeconomic differences in the respective subsamples does not appear to carry through to cattle management performance variables. The small number of degrees of freedom involved, especially for subsample A, contributes to the insignificance in the difference between the subsample means. But quite apart from this, the means themselves are strikingly similar. Both fer this reason and because sample A by itself is too small to yield meaningful results, grade cattle from samples A, B and C have been combined to derive population estimates relating to the cattle themselves. The combined sample is then weighted in proportion to the number of households possessing grade cattle in each of the first stage sample units as suggested by the combined random sample and ten-cell leader listings. Table A-3 indicates that the bias intro- duced by this procedure is minimal and, in any case, not statistical- ly significant. The differences between subsample A and the A-17 TABLE A-J CUiPARISON OF THE MEANS AND THEIR STANDARD ERRORS FOR ELEVEN CATTLE VARIABLES AS ESTIMATED FROM RANDOM SAMPLE A AND A COMPOSITE GRADE CATTLE SAMPLE V ‘ bl Sample Means and Their Degrees of SignificanceE— " ‘ ° (Standard Errors) Freedom P , x _ I '- a com A A+B+C 1. Average age of all 3.13 2.98 382 I .70 cattle (years) (.35) (.17) 2. Average age of grade 2.53 2.64 282 .80 cattle (years) (.41) (.18) 3. Lactation milk production 255 270 46 .83 for Zebu cows (liters) (61) (31) ‘ 4. Lactation milk production 1393 1206 65 .61 for grade cows (liters) (349) (104) 5. Daily milk yield of grade 4.57 4.33 117 .73 cars at three months post- (.64) (.27) parturition (liters) 6. Kilos of grain fed to .83 .58 135 .38 grade cows per day (.27) (.08) 7. Age of grade cow at first 2.77 2.84 76 .74 parturition (years) (.18) (.ll) 8. Number of months since 9.3 10.3 132 .69 graie cow last gave birth (2.30) (.96) 9. Lethh of last lactation 11.5 10.3 66 .58 for grade cows (months) (2.1) (5.7) 10. Age of last grade cow calf 5.86 5.17 80 .54 at weaning (months) (1.05) (.36) 11. Months between last oarturi- 11.0 9.0 56 .62 tion and first insemination (3.9) (.9) for grade cows Household Sample Size Actual 33 ' 251 weightedC 29 144 ‘Assuming equal variance. Differences between degrees of freedom and weighted household sanple size are primarily due to the number of cattle per household. the number of cattle having the charac- teristic in question or don't know responses. t‘Normal approximation. cweighted samples are used in above analysis. A-lB composite means are generally 10 percent or less with the varying directions of the differences suggesting equal management capabili- ties between the two sample populations. E. Statistical Inference Systematic sampling of first stage units is equivalent to sam- pling without replacement with probabilities proportional to size. Computation of the variance fer estimates derived from this type of sample are complex and impractical fer more than two first stage units. Sukhatme [1954] suggests that using the appropriate estimate fer two stage sampling with replacement at the first stage, and introducing the usual finite multiplier may be sufficiently satisfac- tory. Taking first the general case of sampling first stage units with replacement using probabilities proportional to size, Cochran [1963] gives the fellowing unbiased estimate for population means:1 (1) T=Ifi(7~| 1y, 1;") where i a the estimated population mean derived from a sample of first stage sampling units selected with probabili- ty proportional to size; n a member of the N first stage units in the population included in the sample; which is simply the mean of the first stage sample unit means. For the case of cattle means where the distribution of cattle does not 1Most of the notation used herein follows, with slight modifi- cation, that of Cochran [1963] in his discussion of stratified and two stage sampling. A-19 approximate the distribution of households, the first stage means need to be weighted giving the biased but efficient estimate: 2 n ._ (2) Yc ' z wciyci l=1 n 1:1 ”‘1 where the subscript c refers to estimates relating to cattle and "ci = the number of cattle type c per household in the ith first stage unit. Fer zebu cattle "ci is defined as: m J w = 2 CI j=l ‘15 “‘1 and far grade cattle: m 9 w = p 2 g . ci 1 9:] ‘13 mg th where cij = the number of zebu cattle of type c in the j h household of the it first stage unit; m. = the number of households randomly sampled in the ith first stage unit; p1 = the proportion of households in the ith first stage unit having grade cattle as indicated by the ten- cell leader listings; - the number of grade cattle of type 9 in the jth grade cattle household in the ith first stage unit; m = the number of grade cattle-owning households sam- pled in the 1th first stage unit. A-20 The extent of bias in equation (2) is proportional to the error in the wci' Fer zebu cattle, the estimates of the "ci are derived from the random sample and equation (2) is assymptotically unbiased. Fer grade cattle the estimate would be assymptotically unbiased if the pi were estimated on the basis of the random sample rather than the combined random sample and grade cattle listings provided by the ten-cell leaders as they are. However, the sample size per first stage unit is sufficiently small in 26 of the first stage units (n59) and the distribution of grade cattle sufficiently narrow and clus- tered so estimates of the pi derived from a random sample of this size would no doubt be highly unstable. Fer this reason we define p1 = l>911 p where P911 8 the proportion of households having grade cattle in the ith first stage unit, according to the combined random sample and ten-cell leader listings; Pgr 8 the proportion of households in the population having grade cattle based on the random sample; Pg] . the proportion of households in the population having grade cattle based on the combined random sample and ten-cell leader listings. This should improve considerably the accuracy of the estimates of population totals over either the unadjusted P911 or the first stage unit proportions actually found in the random sample. A-21 The variance of (l) is given by 3 (3) Hi) =%:fig('fi -i) +Jriz-M-g—flSZT where Mi = the number of elements in the ith first stage unit; Mo = the number of elements in the population; 7, = the population mean of the ith first stage unit; f21 = the second stage sampling fraction in the ith first stage unit; and M s .2 - j:1 (’11 ' Y‘)2 21 An unbiased estimate of (3) is given by: (1-',]-) n (4) V (T) g'fi-TfijTT'z (yg ‘ 2 -< II) 1 where (l-fi) is the finite population connection factor fer first stage sample units. For cattle, the variance of equation (2) is approximately: 1:. N C = N c (l-f ) s 2 g 1 i -’ - 2 I 1 2C1 ZCT (5) V(Yc) c 2 z 2 (Ya. 1c) + z 2 " ° 1 mco ’1ch where C1 = the number of cattle of type c in the ith first stage units; Co a the number of cattle of type c in the population; 21 a the probability of selecting the ith first stage unit. A-22 An estimate of equation (5), slightly biased, is given by Yateszl n n 2- —2 c) = (lefiln 1:] we, (yci - Yc) - mo epog1mco mzpa Peeves samN com mzpe> mo uses1mco mpuacmm om.P mP.F oo.p w.m m.m ofiv mum om.~ mm. on. m.w o.m omm omp oo.P mm. om. w.m m.m 0mm mmp om. me. mm. m.¢ m.~ cop mm om. mm. om. P.~ m.P 05 cc mvmt—w mmmOLU _.n_ :nmN mUMLw :nmN mDMLO :nmN move: o_ooau pcmpm>wacm :nmN u_:u< coupe: ago +om oMV1¢m dNVtNF vaoo mV1Po. Ammxv xmo cod .mcou .xocaa< mmeco>< Ammxv ugmwmzm>w4 mm< mo mcucoz omH2cm =m- pgsc< oz< mmhpmo no onhmzsmzou mhH4 mo< ~1u umme no: oume_uno uncommon cue.“ auvuuecn we Aucoaaoc» co women nouaqsnu oco m.~ o_noh a? macaw.» oucwn zoom on =o_uopoc cw Loam: mc—nuom do zucuaaoem co comma umuaas~ .upaoconouc uo: w .uouoc om_:coguo unopcs o—QEom o_uuou ocean oupmoqeou cw mzou mcvuauocq no» mo amw com o=~o> umoeo> oaocm>< .uouaum omvzcoguo mmo_== mute; caxvs c. opuueu ocean to» o:_o> mocco>< .ucm: voxns :* we can zoom omega>m cmozuon umuupoqcoucm u u a .opaeom Eocene soc» case». me. ~. N. m. P. o. manna—u uco mcwxocam co.P mu. mm. mm. o. m. Loan: asexucou ape: uppueu con goo: can umucoaxu coneg mo mono: o.a o.~ m.~ ~.P o.p m.o . canoe can mcwaaru mo mucoscocm o.~, e.m a.m m.e e.e e.e moo: can open meeeooc Co adeosooea m.op o.~ o.~ o.o m.m m.¢ gum: can cope: meadow» we zucoaaocd e.. ~.~ e.F 9.. o._ 8., oaecacooeo e. .3.8 cc Longsz meowcmgam vmmecm>< we we ommoco>< apnowue> oueco1_—< onecu1ano~ anoN Po>m4 ”cosmonaut ace mm_cacoucm omh mzo~m<> mom moHu~m4am< oz< mmh<3 ozH>mm¢mam . ..NNN:._. "Na voc..ou N. u:- . sNoA og~ so. x.._.aonosu ..>.>s=N ox. o. «so.os .o .N soanouoo co capo: u.ea o—uuNu as. o. Nsosos N . ...Nxau >0u>=u .coNuosocou - c.3a.) opuuou —-39.>_vc. 0:» cu «souos . .q gucol ac.sav ....ca as. so as... ..N: c—uuau as. no Nsonos :u we. so.» 0:» No zwwou as» o» ns-uos : do: "an vac.»uv u. m.:» .opuuau Nu us» so. a... No ugucoc n.a: o—uuou o» «sou-s :u ssozzv .—.o 0.9-— ...u... g... .g. ... . ...» ... ... a..._m.ao.. N.......a ... .. ..N.x ....c... .. ... .o .... ... >N¢m we. .eo...so¢oo cu. as. c. —.I_¢N g». us» so. u so.» 0:» co..Nsocoo 0:. cu asosos N osogzu .o—uI a a. s—ao was.» - tuna .....o. . .. .... ..... . .... NNN. Nu. .........N. Na.— uuu acocuNo»c. uo oa..> oo.>..m NM. .nn. .... ... Mm: ... .... mm: ... up. .D. m. N. N. N. NN N. .. .N a. N. N zu. .........N. .N N. N. NN .. .N .N .N NN N. o .N. ......N ...a .o ......N No. ... ..N . so. ..N......N. .N. NNN oNN ..N ..N ... ...-..N .... .o ....N o. a N . N N . N N . c 2......UNuo .... us:c..counupna u4nN.. .... .. ... .. .N... ...N ...... . N. .. N. .. N. . ......Nz. .... .. .... .o ....N: .N. NN. . .. a. .. .. N. .....N.. ... .... .o N.. N.. NN. .N.. N... .... N... .N. NNN. N.. w....... .... ... N....... .. ......a. N.. N.. N.. N.. N.. N.. N ......z. .... ... N.......o.. ........x . ..N. ......x ..N 3...: .N ......N. ....> ........o.. N. . N . . . . N . . N .N..._.N..N oo.>_om —N>.>s=m ..... ... a... v...:nv< soauanvacz ....»sam u:..> oo~>pom ugNN . N. . N....Nz .. N. . N...N. .NN N.. .N.. .... NN.. ...... ..N. N.. N N....z ..N: 3 s3» .p son-38 Ea. 2. has. swim“. N. . . . . I.I . . . . . N .255... \U—Qm LO. ¢ nouaaav< vaunaavacs pa»...:m «n.a» «ONNNNm vo::_.coun3~.a man.>s=. cg. NN.—N No sNNN so No.9—N—Naua No saw» cg. c.N ...N ..... a... N..... ..N .. .N.N=N. .... N. ...N.>N. . . .N...Nx . . .N<.N. .NN N.. .N.. .... .N.... ..N. N .N...x N—qso. «u s..» .N .999 use. .N. ..N .. . . NN.N2N. .... N. NN.N..N. . N NN...N: . N NN<.N. ... N.. N.N. NNN. . .N.... ..N. N . ....N: ..Neau .N so.» .— .>Nz cue. NN. . . , NN.:N: ... N.. NN..N:NN ..N NNNN: ... N... .... N.. ...NxN. .N... N.. N.. .. NN.N.>N. . N. N. .. .....N: .. .N .N .. ....N. .N. .... NN.. N... NN.. .... NN..... ..N. .N. .N. N .....x ...-c. a. sac» .— .NN. ego. -. .N... N.......ou. N. N N . . . . N N . . N 2N....NN..N CONN—2s. 23:3 l x»... so. ¢.>s:m wanna NN....” unachoul~na 3o: \J-EB NN..NNN ..N N.N. NN.N: ..N. N . N.E.. NN. N.. N.. N.N NN..>:< N.N. .NN. NNN. . NM..Nx NNN. .N. .N. N ...: N.. N.. N.. N.. N.N ....NNN N.N. ..N. ..N. NN.. ...Nx ... N.. N.. NN. N . .NN..x . ... N.. NN. N.. N.N N....N. N.N. .NN. ..N. N.N. N...x NNN. NNN. NN. NN N N....x ... NN. NNN N....N. .NN. .... N...N NNN. N.. N N..... . N.. N.. N.. N.. N.. NN. N.. NNN .....z. N.N. ..N. N.N. ..N. N.N. N.N. NN. ...N: NN. .N. NN. NN. NN. NN. N.. N .....x ... N.N N.N N.N N.N N.N N.N .....Nx4. .... .. .N... ...: .N....< N.N. ..N. ..N. .NN. NN. N.. . ....z. N.......N.. .N......... N.......: N.. N.. N.. N.. N.. N.. N .....5 ....» .... ..N. 3:238. 5:33.. .....N ....... N. N . . . . . N N . N .N........N NN.. ..N.....x.. 333.. < ...—(u was: a...» 1:: unnamzupxu 3:3 n... 33» 355433 aux—x :2: o.— < «a. 35:5 059 ha ~3<> wu<>._33 .1-59 60:. 2.2 so...) «pal—co .9 ...—a» .5 .0 .75-5‘ ... ... .....z. . ....x u u u c N .N.... ..Nx. 3:339... 33...?! 323.8- 2. 5...: 3 N..—s8 a... No .583. 2» a. ...-s. v... 8. £37333 .0 .32. :3 N ..N ...—2. 2.» a. 2...: ....N ... N.......... ..N...N. . .... ....z ...»... . ....x . ..Nx. a on m- o8. 3o. o— NN .. N ....N ....) ...-... .N .N... .N..... N. .N N ....N ....N ...-... .. ....» .N...x< .N.Nx \ .N..Nx .N. N . N =N.......NN coaczcoQIAA. 32h J-IO N.Nxox N N N N N..N ..N. NN.N N... N.NNNNNN 5.... N..... :3: N.N N.N. N.N. N.N. N... N.N N.. NN. N.NNNNN .N.... N... N... N... N... . NN. NN. NN. N..NNNNN N N. N. N. N. N. N. . N....Nx NN NN .N N. NN NN N. . N...NN NNN NNN .NN. N... N... NN.. NNN. N.N. NNN. NNN. N....N N.N. NN. NN. NN. NN. NN. N.. N N....t N..... .N N... .. .NNN N.NN N.N N.N N.N. N.N. N.N N.. ...NxNN .N.... N... NN. NN. NN. ...NNNNN N N. N. N. . N .....Nx N. N. NN N. N ....NN ..N NNN NN.. .NN. NNN. NN.. NNN. ...NNN N.. N.. N.. NN. N ...NNx ...: n. N... .. N.NN ..NN ..N N N . N N N . N N N .NNN: N N . N. N N . N N N .NNNxNN .NN ..x NNN ..NN .NN: N N. N. N. N. N. N. N. N. N :95. .N... N.. N.. N.. N.. N.. N.. N.. N.. N.. ..NNNNN N N. N. N. N. N. N. N. N. N ....N: N.. NN. .N. N.. NN. .N NN NN NN N. ...NN N.. NNN NNN. NN.. NN.. .NN. NNN. NNN. NNN. .N. NN. NN.. ..NNN . N.. N.. N.. N.. N.. N.. N.. N.. N.. N ..NN: . N..... 5:... ..N 5.3.. NN :22: Na... N....N.NN.N N. N N N N N . N N . N .NN.NN..NNNN 093.3 33:3 all: \o—uw so» ¢ vouNaqv< voNNaqvaca \ «N.N, oo-».~m n.a: < 2 .33 N9: 93 as: n.a. a. no.5. 23:: < no 353: awe—:3: 3:: 3.5.5:; 3:3 358-3: 3:: :2: 0.. 4 go... on“: 0359.; v4. 33— J-l] N .NNzN: N.N .NNNNNNN ..NNN .NNNx N.N ..N. N.N. ..N .NNNNNN .N... N.. N.. N. .NNNNNNN N N. N. N .N..NNN NN .N N. N .N..NN NNN NNN. NN.. NN.. .N.. NNN. .NNNNN NNN. NN. NN. N .N.NNN :3: K ..N.» .— 23.. Eon .~N N.. N.N NN. ...NNNN .NN.. NN. NN. ...NNNNN N N. N .....NN .N N. N . ....NN NNN N.N NNN. NNN. .N.. ...NNN NNN. N.. N ...NNN N.N.-N.. 3 ..N.» .. .Nuo foo ..N 9. 9N. 9N. 9. !N NJQNN .N.... N... NN. NN. N.. N..NN>NN . N. N. N. N N....NN N. .. NN . NN N N.N.NN NN. NNN NNN. NNN. NNN. NN.. N.N. N....N N.. N.. N.. NN. N N..NNN 2! N. ..N.» .N N... Eon N.N 2:; 3:338. N. N N N N N . N N . N NN....NNNNN cool}. .NN.)...» l \O—Qm has “(up "33.2... 3.3325 03—.) . wag—am coacScOuIvuo 3.: J-12 .~.u we. ..a NoNNN» o» N-NocuooN ..N NNNNNuNN: No coNNNNNNov a NNN. N. .N4oxnu .N.. .NNNNNNN N .N..NNx N . .NN.NN NN NNN NNN. .N.NNN NNN. .N.NN: ...: .N. .... .. ...N. ..N. .NN N.N N.. NN.NNNN .N.. N. NN.NN.NN N N NN...N: N. N NN..NN N.. NNN NNN. NN.. NN.NNN NN. N NN.NNN ..N: .9 .N.. .. n.aa cNoo -N .N... N....N.NN.N N. N N . NN.NN..NNNN ooNN.Nm .NN.».N» III! N...N NN. N.N. NNNNNNNN NNNNNNNNNN oa.~> oONN—om uo:c.ucouuuoua u4¢

x< NNN NNN NNN NNN. NNN. NNN. .NN.: NNN. NN. NN. N .NNNNz .NN NN. NNN .....x. NNN. N.N. ..N... NNN. N.. N ..NNN: . v: NNN NNN NNN NNN N....NN NNN. NNN. NNN. N... N.N.: N.. N.. N.. NN. N N.NNNN N.N NNN NNN NNN NNN NNN NN. NNN N....NN N.N. NNN. NNN. N.N. N.N. NNN. NNN. N.NNN N.N. NN. NN. NN. NN. NN. N.. N N.NNNN . N.N NNN NNN NN. NNN ....NNN NNN. NNN. N... NN.. ..NNN N.. N.. N.. NN. N . ...NNN ..N NNN NNN NNN NNN NNN NNN NNN NNN NNN ....N. N.N. N.N. NNN. NNN. NNN. NNN. .NN. NN. N.. ..NN: N.. N.. N.. N.. N.. N.. N.. N.. N.. N ..NNN .33 N. N N N N N . N N . N ..N» .NN.....NN. 353.. < 53 :1: 2: 3:: 3:25;» 3:3 33-3% 3:: :2: o.— < so; 35:: ago no 2.3; 3:44». u...» no zozgzg a... 32» J-M .N-.. N.N-N. 3 8.2.8. NNN 3.5.52... .3323... N .3..- N. N. N. N. NN N. .. NN N. N. N NN.N .N..: ...-..N No .N... .N....N NN NN .N NN N. NN NN NN NN NN N NNNN .N..: N....NN N. .N... NN. .N..Nz< NNN. .NNNx NNN. .N.NN: . _ .NN NNN NN..N:< .NN. NNNNx NN. N \ NN.NNN NNN N. N N N N N . N N N N NN..N.NNNNN NNN. coac.ucou|m-—. 39¢.— APPENDIX K APPENDIX K ENTERPRISE BUDGET SUMMARIES FOr each year, cattle unit months of life, cattle unit months of production period and months of milk production as simulated fbr the separate enterprises are used to calculate costs and benefits fOr the various alternatives considered. The monthly value of manure on the returns side, and labor and cash inputs fOr normal growth and main- tenance on the cost side are then multiplied by cattle unit months of life in each year to get returns from manure and operating inputs fbr maintenance and normal growth. Cattle unit months of production period are multiplied by the monthly value of labor and cash inputs to get the cost of each fbr production. Sales of live animals and the salvage value of animals which have died and of livestock in the herd at the end of the investment period are taken from the simula- tion summary sheet fbr each enterprise. Tables K.l through K.8 present the ten year budget summaries for the eight enterprises analyzed. K-l KN-Z NNN NNN N.. N.N .N... NN. NN NN. .NN. .. N.N ..NN. .NNN....NN. NNNNNNN NN. N.N. .NNN. NN. N.N N.N NN. NN. NN. N.. N.. NN. NNN N.NNN N.NNN nwmmu .NN.. N.N NNNNNNNNNN. .NNNN o o 0 NNNEN.NaN ..... ..NN. NNN N.N..N NN. N NN. NN.NNNN NeunaNNNN. .NN.N~u wwwhl n.bwl Ihhhl INHMI IBFMI Ihnnl ..mmn: -nnn. n.n= m.n mum on. . mhmou 0:..«auao 44.0. 0 o 0 0 0 0 0 0 0 0 0 0 Nuanc. NNNu NNN N. .. an 0. 0n v. o~ N. 0. a. 0 Nona. co..uauo.g Nv. o n. N. N. a. o. o. a. v. ~. a «cane. gmou NNN. no. Nev 00. 00. Nov am. an. own ~on mon «N. None. garage tea oucNNNNNNN: oc..~coao .NNN....NN. N.NNN S.E.: INF Imp. INK: «El IolmuN IGNI BE! Lmhl LP IEI Imp N532. .32 NNN m. «N NN on N~ 0N .n NN o~ o. 0 N.N-.c< uNoo .0 ooN>.~m N.N 0N o~. mm 0 0 00 0 0 0 0 0 N.N¢.c< «N.. .0 NN.NN Noon ~n. n~n NNN NNN .~n ~Nn 0mm .mu o0~ 0N~ o~. assent 0No. on on. NNN 0. ..N no. N~n 0m ... mom 0 ...: .NNN....NN. .NNNNNN 0N.0 00.. NN.. m~.0 00.. a... 00.0 0N.0 00.. 0v.m 0 so.» c. co.au:vocs N..: No Ncho: No.~ 0a.. m0.m NN.. mn.0 0N.~ 00.. 0o.~ 00.. 0..0 0 .N.. c. vo..uN coNNuavoNN .0 NNNNN: N.N: N.NNNu N.N ...... 0.0. 0.0. N.N. 0.... N..... ..N. N..... o... 0.0 .3. ... 3... .9 2.3: N..... :38 .~.. 0~.. a... v... NN.. N~.. .N.. .N.. 0... 00.. 0.. 0:” can» an 05.: 0. NN.:0 N.N.-u N.N.ocNLNN .NcoNNNNaaeou m.<.0. 0. o 0 a 0 a v n u . 0 :0..N.uumu0 ¢ l’loi'il HI I... >420 cum—u: a<.».z. No um.00 000 00 .0. 00 0 00. 0 00 0 0 0 0 0.05.04 .N.. No NN.~0 0000 00. 000 0.0 000 .00 000 000 .00 00w 00~ 0~. 0.000: 0.00 N0 .0. 000 0. 000 00. .00 00 0.. 000 0 ...: .000....:0. 00.0.00 00.. 00.. 00.0 00.0 00.0 00.. 00.0 0..0 00.. 00.0 0 .00. 0. 00.00000c0 ...: .o NgNNN: 00.0 00.. 0..N 00.. 00.0 0..~ 00.0 00.0 00.. 0..0 0 ..N» e. 00.N00 co.NNavo.N .0 NNNNN: N.N: N.NNNu 0.. 0.0 0.0. 0.0. 0.0 0.0. 0... 0.0. 0.0. 0... 0.0 .00. N. 0... .o NchN: N.N: N.N.Nu 00.. 00.. 0... 00.. 0... 0... .0.. .0.. 0... 00.. 0.. 000 ..N. .0 0.0: c. NNNca N.NNNU N.NNNINNNN .Nco.NNNNQENu 0.0.0. 0. 0 0 . 0 0 0 0 0 . 0 20....00000 «<0. 0.0: < 0. 04x. 00.0. 0. 00030.000 000 0.0.00 0.0 00.000: u:0lauo>¢. .0..000 Ibmhl 0.000 02..<¢000 0<.0. 0 0.00:. 0000 0 .0000 00.00300L0 0 0.00:. 0000 0N. 00000 0.3090 0:. ooeocouc.ax 00..-.000 .nog....00. .0000 0.... .MBIIMBI Inn... EEEEFFFE 053.0028. .0. 0. 0. 0. .. n~ ~. .. on a. a. o ...-.=« u..o .0 «.....0 0.0 00. 0.. 00. 0 0.. 00~ o 0 0 0 0 0.0l.=< 00.4 00 00.00 0000 .0. ~00 0~0 000 ~00 .00 000 000 .~0 00~ 00. 0500-: 000~ 0.0 0.0 ~0. 000 0N. 000 0~0 00 0.. 000 . 0 0..: .00:....g0. 00.0.00 00.00000. 05.00 00.000u0. 0:09.. 00..-090. .05.. "00°09 0. 389 C 30.0 g... 0. co..uauos0 0..: .o 00.00: «.009 0. :95 .. :00. .... c. 00.... co.»uau¢g. .0 .0.:0: n.a; .....0 «00.0 0. among 0. 30.3 C :00. .... c. ...4 .o .guco: u.c: 0.00.0 .. 00.. 0... 00.. 0:0 500. .0 use: 0. 00.00 c.0000 O O p U! N O O 0 ~00 0 0° . 00° 0 0°C I 000 I 000 00 0 0. 00 . 00 - O O 00.. 00.0 00.0 00.0 00.0 0~ 0 0. -c: ~¢= . o . -<3 <3 O '00 '00 ~00 000° 00° 00° 0 PK” h0— .U’ 0&0 Q 0 F O 0 ~0.. 0 .. N .. N... .0.. ~0.. 0... 0 . whence-La. ..co...u=maou 000.0. 0. 0 0 N 0 0 0 0 ~ . 0 :0..0.00000 000.0000 .0: 000: a. 000: 000.0: .0 000000.. =..: “0.0.00.20 00.._4 .o mo_~m 2mm .2 z. 2. 3... 3n 3” 2" on an on 3. 22...: 2: «3 Sn 3. 3» SN on an 8 :— 8.” o . .2... 322.23 2:3.“ . . - - . . .. - . - . ...—:3: 2.5 n a o o o a s . n u . cowuauua— Vcoueu oo.— m~.o o o o o o o . o o o co.u~uu-— «as.» "uuogu - 2.. 8." 2.0 2.. o a: o o o o o :95 c o ma.o no.9 cm.c o~.n o oo.v 95.0 ou.— cc.» 0 :aoN ...: ... 338.8... 2:. 2 2:3: 2.» ...... o o o o o o o o o :25 N. «0.. m... om.o «n.a mo.o m~.n m~.— o o o o «nogu fl. o no.9 on.. om.o an.» o oo.v ao.~ 00.. o_.¢ o anon .3» ... 3...... 8.8.62. 3 2:3: :5 :38 0.: ..c o o o o o o o o o «mogu n‘ N. ..N n.» ... a; ... a o o o o :2. N. a.v o.p— n.a. n.a. n.a n.a o.op o.~ n.a v.. o uuogu .. 0 ~.— a.n ~.m o.» ..N a.“ 5.. n.a o.op o.o ano~ ...: ... :3 .o 22.9. :5 2:8 o~.— «5.. no._ -.— p..— ~v.— am.— no.. on.— n—., a.— gem so.» a. who: c. nu_c= o_uu~u s agouuaogug ..co.u~ummeou 3:2 2 a a N o m c n N p o 85:33 aux—aouu box can: a— ua(x Hank‘s no uuzumuac x»_: um—uaxupzu udp~ zu» v.2 wand» Ka-6 NNN» .... .0.. .NN. NNN “NNN veg NNN . __m.— ... N.N ..oNN. .»N¢.N_.gm. .=L=N.N No: onnn .oon.. NNN a.» non NNN .No NNN NNN ooN NNN ocoN apnou 4c~ pou.q~u HQ: kn: k5: IBP IMmP IEI IBIS Ink: IMF IpEr Ink: ImKI “58 «552° 3:: NNN NN oo— oN NN Na 0N .m NN. Nu no, o .N=N¢_ ‘,.u mmN .N NO. ON NN No .N Na NN. Nu no, o ‘09.; . co_uu;u0Nn o... .m o.— N_. NN. ON, mp. N.N N.N .N. co .. .uan=_ ‘,.u ..mm oNN a.“ "on Now cow NNN omm Now man No. .oN ‘05.; nuxogu vca quencouc_ax ocvu-Loao .¢o=___.gm. nufiou “mmmh INmNI Inna“ Imbbm ImnN» Ipmnm Immm» Innnu .Nmnm .INNN» INmN» Imnps azaahwa NNNON .on an NN an an NN NN o. .N NN o— o ...a.=< u.~c .o oo.>_.m NNN. o N.N NN. Na ..N a on, NN. . .NN o o ._.s.=< .>_4 .o .._.m NNNN NN. «an ...I .N. oN. .o. .NN Non NNN NNN N.. .L=¢.x .o.o_ N.N Now, No__ Nmo oo_. o”.— NNN N.N. ..N NN_. o .N.: Aaczpzm. 2:33. mm.o on.N o~.. ON.. mo.N oN.m oo.— a o o o .co.u.uu._ “cone.“na. we. u..gN om.o oN.N 0N.. an." o o 0 ON.» a o o co.u.au._ vacuum o cN.N «N.N NN.. “..N o o o ow.n on.“ o co.u.uu._ “NN.. Loo» .3 6030390.; «Ct ~o 23:9. o..N «n.a. oo.N oN.N «N.N oo.N o—.n 0N.NN oN.m om.o_ o N..» :. no.5.N co.uuauog. .0 ngucox .Nca .Nuu.u o.a N.N, N.N. n.0N o.ON ..N. N.N. N.N. N.N. o.“— N.o N..» a. ...N .9 .Nuco: N.c= ..Nu.u mm.— om._ Nm._ 0N.. .0., “N.. Nm._ on._ an.. 0“,, oN.. saw g... N. uN.= a. ua.c= ..uo.u agouoe~gcm —oco_u~uaqeou mNx. .«.o. QON o oo. oo. .coeq.=au .oom. .o.ON. o.N. .....u 0.0 o oNo oc..=oz «co-.N.>:_ ......u wmmm. Immuu .Iomm. Imwm. Immm. .Ip.n. .Ipmm. In... Iwmm. INN”. Inhm. AINme ”.mou as..‘gugo .<.o. .N.N ... N.N N.N N.N ... om. ..N mnN NNN ..N o ..aoc. ...u ..N. N. 0N. NN. .9. «O. N.. ... an. .N. NN. o .09.. co.uuauoum .00. N.N Now No. ova ova 0N. on .m. o.. .N. ca. «Nana. ...u .N.. o.N ... ooo mom NN. ..N .0. NN. o.» No. .ON .0... garage vga oucocouc.ux oc.u-Looo ..og....gm. ...ou Mwmhm NN.. 0..“ mm.“ No.“ prnm Ibmm. Immpm momm mm.“ «on. mm. mzN:.Na .<.o. N.. 0N .N .n 0N .. N. .N «N on N. o ...s.:< o.~o .9 .N..... No.N O can ... mam ..N o a.» o N.. o o a..a.=< .... .o ...-a com. om. om. NN. o.. N.. ..N mon mon ..N NNN N.. n.ac.x ..m.. .o«. c... on». ..NN a... a... .NON mm.N O... .N.. o ...: .«ag.....m. .c.=..m 3.. o... 3.. mm; on... 3.. ..N.» o o o o .5333— 33333 2.- 3.... a... o... om.. o o o o oN.o m..o o o co....u.. vacuum m... .0.. a... oN.N o o o o o... oo.m o =o....u.. .N... .... .. co..u=uogg ...: .o ...co: «9.. o..o mo.a o..N. mo.N on.» 0.... o..o. oN.o. cm.o. o ..o. c. vo.N.N co..uauosN .o ng.=o: ..c: .....u o.a ..N. N.ON a... n.a. N.N. N.N. .... o... e.m. N.. L... c. .... .o .g.:a: ..c: .....u om.. .N.. .o.. o... .N.. .N.. N... NN.. a... an.. 0N.. ac. L... .- ago: c. u..c= ...... €32.38 .3023338 m.<.o. o. a N . o m c n N . o 20....uumuo ¢<.. .zuzuoN. .<.o. N... o .N .... €93.33 .NNN.. .NNNN. oo.. .....u o.o. a 3.0. ac.N:oz uc’unoii p: ...-u NN.N: ImeI INNNN INNNI INN»: Inn». Inna. Inna» ..m.n. INN»: INN»: I..mI N.Nou N...¢NNNo NN.N. N.N .N .N. N. an N.. N.. No. NN. NN me. o n.aac. NN.N ..N .N NN. NN NN . No. No. N. N.. .N no o .0... . co.NuauogN NN.. cN NN. N.. NN. ... NN. NN. NN. NN. .N .. «Nana. NN.N N.N» NNN NNN NNN NNN N.N NNN NNN NNN N.N NNN N.. .0... garage we. oucacouc.ux oc.uagoac 73......3 3N8 N.N.. NNN NN.N Inna" n... NNNN N..“ INN». NNNN up“. INmN. NN. NzN=.NN .<.o. NN. N. .N N. NN N. NN .. NN .. N. o ...-.=< u..o No .N....N ...N o N.. NNN o ..N. o .N N.N o o o ...-.=« .... .o ....N N.N. .NN NNN NN. N.N NNN N.N N.N NNN NN. NNN N.. ..a:.: N.N.. NNN .oNN .... NNN .... NNN. Noe. N.N. ..N .N.. o ...: .Nae....gm. NcL=NNN o... NN.. NN.N NN.. o NN.N NN.. o o o o Nco.u..u.. NN.:N.NN=N ac. NN.N. a o NN.. NN.. NN.N o a NN.N o o o co....u.. cacao. o NN.. NN.. o NN.N NN.N NN.N o NN.N NN.N o co....u.. .N... .... c. co..u=uo.N ...: .o ...co: NN.N NN.N. NN.N NN.N N.... NN... NN.N. NN.N. NN.N NN.N. o .... c. N..... co..uauog. .0 Ng.cox ..c: .....u ..N. N.NN ..NN ...N N.NN ...N ..NN N.NN N..N N.N. N.. .... c. a... .o ...:o: ..c: .....u NN.N NN.N N..N .N.N .N.. NN.N NN.N NN.N .N.. NN.. NN.. NNN .... a. N..: c. «..N: .....u NgouoeNLNN .Nco.NNN:aamm N.N.o. a. N N N N N N N . o NN....NNNNN m «(my cum: z- u4c~ .ou.nnu .NNN NNN NNN. NN.. N... NN.. NNN N.N NNN. N.. NNN ..N N.NNN NN.. xu» n-a ugu