85732 INVESTING IN THE LIVESTOCK SECTOR Why Good Numbers Matter A Sourcebook for Decision Makers on How to Improve Livestock Data © 2014 International Bank for Reconstruction and Development / The World Bank 1818 H Street NW Washington DC 20433 Telephone: 202-473-1000 Internet: www.worldbank.org This work is a co-publication of The World Bank and the Food and Agriculture Organization of the United Nations (FAO). The findings, interpretations, and conclusions expressed in this work do not necessarily reflect the views of FAO, The World Bank, its Board of Executive Directors, or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. The designations employed and the presentation of material in this information product do not imply the expression of any opinion whatsoever on the part of the Food and Agriculture Organization of the United Nations (FAO) concerning the legal or development status of any country, territory, city or area or of its authorities, or concerning the delimitation of its frontiers or boundaries. The mention of specific companies or products of manufacturers, whether or not these have been patented, does not imply that these have been endorsed or recommended by FAO in preference to others of a similar nature that are not mentioned. Rights and Permissions ©FAO/Giulio Napolitano The material in this work is subject to copyright. Because The World Bank and FAO encourage dissemination of their knowledge, this work may be reproduced, in whole or in part, for noncommercial This report is based on research funded in part by the Bill & Melinda Gates Foundation. purposes as long as full attribution to this work is given. The findings and conclusions contained within are those of the authors and do not Any queries on rights and licenses, including necessarily reflect positions or policies of the Bill & Melinda Gates Foundation, the Food subsidiary rights, should be addressed to World Bank and Agriculture Organizaiton of the United Nations, the World Bank, the International Publications, The World Bank Group, 1818 H Street Livestock Research Institute and the African Union-Interafrican Bureau for Animal NW, Washington, DC 20433, USA; fax: 202-522-2625; Resources. The designations employed and the presentation of material in this information e-mail: pubrights@worldbank.org. product do not imply the expression of any opinion whatsoever on the part of the above Cover photo: © FAO/Simon Maina organizations, concerning the legal or development status of any country, territory, city or Report design: Anne C. Kerns, Anne Likes Red, Inc. area or of its authorities, or concerning the delimitation of its frontiers or boundaries. INVESTING IN THE LIVESTOCK SECTOR Why good numbers matter A Sourcebook for decision makers on how to improve livestock data Ugo Pica-Ciamarra • Derek Baker • Nancy Morgan • Alberto Zezza Carlo Azzarri • Cheikh Ly • Longin Nsiima Simplice Nouala • Patrick Okello • Joseph Sserugga World Bank Report Number 85732-GLB ii  |  Investing in the Livestock Sector: Why Good Numbers Matter TABLE OF CONTENTS PREFACE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . III ACKNOWLEDGEMENTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . IV TABLES, FIGURES AND BOXES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V ABBREVIATIONS AND ACRONYMS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VII INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 PART I. DEMAND AND SUPPLY OF LIVESTOCK DATA: GAPS AND ISSUES . . . . . . . . . . . . . . . . . . . . . . 4 1.1 THE BASICS OF A PROPER LIVESTOCK STATISTICAL SYSTEM . . . . . . . . . . . . . . . . . . . . . . 4 1.2 CORE LIVESTOCK DATA AND INDICATORS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 1.3 DATA AND INDICATORS FOR EVIDENCE-BASED LIVESTOCK POLICIES AND INVESTMENTS . . . . . . . . . 18 DATA COLLECTION SYSTEMS AND LIVESTOCK INDICATORS: GAPS AND PRIORITY ISSUES . . . . . . . . . 30 1.4  PART II. METHODS TO IMPROVE THE QUANTITY AND QUALITY OF LIVESTOCK DATA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43 2.1 COHERENT AND COMPREHENSIVE INFORMATION: DESIGNING A LIVESTOCK QUESTIONNAIRE FOR AGRICULTURAL AND INTEGRATED HOUSEHOLD SURVEYS . . . . . . . . . . . . . . . . . . . . 43 2.2 IMPROVING LIVESTOCK DATA QUALITY: EXPERIMENTS FOR BETTER SURVEY QUESTIONNAIRES . . . . . . 51 2.3 PHYSICAL MEASURES OF PRODUCTION FOR BETTER STATISTICS: THE LIVESTOCK TECHNICAL CONVERSION FACTORS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59 2.4 INSTITUTIONAL CHANGES TO IMPROVE THE QUANTITY AND QUALITY OF ADMINISTRATIVE LIVESTOCK DATA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67 PART III. LIVESTOCK DATA FOR DECISION MAKING: EVIDENCE AND EXAMPLES . . . . . . . . . . . . . . . . . . 78 3.1 ESTIMATING LIVESTOCK NUMBERS: EXAMPLES FROM COUNTING ANIMALS IN WEST AFRICA . . . . . . 78 3.2 PEOPLE AND LIVESTOCK: LIVELIHOOD ANALYSIS USING THE LIVESTOCK MODULE FOR INTEGRATED HOUSEHOLD SURVEYS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90 3.3  DATA INTEGRATION TO MEASURE LIVESTOCK AND LIVELIHOODS IN UGANDA . . . . . . . . . . . . . . 98 3.4  COMPLEMENTING SURVEY DATA ON QUANTITY WITH QUALITATIVE INFORMATION: THE MARKET FOR ANIMAL-SOURCE FOODS IN TANZANIA AND UGANDA . . . . . . . . . . . . . . . 105 3.5 CONSTRAINTS: COMBINING MICRO-DATA WITH FARMERS’ VIEWS . . . . . . . . . . . . . . . . . . . 116 RECOMMENDATIONS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127 REFERENCES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129 QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations PREFACE | iii PREFACE L imited access to quality data is a major constraint to statistical issues, it does represent a unique document for a economic development, making it difficult for public number of reasons. To begin with, it is possibly the first doc- and private actors to design and implement policies and ument which specifically addresses the broad complexity of investments which maximize economic growth while being livestock data collection, taking into consideration the unique smallholder inclusive. This is overwhelmingly the case for characteristics of the sector. Indeed, in most cases livestock agriculture, where output is generated by a series of inputs data are dealt with, if ever, within the context of major agri- directly controlled by the producer, which are often difficult cultural initiatives. Second, the Sourcebook is a joint product to measure, but also influenced by a series of variables of users and suppliers of livestock data, with its overarching beyond his control, such as temperature and rainfall. Within objective being to respond to the information needs of data agriculture, livestock is a key sector which poses considerable users, and primarily the Ministries responsible for livestock challenges for collecting data, and hence designing effective in African countries and the National Statistical Authorities. policies and investments. As far back as 1957, the Chief of Finally, the Sourcebook represents a unique experiment of in- the Agriculture Division of the US Bureau of the Census, Dr. ter-institutional collaboration, which jointly places the World Ray Hurley, observed: “in analysing the [US] census experience Bank, the FAO Animal Production and Health Division, the covering 16 nationwide censuses and almost 120 years, one ILRI and the Africa Union — Interafrican Bureau for Animal concludes that the nationwide collection of satisfactory livestock Resources as well as national governments in Niger, Tanzania data ... is a difficult task and involves a number of problems. Even and Uganda at the forefront of data and statistical innovation the job of obtaining a count of livestock is fraught with difficulties. for evidence-based livestock sector policies and investments. Livestock numbers change every day of the year. Marketing is a continuous process. Livestock inventories are affected by births, This Sourcebook represents a first step towards a deaths, farm slaughter, and by growth and change in age of ani- demand-driven and sustainable approach to enhance the live- mals” (Hurley, 1957, pp. 1420–1). stock information available to decision makers. It is hoped it will provide a useable framework for significantly improving Recognizing that stakeholders contend that data availability the quantity and quality of livestock data and statistics avail- which feed into evidence based livestock policies and invest- able to the public and private sector, and also increase the ments is inadequate and fragmented, the World Bank, the efficacy of investments that country governments and the FAO, the International Livestock Research Institute (ILRI) international community allocate to generate information for and the African Union — Interafrican Bureau for Animal livestock sector policies and investments. Resources (AU-IBAR), with financial support from the Bill & Melinda Gates Foundation (BMGF), implemented the Livestock in Africa: Improving Data for Better Policies Project. The Project, implemented between 2010–2013 in collabora- tion with the pilot countries of Uganda, Tanzania and Niger, World Bank  |  Juergen Voegele, Director, Agriculture and targeted an improvement of the quantity and quality of the Environmental Services Department livestock information available to decision makers through enhanced methods for data collection and analysis within the context of the overall agricultural statistical system. FAO  |  Berhe G. Tekola, Director, Animal Production and This Sourcebook summarizes the outputs and lessons of the Health Division Livestock in Africa: Improving Data for Better Policies Project. It aims to present the challenges facing professionals collecting and analysing livestock data and statistics and possible solu- tions. While the Sourcebook does not address all conceivable ILRI  |  Jimmy Smith, Director General issues related to enhancing livestock data and underlining QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations iv  |  Investing in the Livestock Sector: Why Good Numbers Matter ACKNOWLEDGEMENTS T his Sourcebook was prepared by a core team composed of Nicolas Kauta (MAAIF), Mimako Kobayashi (World Bank), Ugo Pica-Ciamarra (FAO, lead author), Derek Baker (ILRI, Seth Mayinza (UBOS), John McIntire (ILRI), Nadhem Mtimet now at the University of New England), Nancy Morgan (ILRI), Titus Mwisomba (NBS), Vincent Ngendakumana (FAO) and with contributions from Carlo Azzarri (IFPRI), (African Development Bank), Gabriel Simbila (NBS), Morrice Cheikh Ly (FAO RAF), Longin Nsiima (MLFD), Simplice Oyuke (NBS), Steve Staal (ILRI), Diane Steele (World Bank), Nouala (AU-IBAR), Patrick Okello (UBOS), Joseph Sserugga Luca Tasciotti (ISS), Emerson R. Tuttle (Tufts University), (MAAIF) and Alberto Zezza (World Bank). Windy Wilkins (BMGF) and Stanley Wood (BMGF). We are deeply grateful to Bea Spadacini, Anne C. Kerns and Cristiana Special thanks go to the following people, who provide Giovannini for formatting the document and to Clifton constructive and useful comments and suggestions on Wiens for patiently editing it. earlier drafts of the Sourcebook: Gashash Ibrahim Ahmed (AU-IBAR), Gero Carletto (World Bank), Atte Issa (MEL), The authors would like to express their appreciation to the Elisabeth Cross (Washington University), Thomas Emwanu Bill & Melinda Gates Foundation for its financial support (UBOS), Giovanni Federighi (University of Roma II), and for their flexibility in managing the underlying grant, an Kristin Girvetz (BMGF), Massimo Greco (ISTAT), John uncommon feature in development assistance. Jagwe (FarmGain Africa), Catherine Joseph (MLFD), QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations TABLES, FIGURES and BOXES  |  v TABLES, FIGURES AND BOXES TABLES Table 1. Core livestock indicators for sub-Saharan Africa . . . . . . . . . . . . . . . . . . . . . . . . 15 Table 2. Data sources for livestock indicators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41 Table 3. Content of the livestock module for agricultural and multi-topic household surveys . . . . . . . 46 Table 4. Tanzania: summary statistics using different household definitions . . . . . . . . . . . . . . . 57 Table 5. Uganda Livestock Census 2008: questions on milk production . . . . . . . . . . . . . . . . . 61 Table 6. Ethiopia Livestock Sample Survey 2010/11: questions on egg production . . . . . . . . . . . . . 61 Table 7. Niger National Survey of Household Living Conditions 2011: questions on meat production . . . . . 62 Table 8. Tanzania administrative records: data entries on livestock slaughtered and meat production . . . . 63 Table 9. Uganda: Proposed pilots to improve the routine system of livestock data collection . . . . . . . . 76 Table 10. Agricultural/livestock censuses in West Africa: 2000–2012 . . . . . . . . . . . . . . . . . . 82 Table 11. Year to year cattle population growth rate in West African countries, 1990 to 2010 . . . . . . . . 87 Table 12. Year to year sheep/goat population growth rate in West African countries, 1990 to 2010 . . . . . 87 Table 13. Tanzania: Example of a consumer product matrix (beef) . . . . . . . . . . . . . . . . . . . 108 Table 14. Uganda: Example of a production quality scoring table (milk) . . . . . . . . . . . . . . . . . 109 Table 15. Selected example of retail products . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114 Table 16. Uganda: Description of retail outlets . . . . . . . . . . . . . . . . . . . . . . . . . . . 115 Table 17. Example list of nominated constraints (milk, Wakiso District, Uganda). . . . . . . . . . . . . 124 FIGURES Figure 1. The integrated survey framework: a focus on livestock . . . . . . . . . . . . . . . . . . . . . 7 Figure 2. Quality of livestock data as perceived by stakeholders . . . . . . . . . . . . . . . . . . . . 42 Figure 3. Measuring milk production in Niger: Box plots comparing randomized recall methods against physical monitoring . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54 Figure 4. Milk production data experiment: Comparing 6-month recall distribution to lactation curve method. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55 Figure 5. Tanzania: Percentage of households practicing transhumance over the past 15 months by district . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58 Figure 6. Cattle beef slaughtered and beef production in Tanzania, 2001–2011 . . . . . . . . . . . . . . 64 Figure 7. Uganda: Livestock data reports submitted by Districts by month, January–December 2012 . . . . . 72 Figure 9. Uganda: District overall reporting rate . . . . . . . . . . . . . . . . . . . . . . . . . . . 73 Figure 8. Uganda: Frequency of District reporting . . . . . . . . . . . . . . . . . . . . . . . . . . . 73 Figure 9. Uganda: District overall reporting rate . . . . . . . . . . . . . . . . . . . . . . . . . . . 73 Figure 10. Uganda: District conditional reporting rate . . . . . . . . . . . . . . . . . . . . . . . . . 73 Figure 11. Animal life cycle and basic demographic parameters . . . . . . . . . . . . . . . . . . . . 84 Figure 12. Stages for integrating census and survey data using SAE . . . . . . . . . . . . . . . . . . . 99 Figure 13. Uganda: Percentage of households owning livestock by region: 2009/10 NPS and 2008 UNLC (with 95% confidence interval) . . . . . . . . . . . . . . . . . . . 100 QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations vi  |  Investing in the Livestock Sector: Why Good Numbers Matter Figure 14. Uganda: Density of large ruminants actual from survey (left), actual from census (right), and predicted from census (below) at regional and district level . . . . . . . . . . . 101 Figure 15. Uganda: Per Capita Livestock Income Actual from survey and predicted to Census . . . . . . . 102 Figure 16. Uganda: Share of income from livestock Actual from survey and predicted to Census . . . . . 103 Figure 17. Demand analysis: Questions to consumers regarding purchasing behavior . . . . . . . . . . . 110 Figure 18. Demand analysis: Enumerator observations on retail production (beef) . . . . . . . . . . . . 110 Figure 19. Demand analysis: Questions posed to retailers . . . . . . . . . . . . . . . . . . . . . . . 111 Figure 20. Consumers’ retail outlet preferences . . . . . . . . . . . . . . . . . . . . . . . . . . . 112 Figure 21. Quality scored, by retail outlet type . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112 Figure 22. Consumers’ preferences for product type . . . . . . . . . . . . . . . . . . . . . . . . . 112 Figure 23. Flow chart representation of constraint analysis methodology . . . . . . . . . . . . . . . 120 Figure 24. Constraint analysis: Elicitation of local knowledge . . . . . . . . . . . . . . . . . . . . . 121 Figure 25. Constraint analysis: Identification of underlying constraints . . . . . . . . . . . . . . . . 122 Figure 26. Constraint analysis: Excerpts from domain session checklists . . . . . . . . . . . . . . . . 123 Figure 27. Basic constraints identified in Tanzania and Uganda . . . . . . . . . . . . . . . . . . . . 124 Figure 28. Tanzania: Constraints nominated by producers . . . . . . . . . . . . . . . . . . . . . . 125 BOXES Box 1. Livestock’s contribution to gross domestic product . . . . . . . . . . . . . . . . . . . . . . . 16 Box 2. Uganda: the demand for information of a milk processor . . . . . . . . . . . . . . . . . . . . 20 Box 3. A Tool for the Inclusion of Livestock in the CAADP Compacts and Investment Plans . . . . . . . . 29 Box 4. Livestock questions in the Population and Housing Census . . . . . . . . . . . . . . . . . . . 38 Box 5. Issues in measuring pastoral economies . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56 Box 6. Routine livestock data collection in Zanzibar . . . . . . . . . . . . . . . . . . . . . . . . . . 70 Box 7. Livestock population: a critical statistics . . . . . . . . . . . . . . . . . . . . . . . . . . . 80 Box 8. Livestock and livelihoods in Tanzania . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96 Box 9. CAADP Pillar 2: Market access . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106 Box 10. CAADP Pillar 3: Food Supply and Hunge r . . . . . . . . . . . . . . . . . . . . . . . . . . . 117 QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations ABBREVIATIONS and ACRONYMS  |  vii ABBREVIATIONS AND ACRONYMS AI Artificial Insemination LSMS-ISA Living Standards Measurement Study — Integrated Surveys on Agriculture AMD Average Milk per Day LU Livestock Unit AU-IBAR African Union — Interafrican Bureau for Animal Resources MAAIF Ministry of Agriculture, Animal Industry and Fisheries, Uganda BMGF Bill & Melinda Gates Foundation MEL Ministère de Élevage, Niger CAADP Comprehensive Africa Agriculture Development Programme MLF Ministry of Livestock and Fisheries, Zanzibar CBPP Contagious Bovine Pleuropneumonia MLFD Ministry of Livestock and Fisheries Development, Tanzania CCPP Contagious Caprine Pleuropneumonia NDVI Normalized Difference Vegetation Index CCT CAADP Country Team NBS National Bureau of Statistics, Tanzania CIRAD Agricultural Research for Development NCD Newcastle Disease CPI Consumer Price Index NDVI Normalized Difference Vegetation Index EA Enumeration Area NGO Non-governmental Organization EPA Enquête Permanente Agricole, Burkina Faso NLC National Livestock Census FAO Food and Agriculture Organization of the United Nations NPS National Panel Survey FMD Food and Mouth Disease OECD Organization for Economic Co-operation and Development GDP Gross Domestic Product OiE World Organization for Animal Health ILRI International Livestock Research Institute SAE Small Area Estimation ISN Institut National de la Statistique, Niger TCF Technical Conversion Factor JICA Japan International Cooperation Agency TLU Tropical Livestock Unit LC Lactation Curve UBOS Uganda Bureau of Statistics LDIP Livestock Data Innovation in Africa Project UNFPA United Nations Population Fund LID Livestock in Development UNLC Uganda National Livestock Census LSD Lumpy Skin Disease LSMS Living Standards Measurement Study QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations viii  |  Investing in the Livestock Sector: Why Good Numbers Matter ©FAO/Giulio Napolitano QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations INTRODUCTION | 1 INTRODUCTION T he growing demand for food of animal origin in devel- numbers and meat and dairy production, consumption, oping countries, stimulated by population growth, gains and trade flows of a number of livestock products, both in real per capita income, and urbanization, represents a raw and processed (e.g. FAOSTAT, 2013; WAHIS, 2013). major opportunity for poverty reduction, economic growth, The quality of available data, however, is often questioned and overall contribution to the post-2015 Development by livestock stakeholders, even for the most basic indica- Agenda (Delgado et al., 1999). tors such as livestock numbers (see chapter 1.4). This is particularly the case for Africa where aggregate ●● Nationally representative household, agricultural and/or economic growth of over 5 percent per year over the period farm surveys — which are more or less regularly under- 2000–2013 has exceeded growth rates in many other world taken by the National Statistical Authorities — tend to regions due to consolidated macroeconomic and political sta- marginally appreciate livestock. The survey questionnaires bility throughout the continent. Robust economic growth in contain only a few, if any, livestock-related questions, Africa has been and is anticipated to translate into a growing mainly focusing on the number of animals owned and val- demand for animal-source foods. Meat and dairy products ue of production. These surveys, therefore, don’t currently are high-value food products for which consumption is well lend themselves to generating comprehensive information correlated with income level. In 2005/07, the average African on farm, non-farm and off-farm livestock-related activ- citizen consumed about 11 kilos of meat per year and 35 ities (e.g. on livestock trade), which is much needed by liters of milk. This is projected to progressively increase in the policy makers (see chapter 1.3). coming decades, up to 26 kilos and 64 liters in 2050 respec- tively (Pica-Ciamarra et al., 2013). ●● Specialized livestock surveys are rarely undertaken by national governments. These surveys typically target These projections are notable, but definitely more striking technical issues — such as animal breeds, feed, animal if one considers that by 2050 the African population will be diseases, meat production, etc. — with an ultimate objec- 2.2 billion, more than doubling its 2005/07 level (0.9 billion). tive of better understanding the determinants of livestock Overall, between 2005/07 and 2050 total milk consumption production and productivity. They represent a critical will increase from 32 to 83 million tons (+159%), and total input for the design of effective policies and investments meat consumption from 11 to 35 million tons (+218%). at farm level (see chapter 1.4). At constant farm-gate prices, the total market value of meat products will increase from US$ 33 to US$ 108 billion ●● National governments collect on a regular basis data on (+227%), and that of milk from US$ 17 to US$ 44 (+158%) animal diseases which, if uncontrolled, may cause major (Nouala et al., 2011; Pica-Ciamarra et al., 2013). economic and social losses. However, the quality of the collected data, including their timing and accuracy, is Available data on livestock, stakeholders contend, are insuf- uncertain. This limits the capacity of the government to ficient to formulate and implement the necessary public and effectively control and manage the spread of diseases, private sector investments for livestock sector development, including zoonoses (Okello et al., 2013). whose potential contributions to economic growth, poverty reduction and food security risk thus remain untapped. Most ●● Finally, all sources of livestock data and statistics — such countries “lack the capacity to produce and report even the mini- as agricultural censuses, livestock censuses, periodical and mum set of agricultural data necessary to monitor national trends ad hoc agricultural sample surveys, household income or or inform the international development debate” (World Bank, expenditure surveys — rarely if ever generate comprehen- 2011, p. 11). In particular, a review of existing livestock- sive information on pastoral production systems, which related data/datasets for African countries suggests that: is of considerable relevance to many African countries, particularly those in the Sahel and the Horn of Africa (see ●● There exists a variety of livestock-related indicators chapter 1.4). within Africa at country level, including figures on animal QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations 2  |  Investing in the Livestock Sector: Why Good Numbers Matter To sum up, livestock data are not widely collected by national Ministry responsible for livestock development, the National governments and rarely on a regular basis; and the quality Statistical Authority, and other national and pan-African of available data is mixed in its timeliness, completeness, public and private sector data stakeholders. As such, they comparability and accuracy. This makes it difficult the design address data issues which are of broad interest to livestock and implementation of effective investments and policies in stakeholders: the 23rd session of the African Commission for the sector. Agricultural Statistics (AFCAS, December 2013) recommend- ed country governments in the continent adopt some of the Over the past decades a number of initiatives have been tools and methods presented in the following chapters to launched to support the collection and analysis of agri- improve the quantity and quality of the livestock information cultural data and statistics, including the Partnership in available to decision makers. Statistics for Development in the 21st Century (PARIS21), the Wye Group on Statistics on Rural Development and PART I of the Sourcebook reviews the demand and supply of Agriculture Household Income, the UN Global Strategy to livestock data. It first presents the principles underpinning Improve Agricultural and Rural Statistics (World Bank, 2011), an effective agricultural and livestock statistical system, such and the 2010–2013 Livestock in Africa: Improving Data for as presented in the Global Strategy to Improve Agricultural Better Policies Project. The latter, jointly implemented by the and Rural Statistics (chapter 1.1). It then identifies the core African Union — Interafrican Bureau for Animal Resources livestock indicators needed by decision makers, not only for (AU-IBAR), the Food and Agriculture Organization (FAO), the regular monitoring and planning (chapter 1.2) but also for International Livestock Research Institute (ILRI), the World policy and investment purposes (chapter 1.3). It finally in- Bank, and the national governments of Niger, Tanzania and vestigates whether the prevailing agricultural data collection Uganda, is possibly one of the first attempts to specifically systems suffice to generate these indicators (chapter 1.4). address livestock data and statistical issues in Africa. In most cases the answer to this question is no, or only to a limited extent. This Sourcebook on livestock data summarizes the activities and outputs of the Livestock in Africa: Improving Data for PART II presents tools and methods on how to improve live- Better Policies Project. It provides guidance to decision makers stock statistical systems, including the quantity and quality responsible to collect and analyze livestock data from differ- of livestock data. It proposes a livestock module for integrat- ent perspectives on how to systematically address livestock ed household or agricultural surveys, which consists of a set data-related issues within the context of the national agri- of questions aimed at revealing the full role of livestock in the cultural statistical system. In particular, it first develops the household and the farm economy (chapter 2.1); it reviews ex- skeleton of a sound livestock statistical system — consistent periments in survey design, including one on milk production with the demand of livestock information by stakehold- and one on pastoralist livelihoods, which provide guidance ers and the principles of the Global Strategy to Improve on how to develop or improve the content of household or Agricultural and Rural Statistics (World Bank, 2011) — which farm level survey questionnaires (chapter 2.2); it addresses represents the foundation for producing good livestock approaches to better estimate livestock technical conversion data. It then presents a sample of methods and tools – and factors, and hence livestock production (chapter 2.3), and associated examples — designed to improve the quantity and presents an institutional approach to improve the quality of quality of livestock data available to decision makers. These routine livestock data or administrative records, which are a tools and methods target household and farm level data — major source of information on animal diseases in the coun- for example, trade data and the role of expert informants to try (chapter 2.4). generate statistics are not dealt with in the Sourcebook — and to a large extent have been tested in the context of the PART III provides some practical evidence on how country implementation of Living Standards Measurement Studies governments produce or could produce some selected live- and small-scale data collection exercises in Niger, Tanzania stock indicators for the proper formulation of policies and and Uganda. They were jointly identified and developed investments. Chapter 3.1 highlights options for estimating based on dialogue between the Livestock in Africa: Improving livestock population in and in-between surveys, with ex- Data for Better Policies Project and users and suppliers of amples from West Africa. Chapter 3.2 discusses how, using livestock data and statistics at country level, including the data from the implementation of the livestock module for QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations INTRODUCTION | 3 multi-topic household surveys, the contribution of livestock of a methodology to collect data on the quality dimensions to household livelihoods can be properly assessed and feed of the market for animal-sourced foods. This information into the design of policies and investments that maximize is not captured by quantitative data, but it is essential to the impact of sector growth to the broader goal of poverty assess the opportunities for a demand-driven growth of the reduction. Chapter 3.3, 3.4 and 3.5 bring to light that livestock sector which is inclusive of smallholder producers’ livestock data from most surveys — even when an effective participation. Finally, Chapter 3.5 reveals that available data agricultural statistical system is in place — are insufficient are usually sufficient to identify broad categories of symp- on their own to provide detailed guidance to investors and toms of constraints to livestock production and productivity, policy makers and present methods to fill this information but do not suffice to provide clear guidance for policies and gap. Chapter 3.3 gives an example of data integration to investments. It then presents a methodology, implemented obtain statistically robust measures of the contribution of and tested in Uganda and Tanzania, which helps mapping livestock to household income at district level in Uganda, by symptoms with a structured list of core constraints at farm jointly using data from the 2008 Uganda Livestock Census level, thereby assisting decision makers in identifying priority and the 2009/10 Uganda Panel Survey. Chapter 3.4 presents areas for investments to increase livestock production and and discusses the implementation in Tanzania and Uganda productivity. ©FAO/Giulio Napolitano QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations 4  |  Investing in the Livestock Sector: Why Good Numbers Matter PART I. DEMAND AND SUPPLY OF LIVESTOCK DATA: GAPS AND ISSUES 1.1 THE BASICS OF A PROPER LIVESTOCK STATISTICAL SYSTEM KEY MESSAGES THE ISSUE Good livestock data originate from a functional About 60 percent of rural households in developing countries agricultural statistical system. are partially or fully dependent on livestock for their liveli- hoods. Livestock rearing provides them with a wide spectrum of benefits, such as cash income, food, manure, draft power A wide number of livestock data users require a and hauling services, savings and insurance, and social status. multitude of data, but the agricultural statistical The livestock sector currently accounts for about one-third system should prioritize a minimum set of core of agricultural value added in developing countries, and for data as the building block of good livestock over half of the value added in industrialized economies statistics. (FAOSTAT, 2013). While livestock farming might also have some negative effects on society, through animal-human disease transmission and environmental impacts, the sector Data integration, i.e. the use of data originating remains critically important for millions of people in develop- from different livestock, agricultural and non- ing countries (Otte et al., 2012). agricultural surveys, is essential for the design of effective sector policies and investments. The livestock sector, and the role that animals play in the household economy in developing countries, are anticipated to change rapidly in the coming decades. Consumers, includ- Good governance, institutional collaboration ing those in sub-Saharan Africa, are increasingly demanding and capacity building are critical ingredients of a high-value agricultural products such as fruit, vegetables, functional agricultural statistical system, which meat, and dairy products (Delgado et al., 1999; Pica-Ciamarra also includes livestock. et al., 2013; Jabbar et al., 2010). Producers will respond to this growing demand and, as a consequence, livestock will become an increasingly important sector of agriculture. QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations PART I. DEMAND AND SUPPLY OF LIVESTOCK DATA: GAPS AND ISSUES   |  5 “Data not only measures progress, 2009). The budgetary implications for the Uganda Ministry responsible for animal resources cannot be it inspires it.” overstated. Hillary R. Clinton The estimation of livestock value added in the national accounts makes use of so-called technical conversion In this fast-changing context, good quality livestock data are factors. These are coefficients that convert a measured needed for designing and implementing policies and invest- livestock variable into a different unit of measure: for ments that sustain and promote the sector’s socially desirable example, ‘milk yield per cow per day’ allows estimating development. Available livestock data, and the derived statis- milk production by only counting the number of tics or indicators, however, are largely considered inadequate milking cows in the country. In Tanzania, the livestock for effective decision making. technical conversion factors used to estimate the livestock value added in the national accounts have Perry and Sones (2009) present a review of major been kept constant for over ten years, i.e. all possible databases targeting livestock and conclude that “often increases in livestock productivity achieved in recent available data is not adequate to answer the questions years are not captured in the official country statistics being raised or to allow optimal targeting or design of (MLFD, 2012). interventions. Available data is patchy, often old, dispa- rate, scattered and hard to combine and pull together. The above examples, and others available from developing Even seemingly mundane and basic data, such as accurate countries, highlight that livestock sector investments and estimates of the number of poultry in a country, are often policy decisions are often based on inadequate information, unobtainable, let alone more complex questions such as which results in a less than optimal allocation of scarce public what is the impact of a given disease”. resources. Investments that improve the quantity and quality of livestock data can thus generate handsome returns in the A Report on Livestock Data and Information in medium to long-term, provided they produce the information Tanzania released in 2010 by the Ministry of Livestock needed by decision makers to make evidence-based decisions and Fisheries Development reads: “Livestock data are for sector development. currently inadequate in Tanzania … as they lack consisten- cy through time and between sources; and are not complete as they possess a lot of gaps” (MLFD, 2010b). In 1999, LID produced a report on ‘Livestock in Poverty-Focused Development’: it estimated that about 70 percent of the rural poor, about 970 million people, were dependent on livestock for part of their livelihoods (LID, 1999). Ten years later, in 2009, the FAO State of Food and Agriculture ‘Livestock in the Balance’ (FAO, 2009), touching on the livestock and poverty equation, duplicated the table produced by LID, clearly illustrating that livestock poverty data are not updated regularly. A National Livestock Census undertaken in Uganda in 2008 estimated the cattle population at 11.4 million. The day before the Census release, the national herd ©FAO/Giulio Napolitano stood at 7.5 million cattle. In other words, overnight the Census increased the cattle population in the country by 3.9 million heads, with pre-census data underestimating it by 52 percent (MAAIF and UBOS, QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations 6  |  Investing in the Livestock Sector: Why Good Numbers Matter LIVESTOCK IN THE GLOBAL STRATEGY the collection of the following core data as a minimum: inventory and annual births; level of production; imports TO IMPROVE AGRICULTURAL AND and exports; and producer and consumer prices. The Global RURAL STATISTICS Strategy also recommends that country governments should check the consistency of the suggested core items and data Livestock is part of agriculture; livestock data are part of with their own information needs and, in some cases, add agricultural data. Indeed, livestock is usually a component additional items and data. of agricultural surveys, with countries seldom undertaking standalone livestock surveys. Improving the quantity and quality of livestock data available to decision makers requires, therefore, improving the functioning of the agricultural PILLAR 2 Integrating livestock into the national statistical system statistical system which, in turn, is part of the national statis- Several governmental organizations/agencies collect and use tical system. agricultural data. These include, for example, the National Statistical Office, the Ministry responsible for animal re- The Global Strategy to Improve Agricultural and Rural Statistics sources; the Dairy/Meat Board; the Ministry of Trade, and (Global Strategy), endorsed by the UN Statistical Commission others. These actors often collect the same data, but because in 2010, provides broad guidance on how to improve the agri- of little coordination, end up producing indicators that are cultural statistical system, and livestock data therein (World incomparable, or even conflicting in some circumstances. Bank, 2011). The Global Strategy recommends targeting in- There are several reasons for this, such as the use of different vestments to improve agricultural and rural statistics around sampling units and/or different samples; different concepts, three pillars: definitions and classifications; different methods of data 1. The establishment of a minimum set of core data that collection; different questionnaires; and other. country governments should collect on a regular basis; The Global Strategy recommends that country governments 2. The integration of agriculture into the national statistical develop a unique master sample frame for agriculture. system; The frame is the means by which the statistical units to be enumerated in the collection are identified, such as a list 3. Governance and statistical capacity building. of all rural households or agricultural holdings, identifying each unit without omissions or duplication. A unique master PILLAR 1 Establishing a minimum set of core sample will provide the basis for the selection of samples livestock data of farms or households for all surveys, which allows linking farm and household characteristics and connecting both to Different stakeholders demand a variety of data and indica- the land cover and use dimensions. The “area sample” frame tors for a multitude of purposes, which all too often exceed — which is essentially the country land mass divided into the production capabilities of the national statistical system. sampling units — is deemed appropriate to this purpose. The The Global Strategy recommends that the starting point for the improvement of agricultural and rural statistics be the identification of a core set of data to be regularly collected. “The Global Strategy recommends These core data, selected for their importance to agriculture, that country governments should target the social, the productive and the environ- mental dimensions of the sector. They will provide inputs to develop a unique master sample develop several indicators/statistics, including the national frame for agriculture. accounts and the balances of supply and demand for food and The frame is the means by which the other agricultural products. statistical units to be enumerated in The Global Strategy identifies five core livestock items from the collection are identified.” which data should be collected, namely cattle; sheep; pigs; goats; and poultry. For these items, the Global Strategy urges QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations PART I. DEMAND AND SUPPLY OF LIVESTOCK DATA: GAPS AND ISSUES   |  7 “Using common classifications, the integrated survey framework could include, for instance, a light annual agricultural survey with basic questions on concepts and definitions is critical livestock; a specialized survey administered every other year to facilitate the use of data from collecting detailed data on the livestock sector; administra- different surveys.” tive records and community surveys used to collect data on animal diseases on a monthly basis; remote sensing surveys to count animals in pastoral areas at regular year interval; adoption of a unique master sample for agriculture ensures and expert judgments used to estimate and regularly update that data from different surveys, including standalone livestock technical conversion factors. livestock surveys, can be combined and jointly analyzed, thereby facilitating the appreciation of livestock’s role in the Using common classifications, concepts and definitions is crit- micro and macro economy. A unique master sample frame ical to facilitating the use of data from the different surveys demonstrates its value when an integrated survey framework included in the integrated survey framework. For example, (Figure 1) is developed and when data collectors use common milking animals could be defined variously as all females classifications, concepts and definitions. An integrated in reproductive ages, or as females bred especially for milk survey framework ensures that, with no duplication and at production and actually milked during the reference period. minimum cost, all core data, and additional needed data, can Furthermore, milk production could be gross, which includes be collected as demanded by stakeholders. As to livestock, the milk sold and that suckled by young animals, or net, which FIGURE 1. THE INTEGRATED SURVEY FRAMEWORK: A FOCUS ON LIVESTOCK Master Sample Frame (geo-referenced to land cover/use) Basic questions on livestock Specialized livestock survey Annual Periodic surveys surveys Within year surveys Int. data AGRICULTURAL (optional) management DATA AND INDICATORS system Administrative Remote Agri-Business Expert Community Data Sensing Judgment Surveys Animal Pastoralism, Animal disease Technical Disease control/ health/disease wildlife spread conver. marketing information data factors QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations 8  |  Investing in the Livestock Sector: Why Good Numbers Matter excludes milk suckled by young animals. Alternatively, meat and establishing the governance structure underpinning a production could be quantified as dressed carcass weight, functional agricultural statistical system. Capacity building gross carcass weight (including the hide or skin, head, feet and involves the improvement of statistical capacity at the coun- internal organs, but excluding the part of the blood which is try level to ensure that countries successfully implement the not collected in the course of slaughter), or live weight (FAO, Global Strategy. 2000). As far as possible, countries should make use of the FAOSTAT Commodity List, which provides an international classification for agriculture commodities, including live ani- THE SPECIFICITIES OF THE LIVESTOCK mals and livestock primary and processed products. SECTOR While improving the agricultural system is a pre-requisite to PILLAR 3 Governance and capacity building improve the quantity and quality of livestock data, the proper measuring of livestock requires addressing some unique Multiples organizations are involved in the collection and sector characteristics. analysis of agricultural data, including livestock data. A Back in 1957 Hurley observed: “in analysing the [US] functional statistical system requires that the roles and census experience covering 16 nationwide censuses and responsibilities of all actors be clear and agreed upon; that almost 120 years, one concludes that the nationwide common concepts, standards and classifications are used; collection of satisfactory livestock data … is a difficult that samples are drawn from the sample master frame; and task and involves a number of problems. Even the job of that there is no duplication of efforts, as all data collection obtaining a count of livestock is fraught with difficulties. systems will find their logical place in the integrated survey Livestock numbers change every day of the year. Marketing framework. is a continuous process. Livestock inventories are affected Data from livestock are collected not only by the National by births, deaths, farm slaughter, and by growth and Statistical Office but also by other institutions, such as the change in age of animals” (Hurley, 1957, pp. 1420–1). Ministry responsible for animal resources, the Meat and Dairy Board, the Ministry of Industry, and the Ministry of Trade. It follows that any improvement in the quantity and quality of livestock data should involve not only the National Statistical Authority but also other actors, which require targeted statistical capacity building. On the other hand, the Statistical Authority would need to appreciate the peculiar characteristics of livestock, a pre-condition for ensuring that livestock is adequately represented in statistical surveys. Implementing the Global Strategy The Global Strategy to Improve Agricultural and Rural Statistics is implemented through a Global Action Plan which, in turn, is articulated in regional plans, including one for Africa. The Global Action Plan includes three major components: research, technical assistance, and capacity building. The research component aims at developing technical guide- ©FAO/Giulio Napolitano lines and handbooks on methodologies, standards and tools related to the pillars of the Global Strategy. Technical assistance is country specific and aims at assisting country governments in designing agricultural sector statistics plans QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations PART I. DEMAND AND SUPPLY OF LIVESTOCK DATA: GAPS AND ISSUES   |  9 While there are infinite issues to address in successfully as- ■■ How to ask milk production questions, so as to also sessing livestock, from a data collection perspective there are measure the quantity of milk suckled by calves? ultimately three broad areas that should receive attention: sampling; animal biology (zoology) and production systems; ■■ How to quantify manure production in traditional and animal health/diseases. production systems and how to value it? ●● Sampling: The presence of animals across space depends ■■ Other, such as measuring poultry meat production at on a variety of factors, such as agro-ecological conditions farm level, or the value of the transport and draught and animal movements, which means the spatial dis- services provided by animals. tribution of livestock changes throughout the year and ●● Animal health/diseases: The Global Strategy notes that is somewhat uncorrelated to that of rural households “understanding the demand for statistical information at the and farm holdings, which are the typical sampling units. national level […] is a key element of the sustainability of an Selecting appropriate sampling points, appropriate sam- agricultural statistics system. Demand can be supported and ples and sample weights, and identifying the right time strengthened if the statistical system is responsive to users for any survey also targeting livestock can be therefore and provides statistics that are relevant, accessible, timely, challenging, but it is critical for producing reliable live- and with a level of accuracy that meets their needs” (World stock sector statistics. Bank, 2011, p. 27). Regarding livestock, stakeholders ●● Animal biology and production systems: Animals’ life demand a variety of indicators (see chapter 2 and 3 in cycles are affected by the way they are raised, i.e. by the World Bank 2011), among which animal health/disease production system. Measuring the latter is challenging data require special attention for three reasons. First, when rural households — rather than commercial enter- the Ministry responsible for animal resources typically prises — keep animals, as these do not regularly record allocates a large, if not the largest, part of its resources to inputs and outputs along the production process. In these the management and control of epidemic and zoonotic circumstances, a number of data-related issues need to be diseases. Second, the Ministry itself often collects animal addressed before any livestock data collection starts. For health/disease data, i.e. it is both a supplier and user of example: animal health data. Finally, country governments have international obligations to regularly report on their ■■ Which is the appropriate recall period for survey animal disease situation to the World Organisation for questions on the number of animals, given that species Animal Health (OiE) — including immediate notification have different life cycles? (within 48 hours) of an outbreak of an OiE-listed disease. In Africa, they must also send monthly reports on their ■■ How to assess the grade of the animals, considering, animal disease status to the African Union – Interafrican for instance, that the monetary value of a herd of thin Bureau for Animal Resources (AU-IBAR). A statistical cattle differ from that of one of well-fed animals? system that responds to users’ needs, therefore, must be ■■ How to formulate survey questions on animal diseas- able to ensure the collection of timely and reliable animal es? Should one follow an etiological or a symptomatic health/disease data. approach? Are household or community surveys the most appropriate survey tool? “What we measure affects what we ■■ How to quantify labor input, and hence labor produc- do; and if our measurements are tivity, when the herder manages a mixed herd, e.g. flawed, decisions may be distorted.” when s/he jointly takes different animals to water points? Stiglitz Commission on ■■ How to measure the quantity of forage available from the Measurement of roadside hedges, often a major source of animal feed? Economic Performance and Social Progress, 2010 QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations 10  |  Investing in the Livestock Sector: Why Good Numbers Matter CONCLUSIONS addressed for the agricultural data system to generate sufficient good quality livestock data, as livestock present peculiar characteristics that require ad hoc methods In the coming decades, the livestock sector is anticipated and approaches to data collection that need to be devel- to grow rapidly in developing countries. This provides oped and implemented. The next three chapters in the both opportunities and challenges, which are best dealt Sourcebook assess the demand for and availability of with through good quality livestock data and indicators. livestock data, with the objective of identifying the major However, there is evidence that current agricultural data information gaps facing livestock stakeholders. Chapter and indicators — including livestock data — are often 1.2 identifies the core livestock data and indicators that inadequate, which prevents the design of effective policies decision makers need on a regular basis to fulfil their and investment in the sector. mandate. Chapter 1.3 presents the information that de- As recommended by the Global Strategy to Improve cision makers need for policy and investment purposes, Agricultural and Rural Statistics, country governments linking it to the various phases of the policy process, should invest resources to improve the agricultural sta- from agenda setting to policy implementation. Finally, tistical system, starting with identifying a minimum set chapter 1.4 examines whether the prevailing agricul- of core data; developing an integrated survey framework; tural data collection systems suffice to satisfy the data and ensuring cross-institutional collaboration. At the demands of livestock stakeholders and identifies priority same time, some livestock-specific data issues need to be information gaps. ©FAO/Giulio Napolitano QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations PART I. DEMAND AND SUPPLY OF LIVESTOCK DATA: GAPS AND ISSUES   |  11 1.2 CORE LIVESTOCK DATA AND INDICATORS KEY MESSAGES AS MANY LIVESTOCK INDICATORS AS Core livestock data of critical importance LIVESTOCK STAKEHOLDERS identified by the Global Strategy to Improve A multitude of stakeholders make use of livestock data and Agricultural and Rural Statistics include: 1) animal indicators for a variety of purposes. Stakeholders include gov- numbers and births; 2) production of animal ernment ministries and other public or quasi-public agencies, products; 3) trade statistics; and 4) producer and such as dairy boards and statistical authorities; the private consumer prices. sector, encompassing small, medium and large scale livestock producers as well as input suppliers, traders, consumers and other actors along the value chain; livestock researchers and Livestock stakeholders recommend including scientists in national, regional and international institutions; animal disease-related data in the core data, such the civil society, such as NGOs, trade unions and indigenous as number of animals vaccinated and outbreaks peoples movements; international organizations and the of animal diseases. These data are essential for donor community. the Ministry responsible for livestock which, to fulfill its mandate, allocates a large share of its Livestock stakeholders have different objectives and look for budget to control and manage animal diseases. different statistics, in terms of data items, variables, level of representativeness and time dimension. For instance, while indicators on livestock population and its trend at national The needs of livestock data users require that the level are of primary importance for the Ministry responsible institutions involved in the collection of livestock for animal resources, these are of limited relevance for small data provide statistics at different levels of or medium scale producers; while traders look for daily aggregation and with different time frequency. information on market prices of live animals and livestock products in terminal markets, this information is of little use to epidemiologists; while national governments, interna- tional organizations and the donor community have interest in accessing indicators on the incidence and distribution of poverty, including on poor livestock keepers, these statistics “We, the Ministers responsible for are of marginal, if any, significance for consumers. Animal Resources in Africa… Stakeholders are mostly dissatisfied with the quantity and urge Member States to enhance quality of available livestock data and indicators (World Bank, capacity for timely collection, 2011). Public investments are thus called for to enhance their quantity and quality. However, any attempt to improve the analysis and sharing of quality agricultural statistical system so that good data and indi- data to guide policy, strategy and cators are provided to all livestock stakeholders as per each stakeholder’s specific needs is destined to fail. investment programmes.” First, there are many stakeholders with a numerous informa- African Union, 2010 tion needs, i.e. thousands of indicators should be produced to satisfy their demand for information. Second, while some data and indicators are public goods, many others are private goods: these should not be generated by the public sector but by private actors with their own resources. Third, some QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations 12  |  Investing in the Livestock Sector: Why Good Numbers Matter indicators are needed only in specific circumstances, and it These data would help in the estimation of the two major would be inefficient to generate them regularly within the livestock indicators identified in the Global Strategy (World context of the agricultural statistical system, i.e. ad hoc data Bank, 2011, p. 34), namely: collection exercises should be undertaken in these cases. Examples could be indicators on the nutritional value of raw ●● Livestock value added — a critical component of the milk, which are of use when a nutrition policy is formulated; Gross Domestic Product — for the calculation of which or on the breed traits of local animals, which are largely stat- data are needed on animal population, production level ic. Finally, the public sector acts on budget constraints, which and use of inputs; prevent the establishment of a comprehensive agricultural ●● Changes in components of livestock and poultry popula- statistical system capable of generating all conceivable live- tion by species, which encompasses data on trends in the stock-related indicators. livestock population and herd composition by gender, age and purpose (e.g. for breeding or fattening). CORE LIVESTOCK DATA AND Before embarking in any effort to improve agricultural data INDICATORS IN THE GLOBAL systems, country governments — recommends the Global STRATEGY TO IMPROVE Strategy — should check the consistency of the suggested AGRICULTURAL AND RURAL core items and data with their own information needs and, in STATSTICS case, add additional items and data. Camels and alpacas, for instance, could be a livestock item for Sahelian and Andean countries respectively. National governments are also recom- The Global Strategy recommends that a “minimum set of core mended to determine how frequently data for the core items data is to be used as a starting point” to improve the agricul- should be collected and associated indicators generated. tural statistical system. These core data should target three major dimensions of agriculture, namely the social, the production and the environmental dimensions. The livestock PRIORITY LIVESTOCK INFORMATION sector falls under the production dimension and the Global NEEDS IN SUB-SAHARAN AFRICA Strategy identifies five core livestock items for which indica- tors are to be generated (World Bank, 2011, p. 14): The FAO-World Bank-ILRI-AU-IBAR Livestock in Africa: ●● Cattle; Improving Data for Better Policies Project undertook four ●● Sheep; online surveys — two global and two targeting Ugandan and ●● Pigs; Tanzanian stakeholders respectively — and sponsored two ●● Goats; international workshop in East Africa to better appreciate ●● Poultry. the information needs of livestock stakeholders and, in par- ticular, of the National Statistical Authority and the Ministry These items were selected because of their importance to live- responsible for animal resources (LDIA, 2011a, 2011b, stock production globally: they contribute to over 99 percent 2011e; Pica-Ciamarra and Baker, 2011; Pica-Ciamarra et al., of meat, milk and eggs production, with the remaining com- 2012). The latter are the major actors in livestock data col- ing from animals such as camels, yaks, rabbits and equines lection and statistics dissemination in developing countries, (FAOSTAT, 2013). For the above items, the Global Strategy and any improvement in systems of livestock data collection (World Bank, 2011, p. 14) identifies the following core data: should first target their priority information needs (MLFD and LDIP, 2011). Only then will these institutions will be will- ●● Inventory and annual births; ing to invest resources to collect and produce other livestock data and indicators to meet their additional information ●● Production of products such as meat, milk, eggs, and wool, and net trade or imports and exports; needs and/or the demands of other stakeholders. Priority information needs are here defined as the set of data ●● Producer and consumer prices. and indicators that the National Statistical Authority and the QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations PART I. DEMAND AND SUPPLY OF LIVESTOCK DATA: GAPS AND ISSUES   |  13 ©FAO/Giulio Napolitano Ministry responsible for livestock development require on a variety of domains — e.g. social, economic and environment regular basis to properly fulfil their mandate, i.e. those data statistics — in order to meet the information needs of data and indicators that are essential to deliver their monthly, stakeholders, including the government. This involves the quarterly and annual outputs, and whose generation is administration of censuses and sample surveys; analysis of typically funded through the recurrent expenditure in their data and dissemination of statistics and statistical reports; annual budget. Information needed on a larger frequency or the promotion of a coordinated, harmonized and efficient na- irregularly is not considered a priority, even though it may tional statistical system; and training and guidance to other well be of critical importance for livestock stakeholders. providers and users of statistics. Priority livestock information needs for the While the National Statistical Authority has a broad mandate, National Statistical Authority its priority livestock information targets the production of two major indicators, which it generates and disseminates at The National Statistical Authority is mandated to ensure least once per quarter. These are: the production and dissemination of reliable statistics in a ●● The Consumer Price Index (CPI); ●● The Gross Domestic Product (GDP). “CPI is the most relevant measure CPI is estimated monthly and is one of the several price indi- of the cost of living in all countries ces calculated by the National Statistical Authority. It is the and its trend is used to calculate most relevant measure of the cost of living in all countries the inflation rate, a major target of and its trend is used to calculate the inflation rate, a major target of monetary policies. It is also used as a price deflator monetary policies.” in the compilation of real economic statistics, such as GDP at constant prices. QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations 14  |  Investing in the Livestock Sector: Why Good Numbers Matter CPI is a weighted average of prices of a representative basket Priority livestock information needs for the Ministry of consumer goods and services, such as food and non-al- responsible for animal resources coholic beverages; housing water, clothing and footwear; electricity, gas and other fuels; health; transport; etc. Weights The Ministry responsible for animal resources has the overall are (should be) updated every five years at least, based on mandate to promote, regulate and facilitate the sustainable budget/expenditure survey data. The food basket, which development of the livestock sector in the country. This includes animal-source foods, is a major component of CPI. involves the formulation, implementation and monitoring Prices are usually collected by data collectors in a sample of and evaluation of sector programs and policies, as well as the outlets in rural and urban areas (ILO, 2004). delivery of public services and goods, such as vaccinations against epidemic diseases. To fulfill its mandate, the Ministry GDP is the market value of all final goods and services requires a variety of information, but three set of indicators produced in a country and its trend is a major indicator of have been identified as the most needed, namely: growth in the economy. Most countries calculate GDP using the so-called production approach, which is basically the ●● Animal disease-related indicators, e.g. number and pro- difference between the value of outputs for all sectors less the portion of animals affected by a certain epidemic disease, value of goods and services used in producing those outputs number of animals at risk of infection, number of animals over the reference period. This is the so-called ‘value added’. vaccinated against selected diseases, etc.; In developing countries, livestock value added is a relevant ●● Indicators on animal population, e.g. number of animals component of the GDP. GDP estimates are released by the by species, breeds, sex and age over a reference period; National Statistical Authority quarterly and annually. ●● Production and productivity-related indicators, e.g. level of beef production per year and milk yield per cow. In most countries, as Chapter 1.1 noted, the Ministry mandated for livestock development allocates a large share of its resources to animal health-related activities. For in- stance, over 26 percent of the recurrent expenditure of the Tanzania Ministry of Livestock and Fisheries Development is used for this purpose, according to the Medium Term Expenditure Framework 2010/11 – 2012/13 (MLFD, 2010a). The fundamental reason is that the Ministry is responsible for managing and controlling epidemic and zoonotic diseases, and particularly to intervene as rapidly as possible when there are outbreaks, in order to avoid disease spread and the associated socio-economic losses. In addition, country gov- ernments have international obligations to regularly report on their animal disease situation to the World Organisation for Animal Health (OiE) — including immediate notification (within 48 hours) of outbreaks of an OiE listed disease. In Africa, country governments must also send monthly reports on their animal health status to the African Union – Interafrican Bureau for Animal Resources (AU-IBAR). Detection of animal disease outbreaks is of limited value on its own for the Ministry: updated information on the ©FAO/Ami Vitale livestock population in the affected area, and beyond, is essential for designing effective interventions and budgeting them properly. Preventive vaccination or stamping out, for QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations PART I. DEMAND AND SUPPLY OF LIVESTOCK DATA: GAPS AND ISSUES   |  15 example, are best implemented when the number of animals 1. Livestock value added at risk and those (potentially) infected by a certain disease are known with some statistical precision. Indicators on the Livestock value added is a critical component of GDP. Its livestock population, and its distribution across the country, calculation requires (i) data on total number of animals and are also essential for the Ministry to deliver public goods and changes in the number of animals — which can be treated services and formulate sector policies and programs. either as fixed capital (e.g. breeding animals) or as ‘work in progress’ animals (e.g. for slaughter) — over the reference Finally, the Ministry responsible for animal resources does period; (ii) on production of livestock products, such as need with some regularity, at a minimum once per year, indi- meat of various types, milk, eggs, hides & skins, manure, cators on livestock production and productivity, which are a etc; (iii) on the inputs used in the production process, such major piece of information for monitoring and evaluating the as animal feed/fodder and water; animal health services, effects of most interventions on the ground. vaccines, medicines and dips; fuel and electricity; repairs and maintenance; (iv) on imports and exports of live animals and livestock products; (v) on output and input prices. Outputs CORE LIVESTOCK INDICATORS IN are valued at farm-gate prices that reflect the value of goods SUB-SAHARAN AFRICA for the producers; inputs are valued at purchaser’s prices, i.e. the prices that are effectively paid by the producers (see Box The priority information needs by the National Statistical 1 and LDIP 2012a). This information is needed on a quarterly Authority and the Ministry responsible for livestock helps basis at a minimum. Data from nationally representative identify the core livestock indicators for sub-Saharan African sample surveys suffice for estimating livestock value added, countries and, more in general, for developing countries as a as GDP is presented for the country as a whole and, in some whole, including frequency and level of representativeness. circumstances, for its major regions. These are presented in Table 1 and discussed below. TABLE 1. CORE LIVESTOCK INDICATORS FOR SUB-SAHARAN AFRICA INDICATORS FREQUENCY LEVEL OF REPRESENTATIVENESS 1 Livestock value added Quarterly; Annually Country; Major-regions Average market prices for live animals and livestock 2 Quarterly; Annually Country; Major-regions products Outbreaks of animal diseases; Immediately after disease outbreaks; 3 Number of animals affected; District or lower administrative level Monthly Number of animals at risk. 4 Total number of live animals Quarterly; Annually District or lower administrative level 5 Total production quantity of major livestock products Annually Country; Major-regions QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations 16  |  Investing in the Livestock Sector: Why Good Numbers Matter BOX 1. LIVESTOCK’S CONTRIBUTION TO GROSS DOMESTIC PRODUCT T he size of livestock’s contribution to agricultural value added as well as to the gross domestic product (GDP), is a commonly quoted measure of livestock’s role in the are useful for analyses of structural changes in the economy and within sectors. national economy. In all countries, GDP is estimated at least Value added is defined as the value of the output of a sector quarterly and annually by national statistical authorities. minus the value of all intermediate inputs. It is calculated There are three ways of calculating GDP, which include the without making deductions for depreciation of fixed assets production approach, the expenditure approach and the and depletion/degradation of natural resources. Outputs income approach. All should lead to the same result. The from the livestock sector include the increase in the number production approach quantifies the difference between the of animals and the production of livestock products. The value of outputs for all sectors less the value of goods and increase in number of animals is represented by both fixed services used in producing those outputs during one year, capital formation — i.e. animals that are inputs into the pro- i.e. it quantifies the so-called ‘value added’ for all sectors in duction process, such as breeding animals and adult males the economy. The income approach measures the incomes of for breeding or animal traction —­and by so-called ‘work-in all individuals living in the economy over the reference year; progress’ animals, namely those reared for slaughter and the expenditure approach quantifies all expenditures by all young animals reared to become fixed assets. Livestock individuals living in the country in the accounting period. products include meat, milk, eggs, and other by-products, Most country governments estimate GDP using the produc- such as manure, hides and skins, fat, offals, honey, transport tion approach. This method allows for measuring the overall services, etc. Intermediate inputs comprise animal feed/fod- performance of the economy as well as that of each produc- der and water; animal health services, vaccines, medicines tive sector (e.g. livestock) and of specific enterprises within and dips; fuel and electricity; repairs and maintenance, such each sector (e.g. beef and poultry). It also allows for tracking as fences and equipment, etc. Outputs are valued at so- changes in the structure of the economy and within sectors. called basic prices, i.e. farm-gate prices that reflect the value Values added at constant prices are useful to estimate of goods for the producers. Intermediate inputs are valued at growth rates/performances of the economy as a whole or of the purchaser’s prices, i.e. the prices that are effectively paid sector/sub-sectors over time; values added at current prices by the producers. • 2. Average market prices for live animals and for OiE (OiE, 2011). These reports contain detailed information major livestock products on disease outbreaks, with information on latitude and lon- gitude and first administrative division, and actions taken to Average retail market prices, including for live animals, monitor and control the outbreak’s spread. animal-source foods and livestock by-products are needed for the National Statistical Authority to produce the CPI. 4. Total number of live animals by major species at Quarterly data, representative of the country and of its major district or lower administrative level. regions, suffice to produce CPI. These indicators are critical for the Ministry responsible for 3. Outbreaks of select animal diseases; number of livestock not only for efficient interventions when animal animals affected; number of animals at risk. disease outbreaks occur but also for the Ministry or Local Governments to supply other goods and services — such These indicators are essential for the Ministry to control and as the construction and maintenance of market facilities manage the spread of epidemic and/or zoonotic diseases, or the administration of vaccines against Foot and Mouth i.e. to identify outbreaks; treat and destroy animals; and to disease — and to design sector policies and programs, such vaccinate those at risk and/or control animal movement. In as on animal health or water for livestock. Quarterly data are addition, countries must report outbreaks of selected dis- preferred, as this allows monitoring changes in the livestock eases within 48 hours to OiE, send monthly animal-disease population, inclusive of large and small animals. reports to IBAR, and six-monthly and an annual report to QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations PART I. DEMAND AND SUPPLY OF LIVESTOCK DATA: GAPS AND ISSUES   |  17 5. Total quantity of production for major livestock hen. Production and productivity indicators, as said, are the products. basics to measure the performance of whatever intervention undertaken by the Ministry or other livestock stakeholders. Information on production levels is critical to monitor Annual data for the country as whole and its macro-regions trends in the sector and, combined with indicators on animal are typically sufficient. populations, it allows the generation of basic productivity indicators, such as milk yield per cow or eggs per laying CONCLUSIONS ●● While the core indicators for the Statistical Authority should be representative of the country as a whole and of major regions, the population and animal There are few core livestock indicators for sub-Saharan Af- disease-related core indicators for the Ministry respon- rican countries, defined as those needed monthly, quarterly sible for animal resources should be representative at and annually by either the National Statistical Authority district or lower administrative level. or the Ministry responsible for livestock, and which should be generated through the recurrent expenditure budget. ●● The National Statistical Authority demands data on a These are livestock value added, average market prices for quarterly and annual basis. The Ministry of Livestock live animals and livestock products; outbreaks of selected needs data more frequently, often on a monthly basis. animal diseases, number of animals affected, number of animal at risk; total number of live animals by main species ●● The identified core data and indicators correspond to at district or lower administrative level; total quantity of those in the Global Strategy, with the relevant excep- production for major livestock products. tion of animal disease-related indicators that are not mentioned therein. ●● Livestock value added contains, in principle, almost all information needed to monitor sector trends, particu- Investments aimed at improving livestock data systems in larly as it is released quarterly and annually. However, sub-Saharan African countries should first assess the pre- it does not include data on animal diseases, which are vailing agricultural (and livestock) data collection systems critical for the Ministry of Livestock. The details and to evaluate whether they generate enough data to produce precision with which countries estimate livestock value the identified core indicators. If this is not the case, then added vary, e.g. some may differentiate between local investments should be made to strengthen the production and exotic breeds of cattle and some not; some may in- of such indicators (Chapter 1.4 presents a critical review clude manure as one of the outputs of livestock, some of the prevailing agricultural and livestock data collection others may not. system in sub-Saharan Africa). It is also worth noting, however, that the availability of core livestock data and ●● Data needed to estimate the livestock value added, indicators is not sufficient for the statistical system to including on animal population, are of little use for the provide all the information needed by stakeholders to Ministry responsible for animal resources if collected, effectively design and implement livestock sector policies as in most of the cases, from sample surveys. Indeed, and investments. The latter should be based on a much to deliver its services the Ministry needs indicators on wider set of data and indicators, many of which are not to the distribution of the livestock population at district be generated on a regular basis. The next chapter explores or lower administrative level. the kind of information needed for making effective evi- dence-based livestock sector policies and investments. ●● Animal health indicators are of interest only to the Ministry of livestock and should be regularly collected at district or lower administrative level. QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations 18  |  Investing in the Livestock Sector: Why Good Numbers Matter 1.3 DATA AND INDICATORS FOR EVIDENCE-BASED LIVESTOCK POLICIES AND INVESTMENTS KEY MESSAGES INTRODUCTION Different data and indicators are needed The core livestock indicators identified in the previous throughout the various phases of the policy chapter are, on their own, insufficient to provide adequate process, from agenda setting through policy and information for the proper design of livestock sector policies investment design to implementation. and investments. Indeed, so-called evidence-based policies and investments require a wider spectrum of data and indi- cators – e.g. the number of cattle keepers and their average The statistical system provides enough herd; the seasonality of feed available and feed quality; information to broadly depict the livestock marketing facilities and animal health posts along marketing sector, including major trends, opportunities and routes; etc. They also need to be based on participatory and constraints of different segments of producers. inclusive policy processes and, in many circumstances, on some ex ante pilots, primarily to test on a relatively small scale the effects of prospective interventions by comparing The statistical system should provide all outcomes for those (households, communities, etc.) who par- information needed to design and implement ticipate in a given program against those who do not. livestock sector policies and investments. Country governments need to allocate resources A larger set of good-quality data and indicators, participatory for ad hoc data collection when the time comes decision processes and ex ante pilots are complementary to design and implement interventions in the ways to enhance the quality and quantity of information for evidence-based policies and investments. The entry point for livestock sector. their usefulness, however, changes throughout the decision making process. For example, good data are useful in identifying binding constraints to livestock productivity, and hence priority areas for investments; while ex ante pilots are more appropriate “What we measure for identifying effective interventions to remove those con- affects what we do; straints. This chapter systematizes the overall information and if our measurements needed by decision makers to effectively formulate and implement policies and investments in the livestock sector. It are flawed, decisions provides guidance on when and which data and indicators are may be distorted.” needed in the policy/investment dialogue; when participato- ry decision making processes are most valuable; and when ex Stiglitz Commission ante pilots are most appropriate. on the Measurement of It is recognized that the formulation and implementation Economic Performance and of policies and investments is a continuous process and that many development partners condition the final outcome. Social Progress, 2010 For clarity, however, it is assumed here that the decision maker is the Ministry responsible for animal resources, and that the Ministry’s overarching objective is the promotion QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations PART I. DEMAND AND SUPPLY OF LIVESTOCK DATA: GAPS AND ISSUES   |  19 ©FAO/A. Gandolfi of sustainable and inclusive growth in the livestock sector. 4. What to target? Therefore, the Ministry should consider the following questions: Understanding and interpreting the root causes of binding constraints is necessary for the formulation of 1. Why invest in livestock? policies and investments that ease or eliminate those constraints, thereby allowing livestock producers and Allocating resources to the livestock sector makes other stakeholders to capture all the potential benefits sense only if its development contributes to the broad- from livestock production and commerce. er socio-economic development goals of the country. It is therefore necessary to understand the extent and 5. How to design policies and investments? nature of livestock’s development contribution, both negative and positive. Decision makers need to be informed of the pros and cons of alternative ways and means of easing and/ 2. Whom to target? or removing one or more binding constraints. This requires assembly and analysis of information in ap- There is heterogeneity among livestock producers, and propriate forms and formats. variety in their responses to changes in the economic and institutional infrastructure as determined by poli- 6. How to ensure effective implementation? cy. Characterizing livestock producers is thus essential to formulate appropriate policies and investments. Monitoring and evaluation are necessary to ensure Identifying other benefactors from, and stakeholders that policies and investments be properly implemented in, livestock development is also valuable, particularly and that the necessary adjustments can be made. This as conduits to value chain-based change. requires an information and analytic base that is itera- tive with the answers to the questions posed above. 3. Which constraints? The following sections address the above questions. The Identifying the binding constraints that prevent differ- final section synthesizes the main points, focusing on the ent types of livestock producers and stakeholders from importance of accessing data and indicators, which provide making efficient use of their animals is indispensable a statistically precise picture of the country as a whole and in identifying priority areas for investment, and for of its major agro-ecological/administrative regions, a vital policy reform. Such constraints can impede develop- aspect for investment and policy design. This chapter does ment in various ways, at local, national, regional and not specifically deal with the demand for information by the continental levels. private sector, which is briefly discussed in the following box. QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations 20  |  Investing in the Livestock Sector: Why Good Numbers Matter BOX 2. UGANDA: THE DEMAND FOR INFORMATION OF A MILK PROCESSOR T he Sameer Agriculture & Livestock Ltd. (SALL) — a joint venture company established by the Sameer Group of Kenya in conjunction with RJ Corp. of India — took over of Milk production in Uganda is insufficient to satisfy existing demand (the country is a net importer of milk) and SALL finds difficulties in getting sufficient and timely supply of the former government parastatal Uganda Dairy Corporation milk (which leaves over 80 percent of its processing capacity in August 2006. Out of 39 large, medium and small dairy unused). SALL has its own sources of information and, like processing plants in Uganda, SALL is today the largest. SALL all active companies, gets direct and indirect information on is manufacturer of the ‘Fresh Dairy’ range of dairy products. market status and trends through its business partners and These include: fresh pasteurized milk; Ultra-Heat-Treatment through observing daily price trends. However, with the aim (UHT) milk; yogurt; butter; ghee, and powder milk. Fresh of expanding its operation and satisfying the unmet and pasteurized milk represents the major business for SALL, growing demand for milk in Uganda, SALL would appreci- with about 45 to 50 percent of the milk processed daily used ate updated information on districts with relevant surplus to produce pasteurized milk. About 30 to 40 percent of the production of milk as well as on potential trends of milk processed milk goes into UHT milk, and the rest into the production in the country. Some of this information is avail- other dairy products. able, but in most cases is either presented in formats which are of little use to SALL (e.g. only regional data are available SALL is a buyer of milk and a seller of dairy products. It or data are summarized in maps with no detail numbers at- largely buys from district cooperatives in Western and tached) and based on data which are more than a few years Central Uganda, which have established about 135 milk col- old. Delayed availability of data is problematic in a country lection centers equipped with coolers and generators as well where, according to the Uganda Bureau of Statistics, annual as testing kits provided by SALL. The milk is transported to GDP growth averaged over 7 percent over the past ten years, the so-called Bulking Centers, managed by the Cooperatives, a growth which translates into changing consumers’ food where it is chilled a second time. SALL insulated tankers preferences and demand for livestock products. • then take the milk to the processing plant in Kampala. WHY INVEST IN LIVESTOCK? the livestock sector contributes to economic growth, poverty reduction, food security, reduced vulnerability and other socio-economic goals. To this end, the Ministry should be A pre-condition for investment in improved livestock data able to access and package for advocacy purposes the live- systems by the Ministry responsible for animal resources is stock-related and socio-economic data and indicators which access to adequate resources, through the Ministry of Finance reveal sector trends, shares in various aggregates, and their or via other funding sources, such as the Regional Economic correlations with key socio-economic variables. Examples Communities, donors and financial partners, including the of such indicators are listed below; the figures are often private sector. Access to such funds requires demonstrating more illustrative and compelling when comparing between that investment in livestock contributes to the overarching countries. development goals of the country. Such contributions might relate to income generation and/or poverty reduction and ●● Trends and projections in total and per-capita food security, support enhanced resource use efficiency, and/ consumption of animal-source foods, at country and or generate economic gains through stimulating trade. These regional level, and in specific locations or zones. This contributions may also be regional in nature, such as the col- information could provide a rationale for supporting lective contribution to a goal like controlling animal disease. sustainable livestock sector growth in response to ob- Success in generating investment funds to support sector de- served growth in demand for high-value foods, including velopment requires that the following question be answered. animal-source foods. In much of the developing world, a convincing answer to this ●● Trends in livestock value added over the years, in question should provide evidence that the development of absolute terms and as proportion of agricultural value QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations PART I. DEMAND AND SUPPLY OF LIVESTOCK DATA: GAPS AND ISSUES   |  21 added and GDP. Given that the importance of livestock in ●● Bogale et al. (2005) look at the determinants of rural pov- agriculture tends to increase with economic development, erty in three Ethiopian districts, with poverty defined in this information could highlight that investments in the terms of both per capita household calorific consumption sector are needed to ensure its efficient and equitable and per capita household expenditure on basic needs. growth. They show that the probability of a household being poor declines as the number of oxen owned increases. ●● Number and proportion of rural households keeping selected livestock species, disaggregated by income, re- ●● Benin et al. (2008) use an economy-wide model to esti- gion, gender and other variables of development interest. mate the responsiveness of the poverty rate to per capita Available data from developing countries show that, in agricultural GDP growth in Malawi. A one percent in- most cases, the majority of rural dwellers keep livestock, crease in livestock GDP per capita is anticipated to reduce which suggests that broad-based increases in livestock national poverty by 0.34 percent. productivity could directly support their livelihoods, while also increasing the availability of animal protein to urban ●● Pica et al. (2008) show that increases in livestock pro- dwellers. ductivity — as measured by value added per Tropical Livestock Unit — appear to be/have been a cause of per ●● Rates of under-nutrition, daily per capita intake of capita GDP growth in 33 developing countries in Africa, meat and milk, and the proportion and section of Asia and Latin America. the population not consuming animal-source foods. These indicators could highlight the nutritional benefits ●● Bashir et al. (2012) estimate the contribution of livestock available from increasing the availability of affordable to food security in the State of Punjab, Pakistan, using livestock products. data from 12 out of its 36 districts. Food secure house- holds are defined as those with calorie intake at or above ●● Number and type of persons employed along select- 2,450 Kcal/per capita/day. Results show that ownership ed livestock value chains. This provides guidance on the of large and small ruminants has a positive impact on potential for investments in the livestock sector to gener- household food security. ate employment, which represents a major pathway out of poverty for the less well-off, amongst both urban and ●● Otte et al. (2012) estimate household livestock income rural populations, and amongst vulnerable stakeholders multipliers for major world regions, defined as the impact such as women. on total household income of a 1 US$ increase in either Simple data and indicators as the ones mentioned above can help make the case for investing in livestock. However, more powerful advocacy can be achieved by presenting rigorous statistical associations between livestock-based development and overall development. The following list of studies pro- vides examples of such work, which requires high quality data that is standardized within or across countries. This list also supports the development and use of more advanced sets of indicators more geared to advocacy. ●● In a seminal study on agricultural productivity differences across countries, Kawagoe et al. (1985) find that livestock — considered as an input representing long-run capital formation in the agricultural sector — is a significant determinant of agricultural production, as measured by ©FAO/Ami Vitale gross output net of agricultural intermediate products. QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations 22  |  Investing in the Livestock Sector: Why Good Numbers Matter livestock production or livestock processing. Calculated “At present there is a serious multipliers range from 2.0 to 6.8, and are found to be larg- er than those associated with crops, fruits and vegetables, paucity of statistical data on which to manufacturing and the service sector. base marketing, investment, or policy While basic data and indicators on livestock-related and decisions, or with which to assess socio-economic variables are available for most countries the efficacy of current — though often not sufficiently disseminated or adequately analyzed — there are few examples of rigorous statistical commitments or policies.” analysis and modelled projections, and still fewer that can generate causality arguments to demonstrate the contri- Global Strategy to bution of livestock to socio-economic development. This Improve Agricultural and Rural is partly because comprehensive datasets on livestock are Statistics, 2011 not usually available — e.g. in most economy-wide models, livestock is included in the agriculture aggregate. At the same time, the Ministry responsible for livestock is not mandated, ●● Number of commercial livestock enterprises and number/ and often not equipped, to undertake such analyses. Nor share of rural households keeping farm animals; does the Ministry typically have the power to influence significant change in data collection systems by national ●● Herd size and herd composition of livestock producers; authorities, usually the national offices of statistics. However, ●● Livestock production per TLU and/or per unit of labor; it can collate and interpret existing documentation, including from neighbouring countries, and collaborate with regional, ●● Total income and share of total income derived from live- national and international research institutes to rigorously stock for livestock-keeping households, disaggregated into demonstrate that investing in livestock is an effective way to rural/urban, male/female headed, and other variables of contribute to a number of socio-economic goals. development interest; ●● Level of livestock production, including shares of home WHOM TO TARGET? consumption and marketed product, for livestock-keeping households. Once the Ministry responsible for livestock development demonstrates that livestock sector investments can con- These and other indicators should be used to identify a tribute to some broad economic goal, and hence acquires typology of producers, spanning the range from subsistence- resources to invest for sector development, the next relevant oriented to specialized market-oriented livestock producers, question to answer becomes: through to large commercial farms. General typologies avoid pre ante targeting, which is often based on ethnic or Policies and investments are effective when they are con- other socio-cultural dimensions. Different typologies of sistent with the incentives of the livestock stakeholders, producers keep livestock for different purposes, use a variety amongst which the producers are likely to be assigned some of technologies and respond uniquely to changes in the priority. The Ministry, therefore, needs information on economic and institutional infrastructure, as determined current and emerging growth opportunities for animal-based by policy reforms within (and beyond) the sector. Such a food, the distinguishing characteristics of livestock producers typology has been proposed by Nouala et al. (2011): and products, and on the prioritized use of animals in tar- geted households. Basic data and indicators that serve this ●● Mixed subsistence-oriented livestock producers purpose include: are rural households that keep small herds, often mixing animals of different species; they sell a negligible part, if ●● Trends in, and the form of, the demand for various ani- any, of their livestock production; and derive a relatively mal-source foods, including unprocessed and processed small share of their cash income from livestock. For them, products nationally and regionally; any increase in livestock productivity — such as through QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations PART I. DEMAND AND SUPPLY OF LIVESTOCK DATA: GAPS AND ISSUES   |  23 reduction in animal mortality rate — has a positive WHICH CONSTRAINTS? impact on welfare. ●● Specialized market-oriented livestock producers are Once typologies of livestock producers have been construct- rural households that keep a (relatively) homogenous herd ed, the challenge arises as to how to create opportunities for — e.g. they could be specialized in milk or egg production growth and the following question becomes relevant: — sell a significant share of their livestock production; What are the critical and binding constraints that prevent and derive a significant part of their cash income from the different livestock producers from making better use livestock. Improvements in livestock productivity for of their farm animals? specialized market-oriented producers increase their cash income, assuming access to existing and growing market Policies and investments should attempt to relax or remove opportunities. These economic operators can also con- such constraints, particularly for key performance indicators tribute to the generation of off-farm jobs along the value such as livestock productivity, which limit the benefits that chain. producers derive from their animals. Simple data and indica- tors on factors that are deemed to influence production and ●● Commercial farms are specialized enterprises: that productivity provide preliminary information to decision maintain large homogenous herds, some permanent makers. Examples are: employees, and produce only for the market. Policies and investments to increase their productivity — such ●● Prevalence of selected animal diseases, i.e. proportion of as reducing trade barriers to access inputs — make small ruminants affected by goat plague (PPR, Peste des their business more profitable and competitive vis- Petits Ruminants) over the reference period; à-vis imports. Increases in their efficiency could also potentially reduce the real price of animal-source foods ●● Number and proportion of livestock producers with access in national markets — thus contributing to the food to veterinary services; who regularly vaccinate their ani- security of the (majority of) households that are net mals against selected diseases; who use de-wormers; who buyers of food — while generating a number of full time spray/dip animals against tick-borne diseases; on- and off-farm jobs. ●● Number and proportion of livestock producers feeding A variety of indicators can be used to define typologies of their animals with selected feeds or feed concentrates; livestock farms — e.g. herd size and composition, husbandry ●● Number and proportion of livestock producers with access practices, market participation, etc. Depending on the data to extension and financial services; available, countries may define their own typologies. While these data are useful, consultations with expert informants ●● Number and proportion of livestock producers who raise provide a complementary source of information on mean- improved/exotic breeds; ingful producer typologies. Indeed, data alone may generate typologies which are of little use to decision makers — e.g. ●● Number and proportion of livestock producers with a representative dairy farmer with 1.7 cows and selling 12 social networks/capital such as membership in marketing percent of the milk produced may be generated as an average cooperatives; taken across multiple modes in a dataset containing very few ●● Difference between farm-gate and retail-level prices for such individuals. A distinguishing element that in all cases live animals and major livestock products; should be taken into account is the household’s motive for keeping farm animals, in particular whether it is related to ●● Number and types of livestock markets (e.g. primary, subsistence or profit. This one factor will often condition the secondary), including location, frequency of operation and livestock producers’ response to different types of policies size; and investments. ●● Access to common property resources, availability of for- age, and sources and reliability of water used; QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations 24  |  Investing in the Livestock Sector: Why Good Numbers Matter ●● Access to infrastructure such as roads and calculated as a function of total outputs (milk consump- telecommunications; tion, milk sales, animal sales and manure outputs) and total inputs (family and hired labor, fodder and feed, ●● Number of processing plants, including potential and veterinary costs and other). used capacity. ●● Otieno et al. (2012) examine the determinants of techni- While levels, trends and shares of input-, output- and market- cal efficiency in different beef production systems in four ing-related variables provide relevant information to decision Kenyan districts. They conclude that the value of beef makers, more sophisticated analyses — which systematically production would increase if farmers adopted controlled link outputs and inputs — are critical to identify major deter- breeding methods; signed marketing contracts; hired farm minants of production and productivity, and hence to point managers; and if their off-farm income increased (due to to binding constraints and priority areas for investment. its being invested in the cattle operation). Not undertaking this type of more detailed analysis often leads to investments that do not address critical constraints, A critical challenge to formulating targeted interventions/ thus minimizing the impact of overall investment. What investments that ensure development impact is the paucity follows are examples of multivariate analyses that attempted of basic and comprehensive data and indicators on input-, to identify the determinants of livestock production and output- and marketing-related variables. Consequently ad productivity. hoc data collection and participatory processes are essential to identify productivity constraints, but a review of existing ●● Akter et al. (2003) examine the efficiency in poultry and work is also revealing. Such reviews find that, in general: pig production systems in Vietnam. Output is measured as value of production plus the change in inventory. For ●● When livestock data are available from household sur- pigs, it was revealed that land size, herd size, education veys, most subsistence-oriented livestock keepers are of household head and proximity to market are positively shown to lack access to even the simplest production associated with efficiency. Conversely, the age of the inputs, such as animal health services and feed (Bocoum household head, female-headed households, greater access et al., 2013; Covarrubias et al., 2012). This implies that to government supplied inputs, and higher proportion interventions that focus on ensuring access to basic of family-supplied feed materials significantly increase inputs are a straightforward way to improve livelihoods inefficiency. through investments in livestock. Indeed, analyses that target subsistence-oriented livestock keepers invariably ●● Ishaq et al. (2007) find that, in the small ruminant system conclude that increases in the use of basic input— such as of Southern North West Frontier Province of Pakistan, ex- forage, feed and animal vaccines — significantly increase panding the herd size generates larger returns, in terms of production. milk production, than any other investment. In addition, the study indicates that doubling all inputs more than ●● Analyses that target market-oriented specialized rural doubles total milk output. households and commercial enterprises typically con- clude that increases in productivity (efficiency) could be ●● Ashagidigbi et al. (2011) examine the production and pro- triggered by dozens of different actions, many of which ductivity of egg producers in Jos metropolis of Nigeria’s are beyond the control of the Ministry responsible for Plateau State. They find that larger flock sizes and a livestock (e.g. education, credit or year-around access to reduction in the cost of drugs would lead to an increase in roads). This calls for collaboration among government total production, as measured by the total number of eggs agencies, public and private decision makers, and an produced. agreement to use livestock as a catalyst for economic growth. ●● Gelan and Muriithi (2012) assess the economic efficiency of 371 dairy farms in Kenya, Rwanda and Uganda. They show that the adaption of improved breeds in the herd and feed and fodder innovations have significant positive effects on the levels of economic efficiency. The latter is QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations PART I. DEMAND AND SUPPLY OF LIVESTOCK DATA: GAPS AND ISSUES   |  25 ©FAO/Giulio Napolitano WHAT TO TARGET? feed concentrates, the use of which is anticipated to increase productivity, are: Once there is information on whom to target (with a clear ●● Availability of feed concentrates in rural markets; distinction of the intervention’s objectives, i.e. supporting livelihoods or expanding the sector’s contribution to econom- ●● Number of feed producers and their productive capacity; ic growth), and on the binding constraints they face — e.g. limited access to veterinary services for subsistence-oriented ●● Availability of pasture; livestock producers, or lack of credit for market-oriented ●● Relative prices of feed concentrates to the products to be livestock producers — the following area to explore is: produced, including their seasonal fluctuations; The identification of constraints and their subsequent pri- ●● Quality of available feed concentrates; oritization, in practice, provides little guidance on how to relax and remove them, nor the sequencing of interventions ●● Access to information on feed concentrates by livestock that is required to induce positive change. For example, producers. what can or should be done to ensure that farmers feed their animals with concentrates? How can the prevalence of Summary statistics associated with a particular constraint or selected animal diseases be reduced? How to promote the set of constraints, such as those listed above, help disentan- use of controlled breeding methods? In order to address the gle the root cause(s) of a constraint and, therefore, to better root causes of constraints, decision makers need a multitude focus any prospective investment. Analyses that attempt of data and indicators. Indicators relevant to our example of to identify rigorously the root cause of a constraint provide QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations 26  |  Investing in the Livestock Sector: Why Good Numbers Matter additional information for better targeting interventions on “There is... inadequate data the ground. Below are a few such examples of analyses: to demonstrate quantitatively ●● Jabbar et al. (2002) examine the supply and demand for the role of animal resources livestock credit in Ethiopia, Kenya, Nigeria and Uganda. They find that gender of household head, education, train- in African economies, ing, prevalence of outstanding loan and the number of and to use such data to create improved cattle on the farm, all have significant influence broad awareness among on household borrowing and liquidity. policy-makers and investors.” ●● Ajuha et al. (2003) study the demand for veterinary ser- vices in three States of India, namely Gujarat, Rajasthan AU-IBAR Strategic Plan, and Kerala. They show that in all the States the demand for veterinary services, as measured by the number of 2010–2014 veterinary visits over the reference period, is negatively associated with the price of the services and positively possible, invest resources to undertake specialized surveys associated with the service time, a quality indicator. targeting a set of likely constraints. Chapter 3.5, on combin- ing micro data with farmers’ views, presents a methodology ●● Bahta and Bauer (2007) assess the determinants of mar- to identify the root causes of binding constraints, thereby ket participation among small-scale livestock producers in facilitating the identification of priority areas for policies the Free State Province of South Africa. Their results sug- and investments. gest that market information, distance to the preferred marketing outlet, level of training, access to extension services and livestock fertility rate all have positive impact HOW TO INVEST? on farmers’ participation in livestock markets. Once information has been collected on whom to target, the ●● Costales et al. (2008) study the factors that influence constraints they face, and their root causes, the following participation in contract farming of pig producers in process needs to be followed to determine: Northern Vietnam. They conclude that level of education and large physical access holdings facilitate a farmer’s Decision makers should draft an implementation plan — engagement in formal contracts with large integrators. including roles and responsibilities of various actors and an estimated budget — which works to identify actions needed ●● Achoja et al. (2010) examine the determinants of the to relax or remove the root causes of one or more binding demand for veterinary services by commercial poultry constraints. It is clear that the uniqueness of countries’ or producers in the Delta State of Nigeria. They find that localities’ investments and limitations on data and indicators scale of production and distance to the nearest veterinary preclude the drafting of a fully informed evidence-based office significantly influence the use of veterinary services. implementation plan. Indeed, implementation of policy It is not feasible to access detailed information on all con- reforms and investments usually entail or include some form straints affecting livestock producers in all locations and of institutional change — new ways of doing things that have contexts of interest. Often, the most marginalized livestock not been yet tried out and for which data is therefore not systems offer the least amount of information. There are available. not, for example, readily available datasets with information For example, available information is unlikely to be of use in on the quality of animal feeds in a long list of rural markets assessing whether or not the quantity and quality of veteri- or on the price paid by farmers to vaccinate their animals. nary services in rural areas is best improved through forming This makes it challenging to both present basic statistics and a cadre of community animal health workers (a supply side conduct analyses of constraints. In formulation of policies intervention) or, alternatively, through the provision of and investments, decision makers should thus consult expert informants, promote participatory processes and, if QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations PART I. DEMAND AND SUPPLY OF LIVESTOCK DATA: GAPS AND ISSUES   |  27 veterinary vouchers to livestock keepers for the purchase of farmers. Information was given through an eight-page veterinary services and drugs (a demand side intervention). booklet containing pictures with messages on diagnosis and proper treatments. Results show that knowledge This in turn leads to a series of development questions for of trypanosomosis diagnosis and treatment are 23 and which little supporting information is usually available. How 14 percent greater, after 2 weeks and 5 months respec- many animal health workers should be trained? Does a one tively, in the treatment group than in the control group. week training suffice or is a two week course preferable? How Relatively simple information seems sufficient to reduce frequently should refresher courses be held? Should commu- the incidence of selected animal diseases. nity animal health workers be given basic equipment (e.g. needles, thermometers and a small stock of medicines, etc.) ●● Henning et al. (2009) conducted controlled trials in for free, or at cost? 124 randomly selected backyard poultry keepers in nine villages in Myanmar to evaluate two strategies In order to answer these types of questions, decision makers aimed at reducing chicken mortality, namely Newcastle can review development projects and examine past experi- disease (ND) vaccination using a thermostable vaccine ence, conduct participatory decision making processes, or set and changes in the management of chick rearing up pilots by which different alternatives are tested on a small (confinement and supplementary feeding). They scale to identify the most effective, which can then be scaled find that vaccination against ND resulted in a lower up. Some reviews include the following: incidence rate of mortality during ND outbreaks in households with vaccinated birds, but that crude ●● Pica-Ciamarra et al. (2010) provide a comprehensive mortality rate in chicken did not decline and was review of alternative policy instruments, including pros lower in households with altered chick management. and cons for their implementation, in different live- From a policy perspective, investing resources to stock-related domains, such as risk-coping; animal health; reduce mortality incidence due to ND makes sense feed and forage; access to credit; livestock research; trade; only if all-cause mortality incidence is also reduced. and other. They show, for example, that the quantity and quality of veterinary services could be improved through ●● Bandiera et al. (2012) undertook a randomized eval- alternative institutional reforms, such as cost-recovery uation of an entrepreneurship program that provides mechanisms; joint human-animal health service delivery; assets — including cows, goats and poultry birds — and sub-contracting; provision of smart subsidies to service training to run small businesses to the poorest women provides or to livestock farmers; the establishment of in rural Bangladesh. They find that, after two years, community-based animal health workers; and other. women participating in the program allocate more time to self-employment (and less to wage-labor), which results ●● Murphy et al. (2003) compare the efficacy of three school in higher income, higher per-capita expenditure, and snacks in improving growth and cognitive function of improved food security for their families. children in rural Kenya. The snacks are composed of equi-caloric portions of githeri (a vegetable stew), includ- ●● Wanyoike and Baker (2013) analyzed 58 livestock ing githeri alone, githeri plus milk, and githeri plus meat. development projects to identify factors affecting their Total energy intake increases more with the githeri plus effectiveness. Key factors were revealed to be large proj- meat snack than with the other two, because the addition- ect size, specialization in livestock issues, inclusion of al energy provided by the githeri alone and by the githeri government in key communication roles, inclusiveness of plus milk is counterbalanced by a decrease in the energy implementation of exit strategy formulation, and target- content of the food consumed at home. From a policy ing of interventions at several levels of the value chain. perspective, the provision of githeri meat snacks to rural schoolchildren is shown to be an optimal strategy if the To enhance the probability of good intervention design and objective is to improve their nutritional status. implementation, decision makers should assess and rank alternatives, with additional information sourced from ●● Grace et al. (2008) carried out a control trial in South expert informants, through participatory and consultative Mali to assess the effects of providing information on the processes; and from past projects and experience, including diagnosis and treatment of bovine trypanosomiasis by QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations 28  |  Investing in the Livestock Sector: Why Good Numbers Matter those from other countries. As a practical alternative, one ●● Output indicators, which measure the immediate more visible to stakeholders, ex ante evaluations can be effects as determined by access to inputs, e.g. whether undertaken through pilots on a limited scale that are geared more animals are vaccinated against certain diseases as a for scaling up. consequence of increased numbers of veterinarians. ●● Outcome indicators, which quantify the effects generat- HOW TO ENSURE EFFECTIVE ed by the outputs, e.g. reduced incidence of certain animal IMPLEMENTATION? diseases. ●● Impact indicators, which measure the effects of the Once investment choices have been examined and policy outcome beyond its direct and immediate results, e.g. options identified, impact is often determined by anticipating increased animal productivity and improved households’ data and information needs that ensure effective policy im- livelihood. plementation and targeted investments. In general, input and output indicators should be readily Critical to monitoring the effectiveness of development in- accessible and measurable, as they relate and can be collected terventions is the existence and/or establishment of a robust within the daily or regular activities of some actors. Outcome monitoring and evaluation system, which regularly assembles and impact indicators are harder to measure and baselines quantitative and qualitative indicators of success and project more difficult to derive, which often makes it difficult to progress. There exist large numbers of reference documents properly monitor and assess project/policy impact. In addi- on monitoring and evaluation (e.g. EC, 2006; UNDP, 2009), tion, attribution is complicated in many circumstances with which target four types of indicators: outcomes and impacts influenced by a variety of factors, ●● Input indicators, which show whether appropriate including but not restricted to changes in the known inputs financial, human and physical resources are allocated to and outputs. policy and investment implementation. An example is the number and recruitment of public veterinarians. CONCLUSIONS ●● Identify the major constraints that prevent the various types of livestock producers from making the best use of their animals. Decisions on investment and policy formulation in the livestock sector entail a thought process that has been ●● Identify and rank the root causes of the constraints, detailed here in terms of sequencing and specificity of which represent the priority areas for investments. information needs. It is clear that decision makers need information on a variety of data domains in order to: ●● Design effective policy and investment implemen- tation plans, including specification of roles and ●● Demonstrate that livestock sector development can responsibilities of the various actors and an estimated contribute to the broader socio-economic goals of the budget. country. ●● Monitor and evaluate the implementation of policy ●● Define some typologies of livestock stakeholders, reforms and investments. including a clear distinction between market-orient- ed and subsistence-oriented producers, who have different needs and respond differently to policy and institutional change. QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations PART I. DEMAND AND SUPPLY OF LIVESTOCK DATA: GAPS AND ISSUES   |  29 BOX 3. A TOOL FOR THE INCLUSION OF LIVESTOCK IN THE CAADP COMPACTS AND INVESTMENT PLANS T he Comprehensive Africa Agriculture Development Pro- gramme (CAADP) has been endorsed by African heads of state and governments as a vision for the restoration of indicators at national level, which help appreciate whether there are opportunities for livestock sector development to contribute to economic growth, food security and poverty agricultural growth, food security and rural development in reduction. Africa. CAADP aims to stimulate agriculture-led develop- ment that eliminates hunger and reduces poverty and food Module III, Livestock in the Household Economy, rec- insecurity by targeting investments in four pillars: land and ommends that the CCTs collect/generate core livestock water management; market access; increasing food supply indicators at household level, to help understand the role and reducing hunger; and agricultural research. AU-IBAR of livestock in the household economy, including con- is mandated to assist AU member countries to implement straints to productivity. Ultimately, this module aims at the livestock component of the CAADP. To this aim, it identifying priority areas for livestock sector investments. has developed a Tool for the Inclusion of Livestock in the Module IV, Livestock in the CAADP Compacts, clusters CAADP Compacts (AU-IBAR, 2013), which is largely consis- Module I and Module II national and household level in- tent with the stepwise approach presented in this chapter. dicators around the four CAADP pillars, namely land and The Tool identifies a number of core livestock indicators water management; market access; food supply; and agri- that country governments should collect/generate to ade- cultural research. This module assists the CCTs in ensuring quately represent livestock in the CAADP Documents. The that livestock investments are consistent with the CAADP Tool consists of five interrelated modules. framework and priorities. Module I, Mapping and Consulting Stakeholders, assists Module V, Post-Compact Livestock Investments, gives the CAADP Country Teams (CCTs) in identifying and con- some basic indications on the data/indicators needed sulting stakeholders who appreciate the many channels to formulate, implement and monitor & evaluate the through which livestock contribute to economic growth livestock component of the CAADP National Agriculture and livelihoods, including the monetary and non-monetary Investment Plan. It also delves into the importance of value of farm animals. experimenting or testing alternative implementation Module II, Livestock in the National Economy, suggests mechanisms on a small scale before scaling out invest- that the CCTs collect/generate a key set of core livestock ments to the entire country. • In particular, knowing with statistical precision the number the core livestock indicators as identified in chapter 1.2 and of animals and the number of livestock farmers at some low some other data and indicators, complemented by inclusive administrative level, such as the district or county level, is participatory policy processes, consultations with experts, essential information for effectively designing any interven- synthesis of existing experience and analysis, and rigorous tion on the ground. At the same time, it should be recognized ex ante pilots, can assist decision makers in designing and that the data and indicators needed to properly design policy implementing policies and investments that are to a large and investment implementation plans are largely unavailable extent effective in promoting a sustainable livestock sector. or inadequate due to the novelty and uniqueness of the The next chapter presents a critical review of the prevailing intervention. Targeted ad hoc surveys may help reduce this agricultural/livestock data collection system to appreciate information gap at one or more stages of the question-driven what indicators/statistics they are able to produce on a process described here. regular basis. Complete information with all the desired data sets is obviously not achievable, nor economically optimal, and the risk of designing bad policies and investments can never be reduced to zero. However, a statistical system that generates QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations 30  |  Investing in the Livestock Sector: Why Good Numbers Matter 1.4 DATA COLLECTION SYSTEMS AND LIVESTOCK INDICATORS: GAPS AND PRIORITY ISSUES KEY MESSAGES to assess if the collected data suffice to generate the core livestock indicators (as identified in chapter 1.3), namely Numerous methods exist for collecting livestock value added; livestock population; livestock pro- livestock data which range from regular sample duction; average market prices for live animals and livestock surveys and complete enumeration censuses to products; outbreaks of animal diseases, number of animals administrative records and one-off, or ad hoc affected, and number of animals at risk. It also identifies surveys. other relevant livestock indicators that major surveys help generate. Below are the major systems of data collection that are discussed in the following sections: Because the spatial distribution of animals is only partially correlated with the distribution of rural ●● The agricultural/livestock census; households or farms, sampling issues should ●● Agricultural and livestock sample surveys; be given particular attention when designing surveys that aim at generating official livestock ●● Household budget surveys; statistics. ●● Living standards measurement studies; ●● Administrative records or routine data; While a variety of methods exist for collecting livestock data, no single survey satisfies the ●● Others, such as the population and housing census and information needs for policy and investment labor surveys. requirements. Data integration and ad hoc The chapter concludes with a summary table that highlights collection of data are recommended to generate the main core and other livestock indicators available from adequate information on livestock. major agricultural and non-agricultural surveys, and iden- tifies gaps in the demand and supply of livestock data, both from a quantity and quality perspective, as per the findings of a global survey undertaken by the Livestock in Africa: Improving Data for Better Policies Project. MULTIPLE SOURCES OF LIVESTOCK DATA THE AGRICULTURAL CENSUS AND Core livestock indicators and other indicators needed for THE LIVESTOCK CENSUS livestock sector policies and investments could be generated by multiple data collection systems, including regular and The largest agricultural statistical operation in any country is one-off, or ad hoc, surveys. Each country, depending on its the agricultural census. Country governments — namely the priorities and resources, could implement — with some reg- Statistical Authority in collaboration with relevant Ministries ularity — a variety of agricultural surveys, which also target — usually undertake the agricultural census every ten years, livestock, as well as other non-agricultural surveys which may with the objectives to: collect livestock-related information. ●● Generate information which reveals the structure of the This chapter reviews the prevailing and most common agriculture sector, especially for small administrative systems of agricultural and non-agricultural data collection units; implemented across Africa, with the ultimate objective QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations PART I. DEMAND AND SUPPLY OF LIVESTOCK DATA: GAPS AND ISSUES   |  31 “Complete enumeration is, however, costly and difficult to implement. Consequently, many countries have been undertaking sample agricultural censuses or large-scale surveys, which collect information from a sample of agricultural holdings.” ●● Generate data to use as benchmarks for other agricultural the attributes of a full census, even if statistics at the lowest statistics; levels, such as villages, cannot be generated. ●● Provide frames for agricultural sample surveys. The livestock content of the agricultural census always in- cludes information on: The agricultural census collects, processes and disseminates data on a limited range of structural items of agriculture, ●● The number of animals on the holding by species. which change relatively slowly over time. These typically include size of agricultural holdings, land tenure, land use, Species include cattle and buffaloes; sheep and goats; pigs; crop areas, irrigation, livestock numbers, labor, ownership of chicken, ducks, geese and turkeys and other birds; horses, machinery, and use of some agricultural inputs. asses, mules and hinnies; other animals, such rabbits, dogs and cats; and insects such as bees (counted on the basis of Data are collected from agricultural production units, or hives) and silkworms. The number of animals refers to those agricultural holdings. In developing countries, most agricul- animals raised/held by the holding on a specific reference tural holdings are associated with a (small) farm household date, which is usually the day of enumeration. Sometimes and relatively few commercial farms, i.e. data are largely animals are differentiated by age and sex, e.g. cattle are split collected from smallholders. Face-to-face interviews with into cows, bulls, steers, heifers, male and female calves; oc- the agricultural holder or the enterprise manager by trained casionally, differentiation is made between indigenous/local enumerators is the most common technique of data collec- and improved/exotic breeds. tion, though telephone and internet-based interviews have been also utilized. Data are collected in a short time-span, Compared to agricultural censuses, livestock censuses collect occasionally in just one week. more detailed information on livestock, the content of which varies by country and the focus is often dictated by the Data are collected on a complete enumeration basis — i.e. prevailing policies and programs which need to be monitored information is obtained from all production units in the and evaluated. This may include one or more of the follow- country — which allows for the compilation of statistics ing (MAAIF and UBOS, 2009; République du Mali, 2007; even at the lowest administrative units, such as the village. République du Niger, 2007b; URT, 2010): Complete enumeration is, however, costly and difficult to implement. Consequently, many countries have been under- ●● Livestock numbers by type of breed; taking sample agricultural censuses or large-scale surveys, ●● Livestock numbers by production systems (e.g. zero graz- which collect information from a sample of agricultural ing, tethering, communal grazing, stall-fed, etc.); holdings. ●● Economically active population in the livestock sector; For example, the National Sample Census 2007/08 of Tanzania collected data from about 53,000 farming house- ●● Livestock pest and parasite control methods and access to holds, or about 17 percent of all farming households (URT, animal health services/drugs; 2010); the 2008 National Livestock Census of Uganda collected information from about 964,000 households, or ●● Types of animal feed used; 15 percent of all households (MAAIF and UBOS, 2009). ●● Sources of water for animals; Samples of such sizes are usually sufficient to retain many of QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations 32  |  Investing in the Livestock Sector: Why Good Numbers Matter Of course, when sample censuses are conducted, there are sampling errors linked to the estimates of the livestock population. This is more the case when the data are from agricultural sample censuses that collect information from agricultural holdings, which may or may not hold livestock. Sampling errors are less pronounced for data derived from livestock sample censuses, where statistical units are live- stock holdings. These are thus expected to provide a more precise estimate of the livestock population than agricultural sample censuses. AGRICULTURAL AND LIVESTOCK SAMPLE SURVEYS Agricultural sample surveys, including specialized livestock sample surveys, provide governments with structural data on the sector to supplement census information that is usually available every ten years. These surveys provide additional information needed to better design, implement and monitor sector investments. Data from sample surveys: ●● Provide broad indications for development planning and investments in the sector, including public sector interventions; ©FAO/Issouf Sanogo ●● Help monitor trends in structure and assess performance of the agricultural / livestock sector. Agricultural/livestock sample surveys target a relatively small ●● Level of production, i.e. number of animals slaughtered, sample of agricultural holdings. For instance, the sample of litres of milk produced and number of eggs. Usually, the Rwanda National Agricultural Survey (NISR, 2010) and censuses provide information on the quantity of produc- that of the Permanent Survey of Agriculture of Burkina Faso tion, not on the value of production, as price data are not (MAHRH, 2009) both consisted of about 10,000 households. collected; Samples are usually large enough to generate statistics that are representative on a national level and for major ●● Ownership of equipment, such as ox-ploughs, ox-planters agro-ecological zones/administrative regions. In few cases, and ox-carts; such as the 2011–12 Ethiopia Livestock Sample Survey that covered about 68,000 agricultural households, statistics can ●● Consumption of animal-sourced foods. be also generated for lower administrative units, such as Agricultural/livestock censuses provide the ‘gold standard’ local districts (CSA, 2012). Sample surveys may cover the in generating accurate statistics on the livestock population entire livestock sector, or target only some specific livestock in a country, while also providing critical information on sub-sectors and/or geographical areas, such as the 2004 the geographical distribution of animals. They also generate National Cattle Survey in South Africa (Scholtz et al., 2008) information on the structure of the herd, which is required to or the 2005/06 Livestock Survey in the Arid Land Districts estimate and project growth rates of animal populations. of Kenya (ALRMT, 2007). Similar to agricultural censuses, face-to-face interviews by trained enumerators with the agricultural holder is the most common technique of data QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations PART I. DEMAND AND SUPPLY OF LIVESTOCK DATA: GAPS AND ISSUES   |  33 collection. These surveys are usually undertaken by the Four features of agricultural/livestock sample surveys are Statistical Authority, even though the Ministries responsible worth noting. First, they attempt to capture information on for animal resources may also carry out livestock sample both inputs and outputs, which allow building some indi- surveys. cators of productivity. Second, these surveys often include information on prices, both for inputs and outputs, which The livestock content of agricultural and livestock sample are essential to arrive at some measure of profitability and surveys is significant, and particularly comprehensive in the competitiveness of livestock farming. Additionally, this facil- latter. In addition to an agricultural questionnaire, which itates an identification of bottlenecks along the value chain. collects information on basic household characteristics and Third, they capture information about seasonality in live- detailed information on agriculture/livestock, these surveys stock farming through enumerators visiting households in often include a community questionnaire that collects infor- different seasons, or when respondents are asked to provide mation on public services, community infrastructure, market information for selected questions by season. For milk pro- prices, etc. The livestock information available from these duction, disease outbreaks, live animals marketing and other surveys usually comprises (ALRMT, 2007; MAHRH, 2009; dimensions, this seasonal information is important for mon- NISR, 2010; Scholtz et al., 2008; Somda et al. 2004): itoring the sector. Fourth, these surveys occasionally include a question on the household rationale for keeping farm ani- ●● Livestock number, by species, breed and age; mals, which is a crucial consideration when seeking to make ●● Herd dynamics over the reference period (usually one effective investments. Interventions need to be consistent year). Indicators include animal births and deaths, ani- with the incentives influencing households’ objectives for mals lost, slaughtered, marketed and given/received as rearing livestock. Objectives could include self-consumption gifts, etc. This allows projecting herd growth, a critical of animal food, income generation, security/insurance, and piece of information for investment design; input into the agricultural sector (manure/animal traction) among others. ●● Livestock production (meat, milk, eggs, etc.), including both quantity and value, i.e. price data are collected in Agricultural and livestock sample surveys are often per- these surveys; ceived as the best information sources for identifying major constraints to livestock productivity and opportunities for ●● Animal vaccination, diseases outbreaks and treatment, investments at the farm level. However, they rarely cover all and access to animal health services. dimensions of livestock production, nor do governments in sub-Saharan Africa systematically undertake them. Finally, it Supplemental livestock information, dependent on the type is worth noting that there are sampling errors when deriving and objectives of the survey, can include: national/regional/district livestock statistics from agricultur- ●● Feed for animals, e.g. fodder from land and hedges; scat- al and livestock sample surveys. These are more pronounced tered stalks and market purchased feed, etc.; in the case of agricultural sample surveys, where the statisti- cal unit is the agricultural holding that may or not keep farm ●● Water sources, e.g. rivers, boreholes, wells, etc.; animals. ●● Family and employed labor devoted to livestock by type of activity, e.g. feeding, watering, sales and other; “Agricultural and livestock sample ●● Ownership of livestock-related assets, such as ox-carts, surveys are often perceived as the best ox-ploughs, sheds for animals, etc.; information sources for identifying ●● Distance to markets (in time or space); major constraints to livestock ●● Market infrastructure (e.g. animal health posts; slaughter productivity and opportunities for slabs; markets); investments at the farm level.” ●● Consumption of animal-source foods. QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations 34  |  Investing in the Livestock Sector: Why Good Numbers Matter HOUSEHOLD BUDGET SURVEYS from fresh, chilled and frozen beef to dried, salted or smoked meat, and from whole milk to cheese and curd (LBS, 2008). Household Budget Surveys — also called Family Expenditure To measure livestock income, a direct question is usually Surveys, Expenditure and Consumption Surveys, and Income asked about revenues from different activities, including and Expenditure Surveys — collect, process and disseminate wage employment and self-employment in crops and live- information on key components of household’s budget and stock; in a few cases, some details about sales of livestock expenditures with the objective to: and livestock products and expenditures are asked to the respondents, which allows for a better estimate of live- ●● Update the weights in the CPI, a critical piece of informa- stock income. For example, the 2009/10 Uganda National tion to estimate national macro indicators, such as the Household Survey includes a question about income from level of inflation; livestock farming over the last 12 months, differentiated ●● Measure poverty and well-being; by cash and in-kind income (UBOS, 2009); the 2007 Niger Household Budget and Consumption Survey (République du ●● Generate estimates on household consumption, which Niger, 2007b) includes detailed questions about ownership of feed into the calculation of the Gross Domestic Product livestock and sale of live animals and livestock products. (GDP). Statistics on consumption from Household Budget Surveys Household budget surveys are conducted on a sample of are designed to be representative at the national level and for nationally representative households and for agro-ecological macro-regions/agro-ecological zones. Again, challenging the zones/major regions. For example, the sample size of the compilation of results and the reliability of the statistics on 2002/2003 Lesotho Household Budget Survey comprised livestock variables, except for consumption of animal-sourced 5,992 households, which was representative of the country foods, is the issue of potential sampling errors, as all house- and its ten districts (LBS, 2008); the 2001 Household Survey holds and not just livestock-keeping households are the of Senegal included 6,624 households, representative na- statistical units for this type of surveys. tionally and for the 14 regions of the country (DPS, 2004). Similar to other surveys, data are usually collected through face-to-face interviews, but these surveys are unique in that the data is usually collected over a one year period to capture seasonal variations in expenditure patterns. Some informa- tion may be also collected daily, such as food consumption and/or expenditures. The responsible agency for implementa- tion of Household Budget Surveys is the National Statistical Authority. Two relatively unique data sets typically collected through Household Budget Surveys include: ●● Consumption of animal-source foods, an important indi- cator of nutrition and well-being; ●● Livestock income and its contribution to total household income. Questions on consumption of animal foods are usually based ©FAO/Giulio Napolitano on a seven-day recall period. For example, the 2002/03 Lesotho Household Budget Survey includes questions on weekly expenditures on several livestock products, ranging QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations PART I. DEMAND AND SUPPLY OF LIVESTOCK DATA: GAPS AND ISSUES   |  35 LIVING STANDARDS MEASUREMENT ●● Consumption of animal products, including self-consump- tion and market purchases. STUDIES In recent years, with the growing recognition of the role of Living standards measurement studies (LSMS) are multi-top- agriculture for livelihoods, poverty reduction and economic ic household surveys that aim to: growth, the agricultural section of LSMS surveys has been expanding in its coverage, including its livestock content. ●● Measure poverty and well-being and understand their Recent LSMS surveys in Niger (République du Niger, 2010), major determinants; Tanzania (NBS, 2012a) and Uganda (UBOS 2011) include a ●● Provide evidence for planning, monitoring, and evaluating specific section on livestock that collects not only informa- economic policies and social programs in relation to their tion on livestock ownership, herd dynamics and consumption impact on household living standards, especially those of of animal-sourced foods, but also on: the poor. ●● Breeds, differentiated by local/indigenous and improved/ LSMS surveys are administered to a nationally representa- exotic; tive, but relatively small sample of households. This allows ●● Use of inputs, including feed, water, labor; the generation of accurate, or nationally representative, statistics for the country as a whole and for large sub-areas ●● Access to livestock-related services, such as veterinary (e.g. rural and urban areas; macro-regions). For instance, drugs, vaccination, extension; the sample of the 2005 Ghana Living Standard Survey consisted of 8,700 households (GSS, 2008); that of the 2004 ●● Husbandry practices, e.g. housing and breeding practices; Zambia Living Conditions Monitoring Survey comprised ●● Production of livestock products, including not only meat, about 20,000 households (CSO, 2005). Data in these surveys milk and eggs, but also dung and other services provide by are collected by the National Statistical Authority — with livestock, such as transport. increasing use of computer-assisted technologies — through face-to-face interviews, aften over a period of 12 months in LSMS surveys, and particularly those with a comprehensive order to take into account any seasonality. livestock module, are the best sources of information for quantifying the contribution of livestock to household liveli- A unique feature of LSMS surveys is their inclusion of several hoods, including both its monetary and non-monetary value. questionnaires that target a variety of information at the In addition, this type of data can facilitate analysis, ex ante household and community level. They include a household and ex post, of the impact on livelihoods of selected livestock questionnaire, a community questionnaire, a price ques- sector interventions. However, in most cases livestock is still tionnaire and, in some cases, questionnaires on agriculture, unappreciated in LSMS surveys and, given that the sample gender, and/or fisheries. The household questionnaire of agricultural questionnaires targets only rural households comprises sections on education, health, etc.; the agriculture and that sample sizes are small, national level statistics for questionnaire includes modules on crops, extension services, livestock cannot be always generated with precision from and in some countries a significant number of livestock these surveys. questions; the community questionnaire targets information on local infrastructure, availability of public services, and distances to major markets, etc. ADMINISTRATIVE RECORD DATA LSMS surveys include some livestock-related questions, Administrative record data, also referred to as routine data, which target: are regularly collected by national governments, in collabo- ●● Livestock ownership, sometimes with details on herd dy- ration with districts or lower level administrative units, with namics (animals born, death, lost, etc.) over the reference the objective of: period, usually one year; ●● Planning, implementing and monitoring the delivery of public services. QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations ©FAO/Issouf Sanogo 36  |  Investing in the Livestock Sector: Why Good Numbers Matter Within a country, government officers at a specifically des- in general, live cattle by breed (e.g. local/indigenous versus ignated local administrative level (e.g. sub-county, district) improved/exotic), or be by head or weight (kg/live animal). In collect agricultural data, including livestock-related data, on principal, whatever the statistical unit, government officers a regular basis — such as monthly or quarterly. They report are expected to collect data on a complete enumeration basis, to the district administrative unit, which processes the data, i.e. sampling errors are not anticipated in routine data (LDIP, uses it when needed, and then reports to a higher level in the 2010b, 2010c, 2011c, 2012b). administration. The Agriculture and/or Livestock Ministry obtain access to this livestock data and statistics on a regular, In general, routine data primarily target: or occasionally irregular, basis. An example of administrative ●● Outbreaks of animal diseases and other animal-health data includes cross-border trade statistics, with Customs related indicators; Authorities at border points documenting trade flows of imports and exports (quantity and value) of live animals, ●● Livestock population; animal-source foods and other livestock products (e.g. hides and skins), which are then summarized in monthly, quarterly ●● Production of livestock products; and annual reports. ●● Trade of live animals and livestock products; The statistical unit for administrative record data varies and ●● Market prices of major livestock items to be included in is a function of what data is being collected by which admin- the CPI. istrative office. For instance, data on prices of live animals may be collected by extension officers at local markets, or by The content of administrative data varies by country and custom officers at the border; the price may refer to live cattle reporting period (e.g. monthly, quarterly). In Uganda, for QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations PART I. DEMAND AND SUPPLY OF LIVESTOCK DATA: GAPS AND ISSUES   |  37 instance, livestock/veterinary officers at the sub-county level statistical procedures are used to select the sample popula- collect information on a monthly basis at the village level. tion, while concepts and definitions used are often unsuitable This information includes the number of animals by pro- for statistical purposes. Furthermore, they rarely conform duction system and species; animal movements; outbreaks to international standards and may even differ from district of contagious diseases, including the number of animals to district. There is a need for caution, therefore, when using affected, dead/slaughtered and treated, and control measures; administrative records to generate official statistics (Okello et number of animals vaccinated against selected diseases, such al., 2013). as Contagious Bovine Pleuropneumonia (CBPP), Brucellosis and Rift Valley Fever; clinical cases handled by local animal health staff by type, such as diarrhea or mastitis; number OTHER SOURCES OF LIVESTOCK DATA of meat inspections (ante-mortem and post-mortem) and condemnations rate; number of animals slaughtered; sales of There are a number of other sources for livestock-related livestock animals, and prices (average, minimum, maximum); data, including: etc. (MAAIF, no date). ●● The Population and Housing Census; Some of the information and data collected, particularly that ●● Service Delivery Surveys; related to animal disease outbreaks, respond to international ●● Labor Force Surveys; obligations which require African countries to submit month- ●● Marketing Information Systems; ly, quarterly and annual animal health/disease reports to the ●● Experimental Station Records; World Organization for Animal Health (OiE) — the refer- ●● One-off Livestock Surveys. ence organization to WTO for trade-related animal disease The Population and Housing Census, which is conducted matters — the Africa Union-Interafrican Bureau for Animal every ten years by almost all governments, may include one Resources (AU-IBAR); and selected Regional Economic or more screening questions on livestock. Typically, one ques- Communities (RECs). tion will target ownership/non-ownership of farm animals The importance of animal numbers data, in particular, the and a second one the number of animals owned by species. number of animals affected by a disease, is a critical piece This is the case in the 2012 Population and Housing Census of information for emergency interventions related to of Tanzania (NBS, 2012b). Since the Population and Housing animal health, e.g. to assess the number of vaccines needed Censuses target all households, the inclusion of livestock to prevent the spread of some epidemic disease. Data on screening questions help generate an appropriate sample production of livestock products (quantity rather than value) frame for specialized livestock sample surveys and statisti- are collected as a rough measure of the performance of the cally precise estimates of the livestock population. There are sector, which helps monitor the impact of government pol- concerns, however, whether households correctly report their icies and programs. Finally, statistics on trade are a critical livestock assets in the context of such surveys. Another issue piece of information to estimate livestock value added, and is that animals in commercial enterprises are not counted in hence GDP. the census. Routine data provide a major source of information for the livestock sector. Because of the regular information flow, they are essential to deliver public services and monitor the animal health status in a country as well as trade movements. However, there is dissatisfaction with the quality of routine data in African countries. Financial and human resources are limited at the local level, as are incentives for data collectors. There is rarely a systematic and common approach to collect ©FAO/Carl de Souza routine data at local level, with local governments and extension officers using different methods. Routine data are rarely collected from all the relevant statistical units and no QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations 38  |  Investing in the Livestock Sector: Why Good Numbers Matter BOX 4. LIVESTOCK QUESTIONS IN THE POPULATION AND HOUSING CENSUS T he Population and Housing Census is the largest sta- tistical operation undertaken by country governments, every ten years on average. The census collates information census and of agricultural sample surveys, e.g. by improved targeting (minimum farm size). Undertaking the Population and Housing Census jointly with the Agricultural Census or on the quantity and quality of so-called human capital at with agricultural sample surveys, or the latter soon after the the national, regional and small area level, and on housing former, would enable the analysis of a much wider set of and a population’s access to basic services, such as water, data, with the farm household allowing for direct linkages electricity and telephone landlines. Results of the census, between the different datasets. which have very limited sampling errors, are used to ensure efficiency and equity in the distribution of public resources, A number of agricultural data items can be included in the such as for roads, human health facilities and schools. They Population and Housing Census, including on agricultural are also used as benchmarks for statistical compilation and holders and their characteristics (e.g. sex and age); farm as a sampling frame for sample surveys, upon which many area; crops grown; ownership of agricultural machinery; countries rely for the generation of good quality statistics on types of production system and purpose of production; targeted domains. The Population and Housing Census uses ownership and use of livestock; land tenure; agricultural the household as its basic unit. The Census of Agriculture labor force; gender; and other. The FAO UNFA Guidelines for and other agricultural sample surveys use the agricultural Linking Population and Housing Censuses with Agricultural holding as their basic unit. In developing countries, the larg- Censuses present examples of Population Census Question- est share of agricultural holdings are managed by the farm naires (FAO and UNFPA, 2012). These, in most cases, contain household, i.e. a household in which one or more members the following two questions on livestock: are engaged in agricultural production activities. It follows ●● Whether the household rears farm animals and, if yes, that, if farm households were identified in the Population which species (e.g. cattle; pigs; poultry; etc.); and Housing Census, linkages with the census and the Agricultural Census and other agricultural surveys could be ●● The number of animals reared by species. generated, with a multitude of benefits: Responses to the first question are essential to build an The inclusion of farm households in the Population Census effective and up-to-date frame for a livestock census or a allows for identifying all agricultural holdings in the country specialized livestock sample survey, which may even target and, hence, provides a basis to build a sound sample frame one specific sub-sector of livestock (e.g. small ruminants). for the agricultural census and for agricultural sample sur- Responses to the second question provide an estimation of veys. If some questions on agriculture were asked in the the livestock population in the country, which is particularly population census, the agricultural census could be reduced relevant for countries that rarely undertake the Agricultural in scale, thereby generating savings. This information could Census and/or undertake Agricultural Sample Censuses. • also be used to better define the coverage of the agricultural Service delivery surveys aim at providing an assessment Labor force surveys facilitate an understanding of the sta- of quantity/quality trends in public service delivery. They are tus and trends of local labor markets. These sample surveys sample surveys that allow the generation of national level ask questions on the status of employment for the economi- statistics, which are also differentiated by rural and urban ar- cally active population (e.g. full-time or part time; employee eas and macro-regions. Some questions in this type of survey or self-employed; unemployed; etc.). They may include some can target livestock-related services, such as access to animal questions on livestock. For instance, the Botswana Labour health and extension services. Sampling errors, however, Force Survey explicitly estimates the economically-active may make it difficult for these surveys to properly assess the population working in commercial livestock and poultry quality of livestock-related services, which are targeted at a enterprises (CSO, 2008). relative small segment of the population. QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations PART I. DEMAND AND SUPPLY OF LIVESTOCK DATA: GAPS AND ISSUES   |  39 Enterprise surveys are firm level surveys of a representa- DATA COLLECTION COSTS tive sample of commercial private enterprises, which include livestock-related businesses, such as milk processors and Cost of surveys depend on a variety of factors, including commercial ranchers. Unless they specifically target agri- sample size, length and complexity of questionnaire, distri- culture, and livestock within agriculture, these surveys do bution of the population across the territory, and method not supply enough data to produce official livestock-related of data collection (e.g. paper versus computer-assisted data statistics, such as the average number of full and part-time collection). In addition, the budget should also consider costs employees; level of production; share of production sold related to survey preparation, such as sample design and internally, or exported for commercial livestock-related training of enumerators, and for data analysis and dissemi- companies. nation. Major costed activities while undertaking a statistical Market information systems (MISs) aim to provide survey are the following: farmers, traders and other actors along the supply chain with ●● Preparation and testing of the questionnaire; short-term information on price levels (to guide marketing decisions) and generate medium/long-term information on ●● Printing of questionnaire and/or purchase of comput- market trends (to guide investment decisions). Data are usu- er-assisted interviewing equipment; ally collected by so-called market monitors in major markets in the country and disseminated through a variety of means, ●● Training of enumerators; such as market boards, newspapers, radio, and websites, ●● Sampling; such as for the Tanzania Livestock Information Network Knowledge System (LINKS). There hardly any examples of ●● Data collection, including travel; market information systems that have been operational for more than a few years (LDIP, 2011d). ●● Data analysis; Experimental stations are usually mandated by research ●● Report writing and dissemination. agencies/institutions to conduct field- level research with The main budget items include: objectives to assess performance of certain breeds/vaccines/ drugs/feed/ husbandry practices/etc. in targeted agro-eco- ●● Personnel (salaries), including survey designers, enumera- logical zones. Data from these stations cannot be used to tors, drivers, translators, etc.; generate statistics, but are highly valuable in providing indications on the data quality from other statistical sources, ●● Personnel (per diem); and for identifying options for technical investments in the livestock sector. ●● Transportation; Finally, there are one-off livestock surveys, which are un- ●● Consumable, such as papers, pencils, cartridges, etc.; dertaken to respond to specific information needs. These can ●● Equipment, such as weighing scales and meters and, in be quantitative and/or qualitative; target the entire livestock some cases, computers; sector or only specific sub-sectors; review the entire livestock supply chain from input supply to production to consump- ●● Miscellaneous costs, such as phone calls and photocopies. tion of animal sourced foods, or only focus on some of its segments; be nationally representative or be implemented in While identifying major budget items is straightforward, ar- selected regions and zones; target actors along the livestock riving at some general estimation of the costs of agricultural/ supply chain or expert informants. While not implemented livestock surveys is difficult, because costs differ by country. on a regular basis, these surveys provide critical information In general, the largest cost is that for personnel, which can ac- that complement or validate data from regular surveys, count for up to three-quarters or more of the total cost of the thereby contributing to better investment decisions and survey. Transport costs are second. Evaluating the benefits increased understanding of their impact on the ground. of the surveys is even more challenging, as this depends on QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations 40  |  Investing in the Livestock Sector: Why Good Numbers Matter the subsequent constructive use of the data which, alas, often ●● Sampling is a major issue when official livestock statis- remain largely unused. tics are generated from sample surveys, as the spatial distribution of animals is often not well-correlated with the distribution of the sampling units, namely rural AGRICULTURAL LIVESTOCK DATA households and/or farm holdings. This is particularly true COLLECTION SYSTEM AND LIVESTOCK in countries with relatively large tracts of arid/semi-arid INDICATORS areas. ●● The Livestock in Africa: Improving Data for Better Policies Table 2 summarizes the major livestock indicators that the Project undertook four online surveys on livestock data/ reviewed surveys can, on paper, help generate, starting with indicators that also targeted stakeholders’ perception of the core indicators needed by the Ministry responsible for the quantity and quality of livestock data. Data availabil- livestock and the National Statistical Authority. It offers six ity is often highlighted as a problem by international and major comments: national livestock stakeholders, not only because some ●● The prevailing system of agricultural/livestock data collec- indicators are seldom available or not accessible when tion, if functional, could on paper help generate the core needed, but also because most surveys target farm level livestock indicators, in addition to other indicators needed and consumption related issues, with little information for policy and investment purposes; on factors along the input and output value chains. The quality of data, usually ‘fitness for purpose’ amongst most ●● There is no single survey which, on its own, satisfies the National Statistics Office, includes various dimensions demand for livestock data, not even that for core livestock (e.g. relevance, accuracy, timeliness, accessibility and indicators. Data integration, therefore, is essential for interpretability) and qualitative categories (e.g. excellent, ensuring the generation of good quality core livestock good, adequate, poor and very poor), which are subject indicators. to personal interpretation. Again, stakeholders tend to not trust the quality of available livestock data: results of ●● Administrative records are the only data that are regularly a Global Survey (Pica-Ciamarra et al. 2012) on livestock collected and, therefore, they are critical to updating data and indicators indicates that over 41 percent of the the value of core indicators during in-between surveys. 641 respondents rate as poor or very poor the quality of Indeed, censuses are undertaken every five or ten years, available livestock indicators, with only 21 percent assess- and sample surveys are rarely done every year. In addi- ing them as good (Figure 2). tion, once collected, it takes at least one year before the data from these surveys are cleaned, processed and results produced and disseminated. ●● For the design of livestock sector policies and investments that aim at increasing livestock productivity while also contributing to poverty reduction and food security, data from both agricultural/livestock sample surveys and living standards measurement studies are needed: the former help appreciate constraints to livestock produc- tivity/profitability and the latter the role of livestock in the household economy, and hence the incentives and disincentives that underpin household’s livestock-related decisions. However, as said above, neither agricultural/ sample surveys nor living standard measurements studies ©FAO/Thomas Hug are regularly undertaken in sub-Saharan African countries and, when they are, the livestock sector is often unappre- ciated in the survey questionnaires. QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations PART I. DEMAND AND SUPPLY OF LIVESTOCK DATA: GAPS AND ISSUES   |  41 TABLE 2. DATA SOURCES FOR LIVESTOCK INDICATORS Living Survey Agricultural / Agricultural / Household Standards Administrative Livestock Livestock Budget Surveys Measur. Records Core indicator  Census Sample Surveys Studies Livestock Population *** ** No * ** Livestock production * *** No * ** Market prices * *** *** ** *** Outbreaks of animal diseases / animals no no No no *** affected / animals at risk Animal stock, beginning and * ** No ** *** end of reference period Livestock value added Production, quantity * *** No ** ** Input, prices no ** No * no Production, prices * ** No * *** Input, prices no ** No * no Imports / exports no no No no **               Living Survey Agricultural / Agricultural / Household Standards Administrative Livestock Livestock Budget Surveys Measur. Records Core indicator  Census Sample Surveys Studies Productivity-related indicators * *** No * * Profitability-related indicators no *** No * no Constraint-related indicators * *** No * no Livestock-livelihoods indicators no * No *** no *** very likely; ** likely; * possible QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations 42  |  Investing in the Livestock Sector: Why Good Numbers Matter FIGURE 2.  QUALITY OF LIVESTOCK DATA AS PERCEIVED BY STAKEHOLDERS 0.4 0.3 % OF RESPONDENTS 0.2 0.1 0 Very poor Poor Adequate Good Excellent Source: Pica-Ciamarra et al., 2012 CONCLUSIONS It is clear that a multitude of surveys regularly collect data on livestock and that, on paper, a functional agricultural/livestock statistical system could support the generation of the core livestock indicators and ©FAO/Thomas Hug some other key livestock policy/investment indicators. However, given that there is no single survey that fully responds to the information needs of major livestock stakeholders, the possibility of making effective invest- ments in the sector strongly depends on undertaking specialized surveys when policies and investments are designed and on the possibility of jointly using data from different surveys; in other words, on the possibil- ity of data integration. Currently stakeholders contend that their demand for information remains often unmet, including both the quantity and quality of available livestock data. This suggests the need for investments to improve the agri- cultural data collection system targeting livestock and/ or addressing livestock-specific data issues. Part II of this Sourcebook presents examples of methodologies that governments can apply/adapt to produce more and better quality livestock data. QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations PART II. METHODS TO IMPROVE THE QUANTITY AND QUALITY OF LIVESTOCK DATA   |  43 PART II. METHODS TO IMPROVE THE QUANTITY AND QUALITY OF LIVESTOCK DATA 2.1 COHERENT AND COMPREHENSIVE INFORMATION: DESIGNING A LIVESTOCK QUESTIONNAIRE FOR AGRICULTURAL AND INTEGRATED HOUSEHOLD SURVEYS KEY MESSAGES Neither agricultural nor living standards Challenges in developing a livestock measurement surveys are regularly undertaken questionnaire include the different objectives in sub-Saharan African countries. When they are of the National Statistical Authority and the implemented, the livestock sector is often under- Ministry responsible for livestock, the former appreciated in the survey. willing to keep the questionnaire as simple as possible and targeting few data items, while the latter aims to have it as detailed as possible, A standardized questionnaire including livestock targeting broad information on livestock. in agricultural and household surveys allows a better appreciation of the role of animals in the farm and household economy, which is a pre-condition for the effective design of sector policies and investments. INTRODUCTION Stakeholders contend that available agricultural data collection systems, as chapter 1.4 shows, are to a large extent insufficient to generate adequate livestock-related information, because of both a lack of and insufficient quality data. The most straightforward way to increase the available QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations 44  |  Investing in the Livestock Sector: Why Good Numbers Matter information on livestock is to ensure the adequate inclusion achieve the overarching goals of poverty reduction and food of livestock in the questionnaires of surveys, which are security, and other broad socio-economic goals. regularly undertaken by national governments, such as the agricultural census, the agricultural sample survey or the A review of a handful of both agricultural/sample survey living standards measurement study (LSMS). and multi-topic household survey questionnaires, however, reveals that livestock is, in most cases, inadequately repre- This chapter presents a set of livestock questions — so-called sented. For example: ‘livestock module’ — to be considered for inclusion in agricul- tural/livestock sample surveys and in multi-topic household ●● The 2008 Rwanda National Agricultural Survey includes surveys. The focus is on farm and multi-topic household sur- only a few livestock-related questions: the number of an- veys — and not on surveys targeting commercial enterprises imals by species; type of feed; farming methods, notably — as in most developing countries the largest share of ani- stabling or roaming; ownership of a cowshed; and then mals are kept by farm households or livestock keepers. Data information on sales of animals and home slaughtering from farm and multi-topic surveys, as Table 2 in chapter 1.4 (NISR, 2010); illustrates, can on paper generate almost all the livestock-re- ●● The 2010/11 Livestock Sample Survey of Ethiopia, one lated indicators needed by stakeholders, though they are to of the few countries in sub-Saharan Africa that regularly be complemented by data from other sources when policy undertakes agricultural sample surveys, includes ques- and investment plans are to be detailed (chapter 1.3). tions on animal population by breed, age and purpose for The next section provides the rationale for developing a keeping; on births, purchases, death and slaughters of livestock module for agricultural/livestock sample surveys animals; on livestock diseases, vaccination and treatment and multi-topic household surveys. A section that highlights over the reference period; on utilization of livestock feed; the salient features of the livestock module follows, including and on participation in a livestock extension program the approach used to develop it. Then lessons from the imple- (CSA, 2010); mentation of the module in multi-topic household surveys ●● The 2008 Livestock Survey in the Arid Land Districts of in Niger, Tanzania and Uganda are presented, followed by Kenya collected information on livestock numbers by spe- recommendations on how to apply and improve it. cies and, within species, by breed, age and sex; on changes in stock due to births, deaths, purchases, sales, social rea- LIVESTOCK IN AGRICULTURE sons (gifts), slaughter and theft; on production and sale of SURVEYS AND IN MULTI-TOPIC milk, ghee, honey and hides and skins (ALRMT, 2007); SURVEYS: A SNAPSHOT ●● The 2005 Ghana Living Standard Measurement Survey includes questions on livestock ownership by species, as Livestock keeping is a multi-functional activity in developing well as sales and purchases of live animals over the last countries: farm animals generate food and income, are a store months; questions on expenditure for raising livestock, of wealth and act as a safety net in times of crisis. They pro- including feed, veterinary services and drugs, hired labor vide draught power and hauling services, manure, fuel and and some other; revenue from selling milk and eggs; and building material; transform crop residues and food wastes in self-consumption of animal products (GSS, 2008); valuable protein and contribute to social capital (FAO, 2009). Rural households have thus a variety of incentives for keep- ●● The Malawi Integrated Household Survey 2010/11, which ing livestock and, indeed, data from 12 developing countries does have a specific focus on agriculture, includes ques- in Africa, Asia and Latin America show that between 46 to 85 tions on livestock ownership by species; change in stock percent of rural households keep farm animals, with a coun- over the past 12 months (purchases, sales, slaughter, try average of about 60 percent (FAO, 2009). Many of these given away as gift, etc); disease and vaccination; and total households are poor and, given the important role livestock expenditure on hired labor; feed, vaccines; veterinary ser- plays in their household economy and that many livestock vices and other; production of milk, meat, eggs, manure animals are not meeting their full productivity potential, it is and honey (NSO, 2010); anticipated that increases in livestock productivity can help QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations PART II. METHODS TO IMPROVE THE QUANTITY AND QUALITY OF LIVESTOCK DATA   |  45 ●● The 2010/11 Nigeria General Household Survey contains Overall, insights into the rationale for investing in livestock questions on animal holdings, including change in stock to reduce poverty, including identification of major produc- in the past 12 months due to births, sales, slaughter and tion-related constraints, are in many cases challenged by a other reasons; on major diseases affecting animals and lack of adequate information on the role and use of livestock vaccination; and a final question on the expenses incurred in the household/farm economy. for tending the entire herd, such as on hired labor; animal feed; maintenance of pens and stables; and commission on sale of animals and a few others (NBS, 2010). A LIVESTOCK MODULE FOR AGRICULTURAL AND MULTI-TOPIC In general: HOUSEHOLD SURVEYS ●● Available data in agricultural/livestock sample surveys and in integrated household surveys are sufficient to With the objective to assist decision makers in collecting generate descriptive statistics on livestock ownership; more comprehensive livestock-related information at house- sometimes on production and, occasionally, on inputs hold level, the FAO, the World Bank, the ILRI and AU-IBAR, with a focus on access to animal health services. Data in collaboration with national governments in Niger, Uganda from integrated household surveys do also allow classify- and Tanzania, developed a short, a standard and an expanded ing/grouping households according to some livelihoods version of a livestock module for multi-topic household sur- criterion (e.g. income level). veys and agricultural surveys. ●● However, data are rarely sufficient to provide a systematic The module was developed as follows. First, a variety picture of the livestock sector of the country because of multi-topic household survey questionnaires and of limited/missing information on husbandry practices, inputs and outputs, such as breeding practices; feed and water access; production and use of manure; the use of animals for hauling services and draught power; and oth- er. The implication is that the overall understanding of the livestock sector is patchy at best. ●● Data from both surveys do not provide a good under- standing of the determinants of livestock productivity, which involves some ratio between outputs and inputs. Even when information is asked about inputs, this targets mainly value (and not quantity), and in most cases is asked regarding the herd as a whole, i.e. it is not possible to attach inputs to the different animal species or to indi- vidual animals. ●● Data from integrated household surveys provides some ability to measure the contribution of livestock to house- hold livelihoods and to investigate the basic determinants of livestock ownership, such as family size, land owner- ship, level of education, level of income; etc. However, this data neither captures the non-monetary livestock services provided by livestock, such as manure, draft power and insurance, nor allows exploring the livestock-gender and ©FAO/Ami Vitale livestock-youth relations. QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations 46  |  Investing in the Livestock Sector: Why Good Numbers Matter agricultural/livestock survey questionnaires implemented in agricultural/livestock surveys and multi-topic household developing and transition countries were collected. Survey surveys. This expanded module consists of over 200 live- questionnaires are often included as appendices of statistical stock-related questions, which makes its inclusion in typical reports; are sometimes available on the website of the nation- agricultural and household surveys impossible. A standard al statistical office; and some are made publicly available by and a short version of the module were therefore developed, the International Household Survey Network. which national governments may easily adapt and include in their survey questionnaires. The three versions of the module Second, a production function approach was used to identify vary by size, but have four common, overarching goals: the information set needed to provide a satisfactory picture of the livestock sector. This involved systematizing all inputs ●● Generate basic statistics on key livestock-related variables, and outputs associated with animal keeping, such as feed, such as livestock ownership and access to animal health water, animal housing, animal health, animal slaughtering, services; milk production and marketing. ●● Measure the value of household’s livestock, which are an Third, working groups were formed around each component important economic asset; of the production function and tasked to identify a set of questions to possibly include in agricultural and integrated ●● Measure the cash and in-kind income from livestock; household surveys, using the collated questionnaires as a ●● Model household’s livestock husbandry and production starting point. No upper limit was set to the number of ques- practices. tions to propose, but the scope, content and typical length of agricultural/livestock and integrated household survey The module solicits information in three major domains: questionnaires were illustrated to group members. livestock ownership; livestock inputs, i.e. husbandry prac- tices; and livestock outputs. Processing is omitted (but for Finally, the questions proposed by the working groups for the one question) as it is a non-farm enterprise activity that is various segments of the production function were assembled typically addressed in other types of surveys. and made consistent to generate an expanded module for TABLE 3. C  ONTENT OF THE LIVESTOCK MODULE FOR AGRICULTURAL AND MULTI-TOPIC HOUSEHOLD SURVEYS Livestock domain Sections Remarks Questions are asked for individual animals, often differentiated by age, • Number of animals Livestock ownership gender and breeds (local/indigenous and improved/exotic), which • Change in stock in past 12 months helps to appreciate herd structure and inter-species composition. • Breeding • Feeding Questions are asked for major groups of animals (e.g. large ruminants, Inputs and husbandry practices • Watering small ruminants, pigs, poultry birds, equines, other), as management • Animal health practices usually do not differ between animals of the same species. • Housing • Meat production • Egg production Monetary and non-monetary Questions are asked for major groups of animals, including both the • Milk production outputs monetary and non-monetary value of production. • Animal power • Dung QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations PART II. METHODS TO IMPROVE THE QUANTITY AND QUALITY OF LIVESTOCK DATA   |  47 Short version sub-sections include questions on the use of family labor by gender, and on the non-family labor hired for raising animals. The short version of the module includes questions on live- stock ownership by species (e.g. cattle) and type of animals The standard version of the module supports generating de- within species (e.g. bulls, steers, cows, etc.), and a question scriptive statistics for key livestock-related variables, for which on the major purposes for keeping animals. It inquires about nationally representative indicators are often unavailable. sales of animals by species over the reference period, which is Examples include ownership of exotic breeds; prevailing breed- 12 months for large and medium animals (e.g. cattle, sheep ing practices; and access to veterinary services. It also allows and goats) and three months for small animals, namely short quantifying with accuracy not only a household’s livestock cycle animals (e.g. chicken, ducks and rabbits). It includes wealth, but also the contribution of livestock to household some questions on meat, milk and egg production, and one livelihoods, including both their monetary and non-monetary only question on husbandry practices. The latter targets ani- value. In addition, depending on the sample size and the spe- mal vaccination which, in most countries, is provided for free cies at hand, it can be used to estimate production functions or subsidized by the public sector. using the animals as unit of observation, particularly when it is included in specialized livestock surveys. The standard The short version of the module allows quantifying with version of the module comprises about 95 questions. some accuracy a household’s livestock wealth, and hence classifying households into different types; it also provides a Expanded version rough measure of the cash income derived from livestock. It does not provide a comprehensive picture of husbandry and The expanded version of the livestock module includes all the production practices. This version comprises about 30 ques- questions in the standard version, plus additional informa- tions and is intended for use in surveys for which livestock is tion in all sub-sections. In particular, it allows differentiating a minor interest. between animal ownership and animal keeping, as not all households owning livestock raise them on the farm; it in- Standard version cludes questions on the providers of goods and services, such as the public and private sector, and NGOs; it asks details The standard version of the module collects a large amount about the role of family members in selling animals and live- of livestock-related information, including ownership of ani- stock products, including who controls the earnings. mals, inputs and husbandry practices, and livestock outputs by product, by-product and service, such as milk, manure and The expanded version of the module allows generating key draft power. As in the short version, questions on livestock livestock statistics and undertaking analyses as with data ownership target species and types of animals; while all other from the standard version, but with higher accuracy. It’s a questions only inquire about animal species, such as large long and heavy version and, as such, it should be seen as a ruminants, small ruminants and equines. rotational module that country governments implement only when they need comprehensive and detailed information Questions on change in animal stock over the reference on livestock, most likely for a specific sub-sample of the period collect information on the causes of herd reduction/ population (e.g. the cattle keepers). In response to specific expansion, including purchases, sales, slaughters, gifts information needs, however, survey designers may wish to and loss of animals for different reasons (e.g. death due to include only one or selected sub-sections of the expanded disease; theft; etc.). Questions on inputs and husbandry version of the module in their survey questionnaires, such as practices target housing and breeding practices; access to and those on breeding and animal health. use of water and forage/feed; and animal health, including vaccination, deworming and treatment of sick animals. Finally, questions on outputs inquire not only about meat, “The expanded version of the livestock milk and egg production, but also about the use of animal module includes all the questions in power (draft and transport services) and the production of dung, mainly but not only, used as manure. Most the standard version, plus additional information in all sub-sections.” QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations ©FAO/Giulio Napolitano 48  |  Investing in the Livestock Sector: Why Good Numbers Matter IMPLEMENTING THE LIVESTOCK ●● While the Ministry responsible for livestock prefers to include as many questions as possible in survey ques- MODULE: LESSONS tionnaires, the Statistical Authority prefers keeping the livestock module as short as possible, for at least three The three versions of the livestock module for agricultural reasons. The first is savings: not only does a longer live- and multi-topic household surveys are starting points for stock module involve more costs, but it could also give developing questionnaires that fit the needs of the country. non-livestock stakeholders arguments for expanding Survey designers are expected to build their own module that other sections of the questionnaire, such as those on adapts to the country livestock sector, including its structural health or education. The second is a statistical reason: and transitory features. agricultural/livestock and integrated household survey Three sub-Saharan African countries so far have used the questionnaires are administered to a relatively small sam- livestock module to improve the livestock content of their ple of households, and detailed questions are sometimes multi-topic survey questionnaires, including Niger (Enquête answered by just a few households, which make the col- Nationale sur Les Conditions de Vie des Ménages 2011/12), lected data insufficient for any robust statistical analysis. Tanzania (National Panel Survey 2011/12) and Uganda For example, a question on the sale of dung cakes would (National Panel Survey 2011/12). Some lessons drawn out of make little sense in the context of multi-topic household questionnaire design and administration and from a descrip- surveys. Third, Statistical Authorities analyze — because tive analysis of the Niger data are as follows: of their specific mandate — only part of the collated data: for example, they have little interest in studying the QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations PART II. METHODS TO IMPROVE THE QUANTITY AND QUALITY OF LIVESTOCK DATA   |  49 preferred outlet markets used by farmers or in exploring details about inputs and outputs, as the sub-sample of pig the correlation between household size and structure producers is not large enough to generate data for robust and herd size and composition. In addition, they are well descriptive statistics or causal analysis. aware that there are few other actors in the country capa- ble of analyzing the data. Indeed, there are several surveys ●● Animal health/disease information is critical for country for which most of the data remain unutilized, a net waste governments, particularly that pertaining to trans-bound- of public resources. ary and zoonotic diseases. Following a standard approach, the module suggests asking direct questions about animal ●● The Ministry responsible for livestock has three argu- diseases, such as brucellosis, ovine rinderpest (Peste ments for advocating the adequate inclusion of livestock des petits ruminants) and Newcastle disease in poultry. in multi-topic household surveys. The first is based on However, not all farmers are fully aware of the types data showing that, as is the case in most developing of diseases that affect their animals. Complementary countries, the majority of rural households keep some information, such as from veterinary officers, could thus farm animals and that livestock contribute over one third be gathered while analyzing the animal health section of to the value added of agriculture. The implications are that the module. Alternative options to collect animal health it is important to ask questions on livestock, as these are information also could be designed and tested. One likely to be answered by the majority of households; and possibility is to use a syndromic approach, which implies that a crop-focused questionnaire would be largely unable asking syndrome-related questions on the basis of clinical to properly appreciate the livelihoods of rural households. features (e.g. neurological, respiratory, dermatological The second argument is that, even though some questions and diarrheal syndromes); the collated data should be might be of little statistical relevance, these are poten- interpreted jointly with local animal health authorities. A tially important for decision makers because they provide second possibility is to include animal disease questions critical policy information, such as data on the proportion in both the household and community questionnaire of of households with exotic breeds of animals. Finally, the the multi-topic surveys, along the lines of participatory Ministry responsible for livestock must show a commit- epidemiology. ment to collaborating with the Statistical Authority to examine the livestock content of the surveys. It should ●● Measuring labor has been found to be particularly chal- be noted that, in almost all developing countries, staff in lenging for two reasons. First, in many circumstances, the Ministry responsible for livestock are not equipped with the possible exception of milking, the labor force to analyze the data collected through household surveys; performs the same task (e.g. taking animals to graze) however, they are the most important users of the data. simultaneously for all animals in the herd, and in partic- ular for large and small ruminants (e.g. cattle and sheep). ●● While implementing the livestock module, survey de- Second, watering and feeding animals are often joint signers should adjust the suggested list of animals in the activities, with livestock taken to pastures where water module, which is comprehensive, to be consistent with sources are available. The implication is that attaching the prevailing livestock production systems. This could labor to a specific task or an individual animal is difficult, be done at three levels. First, some animals are simply thereby making it challenging to measure labor produc- not present in a given country, such as yaks in Uganda, tivity. The module presents one way to address this issue: and should not be included in the survey questionnaire. by first asking whether animals of different species are Second, while the module allows separating local/indig- fed and watered jointly; and then asking questions on enous from improved/exotic breeds, in many countries the time allocated to feed/water animals by family and the diffusion of the latter is so minimal that it may make non-family labor. Other options could be designed and sense to only differentiate animals by breed in the section tested. on animal ownership. In the same vein, there are animals that are not widely held by households, such as pigs in ●● When collecting information on livestock production, Niger. Again, in these circumstances, it makes more sense the module proposes an approach which differs from the to collect minimal information on ownership of pigs in one typically used in multi-topic household and agricul- order to generate some basic statistics, but not to ask tural surveys. In particular, rather than directly asking QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations 50  |  Investing in the Livestock Sector: Why Good Numbers Matter information on meat, milk and egg production, the mod- ule asks a sequence of questions that link animals with production levels. This helps the interviewee to provide accurate information on production levels and to arrive at some measure of partial productivity (e.g. eggs per hen over the reference period). For milk, for instance, ques- ©FAO/Giulio Napolitano tions are included about the number of milked animals over a reference period; the number of months during which the animals were milked; whether suckling was allowed when the animals were milked; and the average quantity of milk produced per day during the milking period. Similar series of questions are suggested to obtain meat and egg production information. module and priority areas for improvement. In any case, the Niger, Uganda and Tanzania surveys represent the most The above are the major lessons emerging from the adminis- comprehensive household-level livestock datasets available tration of the livestock module in the multi-topic household in sub-Saharan Africa, thus facilitating the analysis and doc- surveys of Niger, Uganda and Tanzania. Additional insights umentation of the many connections between livestock and on strengths and weaknesses of the module will become livelihoods. The forthcoming insights from these surveys are clear as the country data for Uganda and Tanzania is ana- expected to significantly enhance our understanding of the lyzed. The analysis will highlight possible weaknesses in the role of livestock in the household economy. CONCLUSIONS including Niger, for the Enquête Nationale sur Les Conditions de Vie des Ménages 2011/12, Uganda, for the National Panel Survey 2011/12, and Tanzania, for the Traditional agricultural/livestock sample surveys and National Panel Survey 2010/11. multi-topic household surveys inadequately represent livestock, despite the fact that livestock are a widely Lessons drawn from the design and administration of the owned asset among rural households in developing coun- survey questionnaires indicate that, unless the Ministry tries, including the less well-off. This challenges the design responsible of livestock is aware of the content and and implementation of equitable and efficient interven- scope of the survey questionnaire and commits itself to tions in the sector. analyzing the produced data, the Statistical Authority will prefer avoiding expanding the livestock section of any This chapter presented a short, a standard and an expand- survey. As to the implementation of the module, at least ed version of a livestock module for agricultural surveys in the context of multi-topic household surveys, the major and for multi-topic household surveys. The three versions challenges relate to measuring labor and animal health/ of the module, with different level of details, aim at col- diseases. These represent areas for further research. lecting data to generate statistics on key livestock-related variables; measuring the value of a household’s livestock; The short, standard and expanded versions of the live- measuring cash and in-kind income from livestock; and stock module for multi-topic household surveys and the understanding and modeling the household’s livestock survey questionnaires for Niger, Tanzania and Uganda are husbandry and production practices. available to download from the websites of the FAO-WB- ILRI-AU-IBAR Livestock in Africa: Improving Data for Better The three versions of the livestock module are starting Policies Project and the World Bank LSMS-ISA Project. The points for developing country modules that fit the needs data from the livestock module implemented in Niger, of the country at hand. Three sub-Saharan African coun- Tanzania and Uganda are also freely available for down- tries have so far used the module to improve the livestock load and use. content of their multi-topic survey questionnaires, QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations PART II. METHODS TO IMPROVE THE QUANTITY AND QUALITY OF LIVESTOCK DATA   |  51 2.2 IMPROVING LIVESTOCK DATA QUALITY: EXPERIMENTS FOR BETTER SURVEY QUESTIONNAIRES KEY MESSAGES INTRODUCTION Asking questions that generate accurate livestock The design of a livestock survey is not necessarily straightfor- data — on animal diseases, labor inputs and ward, due to the complexity in the production and marketing milk production — is sometimes challenging, as processes, in the management of livestock assets, and in the farmers might have imprecise information on lifestyle of some population groups that are especially reliant those and other variables. on livestock for their livelihoods (e.g. nomadic, semi-nomad- ic, or transhumant livestock keepers). All of these factors pose particular challenges to data collection. Randomized experiments, by which different questions targeting the same information are When designing survey questionnaires, therefore, decision asked to farmers, are an effective method for makers should take into account both livestock-specific and identifying the best way to formulate specific system-specific characteristics. However, in most cases, questions and improve survey questionnaires practitioners who are tasked designing a new survey often have little to rely on other than their own technical expertise, content. experience and common sense. Moreover, the lack of a sys- tematic approach to survey design often results in less than Transparent dialogue and collaboration with optional survey questionnaires, and hence in the generation livestock stakeholders is necessary to effectively of inaccurate data. formulate livestock survey questionnaires, This chapter proposes that there is much to be gained by particularly those targeting sub-segments of the developing, adopting and disseminating good practices for population, such as pastoralists. survey construction which facilitates the systematic assess- ment of the choices made in questionnaire design and feeds into an understanding of how those choices influence the quality of the data collected. Drawing on survey experiments in Niger and Tanzania focused on milk production and pasto- “When designing ralist livelihoods respectively, this chapter sketches possible practical approaches to conducting various types of survey survey questionnaires, validation exercises. decision makers should take into account both PRE-TESTING: livestock-specific DO AS WE SAY, NOT AS WE DO and system-specific In their guidelines on methods for testing and evaluating characteristics.” survey questions, Presser et al. (2004a, p. 109) note that “pre- testing’s universally acknowledged importance has been honoured more in the breach than in the practice.” Even in countries with well-managed and financed statistical systems, pretesting is often limited to a dry run of survey interviews, usually tar- geting a fairly limited number of households, which are then qualitatively evaluated by the survey teams so as to draw QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations 52  |  Investing in the Livestock Sector: Why Good Numbers Matter lessons from questions that seem to pose problems to inter- is a randomized ‘experiment’ in which randomly selected viewers or respondents. Sometimes this is complemented sub-samples were asked alternative sets of questions aimed by a quantitative analysis of response frequencies and other at capturing household milk production. The other is a more simple statistics from the data collected during a pilot. qualitative, but systematic and documented, pilot test of a questionnaire on pastoral households in Northern Tanzania. Often there is little that is systematic about these tests, despite the use of techniques which assess the performance It is important to note that the decision on the empirical of survey instruments (see e.g. those reviewed in Presser et approach to take is a function of the type of research ob- al., 2004b, and Iarossi, 2006). This is aggravated by a lack of jectives and the underlining questions being asked in each documentation on the process and results of such tests. The exercise. For reasons that will become clearer in the discus- evaluation of what ‘works’ is mostly left to the judgment and sion that follows, randomized experiments can be useful to experience of the survey team. compare ‘discrete’ approaches, less so to fine tune a draft questionnaire where there are several interrelated and maybe Increasingly, however, survey practitioners are paying atten- far-reaching design questions that need to be pinned down. tion to pre-tests as a means of improving data quality. Also, specific methods are being developed, tested and codified and increasingly applied in survey practice. The interested reader is referred to Presser et al. (2004b) for a review of methods such as cognitive interviews, behavior coding, response laten- cy, vignette analysis, experiments, and statistical modeling. While the use of such methods, and their documentation, is more commonly found in OECD country surveys, their appli- cation is being adopted in low-income countries, including in Africa. A literature is slowly emerging, which includes tests of consumption expenditure data (Joliffe, 2001; Beegle et al., 2012), recall methods in agricultural surveys (Beegle et al., 2011), agricultural production diaries (Deininger et al., 2012), child labor (Dillon et al., 2012), labor statistics (Bardasi et al., 2010), and micro-enterprise profits (de Mel et al., 2009). Within the livestock sector, numerous areas have been highlighted as particularly challenging for survey design. In consultations with livestock and household survey experts, the two specific topics which were cited as particularly problematic were the collection of data which feed into calculations of milk production, and the collection of data on mobile (pastoral) households/herders. This chapter reviews experiments in livestock questionnaire elaboration within the context of household surveys in specif- ic African countries, namely Tanzania and Niger. The process of conceptualization, design, implementation and analysis of these exercises is described for survey practitioners interested in potentially employing similar approaches to the pre-tests of new livestock-related questionnaires. The ©FAO/Ami Vitale methods employed in these two examples represent distinct ends of the spectrum of possible approaches. The one tar- geting improved survey data on milk production in Niger QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations PART II. METHODS TO IMPROVE THE QUANTITY AND QUALITY OF LIVESTOCK DATA   |  53 RANDOMIZED EXPERIMENTS: It is beyond the scope of nationally representative household surveys, in terms of both objective and logistics, to collect MILK PRODUCTION IN NIGER milk production data over extensive time periods, or in a way that allows calculating the complex milk productivity Nationally representative household surveys typically lump parameters often required by livestock sector specialists. The the data collected on livestock products into one table listing objective of a nationally representative household survey is the different products on the rows and a set of standard more modest, and limited to collecting a reliable measure of questions, common to all products and based on a 12-month milk production that can accurately portray the role that milk recall period, in the columns. The module usually asks a production has in the overall household livelihood strategy. variation on two rather simple questions: (1) “Number of production months in the last 12 months”, and (2) At the same time, surveys aim to look at the heterogeneity “Average production per month during production months.” across households. This implies that methods that rely on Sometimes these questions are asked for milk as a homo- the application of technical production factors from the lit- geneous product, sometimes the product is broken down in erature (e.g. average milk production per animal in a certain different types of milk (cow, sheep, goat). environment) combined with variables that may be easier to measure in a survey (such as the number of animals milked Because of the peculiarities of milk production1, it is a by the household) may result in accurate ‘average’ estimates, well-known fact among livestock experts and statistical prac- but may artificially reduce the observed differences in milk titioners that collecting reliable milk production data with production (both in physical and value terms) across house- such simple recall questions is likely subject to errors. This holds. For most of the analyses performed with household has led livestock researchers and livestock survey specialists level data, the analysis of the dispersion of the distribution to devise more complex strategies to generate more accurate is often as important, if not more so, than the analysis of milk production data and additional information useful to the measures of central tendency (means, medians). For evaluate milk production systems. these reasons, alternative data collection methods need to be Examples of these alternative approaches include the 12_mo evaluated, not only on the basis of their ability to yield an ac- method developed by researchers in CIRAD (see Lesnoff curate point estimate of, say, mean milk production, but also et al., 2010) which relies on the monitoring/recording of on their ability to return a distribution of observations that production over extended periods of time. To increase the resembles as much as possible the ‘true’ distribution. accuracy of the responses, techniques are introduced that, In view of these considerations, an experiment was imple- while based on recall approaches, prompt more in-depth mented in Niger which reviewed and compared two methods information from the respondent about the milk production that are often applied in livestock sector surveys. These two system. In developing new survey approaches to integrate methods, supported by different questionnaires, are referred into household surveys that include an expanded agricultural to as the “Average milk per day” (AMD) and the “Lactation focus, these methods are useful, but need to be adapted curve” (LC) methods. Both seem to hold the promise of being to conform to both the objective of the survey and to the adaptable to both the questionnaire design and logistics of a survey operations. The only way to assess whether a change nationally representative multi-topic household survey. in approach results in an actual improvement in data quality is to validate the new method via fieldwork, ideally in an The two questionnaires are amenable to testing in an exper- experimental setting, while reproducing as closely as possible imental setting because they represent a discrete change in real survey conditions. survey design. In a broad sense, they are virtually identical, except for questions related to milk production. Both ques- tionnaires start off by prompting the respondents about 1 There are a host of features of milk production for human consumption the number of months during which animals were milked that make recall particularly hard: Milk is produced continuously, but with seasonal patterns. The lactating capacity of animals varies over time, across for human consumption, and how many animals, by animal animals, and is dependent on the management of the animals. The farmer type (bovines, sheep, goats, camels), were milked on average may additionally decide not to collect milk independently of the production during each of those months. capacity of the animals, and often part of the milk is used for suckling offspring. QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations 54  |  Investing in the Livestock Sector: Why Good Numbers Matter The questionnaires themselves differed in that the AMD production while also displaying a low correlation coefficient asked for the average quantity per day produced by each with the monitoring variable (r=0.38). Shortening the recall milked animal during the period, whereas the LC question- period to six months appeared to result in the most accurate naire asked about the amount of milk produced by each estimate (about 3 percent difference in mean value compared animal at three (four) different points in time: one week, one to 5 to 6 percent with the 12 month recall). The six-month month, and three (and six) months after parturition, e.g. recall also showed the highest correlation to the benchmark after reproducing. The two modules then continue asking at 0.71. When using a 12 month reference period for the the same set of questions on issues of whether calves/lambs/ AMD method, it appears that also including questions on the kids were allowed to suckle, about the time gap between level of production at different points in the lactation can aid parturitions, and about the disposition of milk production recall, resulting in a marginal difference in mean values, but (sales, consumption, and transformation into dairy prod- in a substantial improvement in the correlation coefficient ucts). Annual milk production can be calculated from both (from 0.44 to 0.61). questionnaires. In the AMD, this involves simply multiplying the average daily production by 30 days (to get to monthly The experiment therefore revealed a clear ranking of methods production per animal), then by the number of months of in terms of their accuracy, and a clear idea of the extent to milk production. Using the LC method, the calculation is which the range and distribution of the estimates produced more complicated with annual production derived as the area with each of the survey methods deviates from the bench- under each animal’s lactation curve, or the milk production mark value of choice. curve. One challenge in assessing data quality is that of identifying FIGURE 3. M  EASURING MILK PRODUCTION a benchmark, or a ‘gold standard’ against which the survey IN NIGER: BOX PLOTS COMPARING measures can be compared to assess their accuracy. In the RANDOMIZED RECALL METHODS experiment in Niger, such a gold standard was constructed by AGAINST PHYSICAL MONITORING performing a physical monitoring of actual milk production 1000 every other week for 12 months, using a sample of around 900 300 households. The same households were then interviewed 800 using the two recall methods. The comparison yielded inter- 700 esting insights into the relative performance of the candidate 600 recall methods. Statistical analyses were later used to analyze 500 not only the relative performance of the alternative recall 400 methods, but also, and perhaps more importantly, to review how measurement error (or the deviation from the bench- 300 mark) varied by household and respondent characteristics, as 200 well as with specific variables of interest (e.g. does measure- 100 ment error increase or decrease with larger herd size, or with 0 g respondent’s education?). rin D ) o. all ed ule ito (LC AM 6-M rec lliz n od Mo rve ua Cm o. — nn Cu 6-M nL ng l, a ct. In the case of the Niger milk production example, a compar- Di ori La cal AM nit Re Mo o. ison was drawn between four competing recall methods: the 6-M AMD and LC methods over 12 months; the AMD, but based on a combination with the LC questions; and the AMD, but based on a shorter recall2. The results allowed for ranking of the methods, based on their variance from the results of the monitoring. The AMD recall performed better, in all its vari- ants, than the LC method, which appeared to underestimate 2 The results are discussed in full in Zezza et al. (2013). QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations PART II. METHODS TO IMPROVE THE QUANTITY AND QUALITY OF LIVESTOCK DATA   |  55 FIGURE 4.  MILK PRODUCTION DATA EXPERIMENT: SYSTEMATIC PILOTS: PASTORAL COMPARING 6-MONTH RECALL HOUSEHOLDS IN ARUSHA, DISTRIBUTION TO LACTATION CURVE TANZANIA METHOD. 1.20 The above example highlights the complexity of survey de- Correlation coefficient OLS coefficient sign and lends itself to examining other challenges which are 1.00 potentially more complicated and require different methods. 0.80 Broader information needs often are required which cannot be generated by simply adjusting the survey design through 0.60 refining how one specific (albeit crucial) piece of information is collected. 0.40 A critical example facing the African livestock sector is ensur- 0.20 ing inclusion of special populations such as mobile herders (nomadic, semi-nomadic, transhumant) which are often 0.00 6-Mo. recall, AMD in LC AMD Lact. Curve 6-Mo. recall not captured in national household surveys because of the annuallized module (LC) CORRELATION problems posed with integrating them in the sample, and of CORRELATION WITH 12 MONTHS MONITORING WITH 6 MONTHS finding them in a specific location at the time of the survey. MONITORING The little data that exist on pastoralists is therefore usually the product of surveys geared specifically at surveying those populations or communities, which most likely invalidates any direct comparison with the population at large. ©FAO/Giulio Napolitano QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations 56  |  Investing in the Livestock Sector: Why Good Numbers Matter BOX 5. ISSUES IN MEASURING PASTORAL ECONOMIES L ack of panel data on pastoral production systems thwarts the possibilities of formulating investments which promote an efficient use of resources available in arid and Second, pastoralists’ regular or opportunistic movements during the year makes it difficult to set up a system of standard data collection. Trekking routes may change from semi-arid lands, including livestock. Whereas several studies year to year (nomads may even change animal movements have documented pastoralist production systems and pas- after being informed of survey operations) and counting all toralist livelihoods in detail, the tools these studies use are animals that pass along a route is difficult; aerial or satellite time- and cost-intensive and not appropriate for monitoring surveys are powerful instruments to measure livestock pop- trends in the pastoral economy on a regular basis. More ulations in vast arid and semi-arid areas, but they produce practical ways need to be developed if Statistical Authorities little information on the pastoral economy, i.e. on their own are to collect, process and disseminate data and statistics on they are an ineffective tool for designing programs and in- pastoral production systems. vestments. Water points, which have been used as sampling units in some countries (e.g. Southern Ethiopia and Iran), There are at least three key issues associated with measur- are often unknown to statistical authorities and also pres- ing pastoral economies. First, there is no standard definition ent high seasonal variability, both in numbers and capacity of pastoralism, which may be identified on the basis of of watering livestock, i.e. livestock data collected at water economic parameters (how much does livestock contribute points may produce highly variable results across the years. to household income?), agro-ecological parameters (where is the household situated?), ethnic dimensions (to what tribe The third issue relates to data interpretation focused on does the household belong ?), by exclusion (e.g. by defining pastoral people which prioritizes investment options consis- crop and mixed crop-livestock farmers) or by combination tent with their livelihood system. Given the multiple roles of of more than one variable. Each of the different approaches livestock in pastoral economies, and the oftentimes oppor- has its own advantages and weaknesses: for instance, using tunistic use of markets by pastoral peoples, using standard an economic definition could produce high variability in the production or profit functions to identify key constraints number of pastoralists across the years because of rapidly affecting their livelihoods may lead to biased conclusions changing livelihood strategies associated in response to and policy indications. • weather fluctuations. As noted by Presser et al. (2004a: p. 122) pre-tests are espe- existing questionnaires from both sedentary and pastoral cially lacking for special populations, which is where they are livestock and other living standard surveys, and putting most needed given the special difficulties posed in surveying together an entirely new questionnaire to be tested and these populations. Survey challenges linked to pastoral validated before it can be applied on a larger scale. While it households include two broad classes of difficulties: (1) cap- may not be possible to identify a ‘gold standard’ for compar- turing them in the sample, and (2) asking the right questions. ison, one can, however, attempt to develop new sections of a survey instrument to address key questions for analysis, The experiment summarized in the following section focuses systematically pilot them in the field, and document the dif- on the latter: assuming access to pastoral households, what ficulties, successes and failures. Consolidating, collating and are the priority questions? Given that the livestock man- disseminating this learning can contribute towards estab- agement practices practiced by pastoralists (as well as many lishing a body of knowledge that will incrementally improve other challenges to their livelihoods) are profoundly different survey design efforts. The objective should not be that of from those of sedentary livestock keepers (and households in arriving at a blue-print, off-the shelf type of questionnaire, general) relevant information cannot be extracted by asking but rather to offer a starting point for other practitioners them the same set of questions posed to other households. to adapt to the specific features, goals and circumstances of each survey. Developing a pastoral specific questionnaire therefore re- quires carefully thinking about the key questions, adapting QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations PART II. METHODS TO IMPROVE THE QUANTITY AND QUALITY OF LIVESTOCK DATA   |  57 In the Arusha region of Tanzania, an exercise was conducted socio-economic characteristics. Comprehensive results are to adapt key sections of the Tanzania National Panel Survey documented in a detailed report (Loos and Zezza, 2013). (NPS) questionnaire for use with pastoral populations (Maasai communities, in this case). An initial draft module This systematic piloting of the new survey instrument pro- was developed which started from the NPS questionnaire vided some clear indications of the specific traits of pastoral and was then adapted to address key features which appeared livelihoods in Northern Tanzania that may be more amenable not to work well with pastoral Maasai communities. The to inclusion in a national survey like the NPS while also new questionnaire had a modified household roster which revealing those that may not, or that would require consid- attempted to capture the complex organization of the Maasai erable extra effort. Adjusting the household roster to reflect household which was not adequately represented by a ques- the complex structure of Maasai households, for instance, tionnaire built around a nuclear family. It also included a set appears doable, and may have important implications for of questions which related livestock ownership to the specific the analysis of livestock management. Table 4 shows the sub-households, questions on household and livestock mobil- implications of using the Maasai definition of household ity, sedentarization, grazing practices, and conditions which (the “olmarei” in Maa language) versus one based on the are not relevant to sedentary livestock keepers in Tanzania nuclear family definition implied by the standard household but are fundamental to interpreting the challenges to Maasai definition used by the National Bureau of Statistics (NBS) livelihoods. in their National Panel Survey (NPS). (The latter would be identified by Maasai respondents mostly as a sub-household, While conducting fieldwork, the field team iteratively revised referred to by its Kiswahili term, “kaya”). Because of the way the questionnaire, documenting the underlining rational livestock are assigned to different households members, motivating the changes, and providing an account of how and across sub-households, the key descriptive for the same the questionnaires performed in the interviews. This was sample would change dramatically. This would clearly have combined with a quantitative analysis of the data collected implications for any analysis of livestock management, in from about 200 households located in different commu- particular those related to animal movement, because of the nities with a wide range of underlying agro-ecological and way livestock is distributed across sub-households, as well TABLE 4. TANZANIA: SUMMARY STATISTICS USING DIFFERENT HOUSEHOLD DEFINITIONS Self-defined Olmarei NPS definition Kaya % difference Number of households 200.00 372.00 86.00 Household size 9.50 5.50 -42.00 Dependency ratio 1.31 1.18 -9.90 Female headed HH (%) 1.50 3.80 153.00 Age of head of household(years) 46.20 48.40 4.80 Head attended school (%) 28.00 23.70 -15.40 / household 99.20 53.30 -46.30 Animals /capita 10.43 9.71 -6.00 / household 23.33 12.54 -46.20 TLU /capita 2.45 2.29 -6.50 Source: Loos and Zezza, 2013 QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations 58  |  Investing in the Livestock Sector: Why Good Numbers Matter as per any per capita measure of welfare (because of the way households that cannot be contacted, in particular by under- household size needs to be computed to take into account the standing the expected timing of mobility so as to identify a different eating and sleeping arrangements prevalent among suitable time for the second visit. This pilot has shown that the Maasai). it is possible to gather useful information for the analysis of pastoral livelihoods in a complex household survey, such as Gathering basic information on the extent and timing of integrated household surveys. While it would have been quite mobility, and on the state of grazing areas also seems possi- challenging for the NPS operations to undertake such a pilot ble. Identifying the specific grazing areas used may be more targeting such a relatively small population, the independent challenging, although this may be feasible where community undertaking of the survey and the documentation and land use maps have been developed. Asking households in sharing of results with in-country stakeholders will increase different communities about the extent, duration and mo- the likelihood that the Statistical Authority will afford more bility of households and livestock, responses were obtained specific attention to pastoral populations in future national that seemed to tally with the qualitative perceptions. This ap- surveys. Without such a focus, national level data will miss an proach seems better able to capture the heterogeneity across opportunity to discuss policy options for the development of households and communities (see Figure 5 for a graphic pastoral communities. depiction of the responses). A critical challenge to overall survey design is to ensure that all households can be found at the time of the sur- vey. Surveys organized in two visits during a 12 month CONCLUSIONS period may be more successful in reducing the number of Surveys are conducted routinely on a wide range of topics in countries around the world. The amount FIGURE 5.  TANZANIA: PERCENTAGE OF of learning that is accumulated from each survey HOUSEHOLDS PRACTICING performed is arguably much less than what it could be. TRANSHUMANCE OVER THE PAST 15 Pressed for time, resources and results, survey practi- MONTHS BY DISTRICT tioners often draw on their own experiences, or those 40 of their associates, as the main source of guidance. Longido Kiteto Total A systematic approach to learning, as presented in 35 this chapter, can contribute to improving the quality % OF HOUSEHOLDS PRACTICING TRANSHUMANCE of the data that are generated by household surveys, 30 and transform the learning process whereby best practices are adopted by others. This avoids reinvent- 25 ing the wheel every time a new survey is designed. Documentation and dissemination of lessons learned 20 are crucial in that respect. 15 Targeted efforts at experimentation and documenta- tion of innovative survey designs can have a positive 10 impact not only on the quality of the data being pro- duced, but also in the confidence that data users have 5 in those data. While expert judgment and experience will continue to be an important input into designing 0 surveys, a range of methods, drawn from experimental designs to systematic pilots, can feed into improved 11-Sep 11-Dec Jan Feb Mar Apr May Aug Jun Jul Sep 11-Oct 11-Nov survey practices, generate better quality data, and Source: Loos and Zezza (2013) contribute to innovative learning processes. QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations PART II. METHODS TO IMPROVE THE QUANTITY AND QUALITY OF LIVESTOCK DATA   |  59 2.3 PHYSICAL MEASURES OF PRODUCTION FOR BETTER STATISTICS: THE LIVESTOCK TECHNICAL CONVERSION FACTORS KEY MESSAGES of a set of factors of production (e.g. animal stock, feed, wa- ter, etc.), and gives a single overall measure of productivity. Face-to-face interviews are often unsuitable for Total factor productivity is calculated using indices of outputs obtaining accurate data on the production level. and inputs (e.g. the weighted sum) or by some econometric technique that links output(s) to a set of inputs. Both partial and total livestock productivity measures are either based on Physically measuring at the farm level and the physical quantities of inputs and outputs (primal mea- in abattoirs/slaughterhouses is necessary sures of productivity) or on price, profit and cost information for properly quantifying production levels in (dual measures of productivity) (Chambers, 1988; Nin et al., traditional livestock production systems. 2007). The quality of any livestock productivity measure strongly de- Unless production levels are physically measured pends on the quality of the data available to measure inputs at regular year intervals, official statistics on and outputs. Data quality is typically high in research insti- livestock risk being biased. tutions or stations mandated to undertake scientific studies. It is relatively good when ad hoc data collection activities are undertaken for some investment purpose, such as for Methods to physically measure production level implementing a time-bound project in a given geographical at farm level and in abattoirs/slaughterhouses area. It is less good, and often poor, when nationally repre- are relatively straightforward, though they might sentative livestock statistics or indicators are to be generated: be expensive. limited financial and human resources devoted to data collection; limited focus on livestock in most surveys, i.e. lack of livestock data; sampling errors; non-sampling errors (e.g. improper survey livestock question formulation); and low frequency of livestock data collection, all make it difficult to INTRODUCTION generate good quality livestock productivity measures. The consequences of not correctly measuring livestock Increases in agricultural productivity, including in livestock, productivity in nationally representative statistics can be are essential for economic growth and poverty reduction serious. First, the Ministry responsible for livestock devel- in much of the developing world. Measuring livestock pro- opment will not be able to fully assess the returns to sector ductivity, and understanding its determinants, is therefore policies, including investments on the ground, which could critical to design and making investments that maximize the lead to a biased allocation of ministerial resources. Second, contribution of livestock to socio-economic development. livestock value added or the contribution of livestock to Livestock productivity connects inputs to outputs. Partial the Gross Domestic Product is unappreciated, which again livestock productivity is the amount of output produced by could result in a less-than-optimal allocation of government one unit of a given production factor over a reference period, resources. e.g. labor productivity could be calculated as liters of milk This chapter presents some methodologies for improving produced/hours of labor devoted to milking per cow per day; livestock productivity indicators at country level. The focus feed productivity could be computed as kg weight gain/kg is on the enumerator of all productivity measures, i.e. on the of dry matter fed to the animal over a stated period of time. level of production, and in particular on parameters used Total factor or multi-factor livestock productivity measures to calculate so-called livestock technical conversion factors, output(s) (e.g. milk, manure, transport services; etc.) per unit QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations 60  |  Investing in the Livestock Sector: Why Good Numbers Matter which convert a measured livestock parameter to a different data on livestock production levels, including of all major unit of measure: for example, ‘milk yield per cow per day’ al- livestock products. lows estimating the level of milk production by only counting the number of milking cows over a given period/area. Whichever the survey instrument, there are two main methodologies of data collection. The first consists of direct The next section briefly reviews methods and challenges to interviews, whereby an enumerator visits the (farm) house- collecting data on livestock production to generate nationally hold or some other stakeholder and asks him/her detailed representative statistics; section three introduces livestock questions on some livestock production variables. The second technical conversion factors and their role in producing consists of visual observations, whereby some actor, such as good quality livestock statistics; section four presents some an extension officer or a market agent, observes (in a more low-cost data collection methodologies to estimate selected or less structured way) production-related variables and livestock technical conversion factors, which have been fills a data spreadsheet (MLFD, 2012). Tables 5 to 8 provide recently applied by the Tanzanian government. Section five examples of survey questionnaires and data sheets used by presents conclusions. sub-Saharan African governments to collect data on livestock production levels. CHALLENGES IN COLLECTING DATA Assuming that no actor has incentives to misreport, direct ON LIVESTOCK PRODUCTION interviews and visual observations are appropriate to capture with statistical precision information on categorical vari- Four major survey instruments can be used to collect data ables which are slowly moving, such as the number of large useful to generate statistics on livestock production (see and small ruminants owned by a household, or main water chapter 1.4): sources. They can also be used to capture, although with less accuracy, information on variables for which the respondent ●● The agricultural census and, in some cases, the livestock is likely to have some, but not full, knowledge/memory, such census. These collate, process and disseminate data on as the number of animals affected by a certain type of disease a complete enumeration basis on a limited range of over the past 12 months or the amount of resources spent to structural items of agriculture, which change relatively treat sick animals over the reference period. slowly over time. The agricultural/livestock census usually collects data on milk and egg production and, in some Direct interviews and visual observations, however, are circumstances, on meat production. not the best methods to collect data on variables which are difficult to measure: these are typically continuous variables ●● Agricultural sample surveys, including specialized with relatively high variability, and whose value also depends livestock sample surveys, provide governments with com- on factors that are not under the control of the household, prehensive data on the livestock sector, which supplement such as rainfall. Cases in point are livestock production vari- census information. These surveys usually collect data on ables, such as meat, manure and milk production. In these production levels of all major livestock products. circumstances, technical conversion factors are often used or should be to generate statistically robust livestock production ●● Living standards measurement studies (LSMS) are indicators. multi-topic household surveys that aim to measure poverty and well-being and understand their major determinants. They collect data on livestock production, an important contributor of household livelihoods in “Whichever the survey instrument, developing countries. there are two main ●● Administrative record data, also referred to as routine methodologies of data collection. data, are regularly collected by national governments The first consists of direct with the objective of planning, implementing and moni- toring the delivery of public services. They often include interviews… The second consists of visual observations.” QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations PART II. METHODS TO IMPROVE THE QUANTITY AND QUALITY OF LIVESTOCK DATA   |  61 ©FAO/Giulio Napolitano TABLE 5. UGANDA LIVESTOCK CENSUS 2008: QUESTIONS ON MILK PRODUCTION Cattle Household Exotic Milk production identification number (ID) Indigenous Dairy Beef (litres) Household ID Household ID Household ID Household ID — TABLE 6. ETHIOPIA LIVESTOCK SAMPLE SURVEY 2010/11: QUESTIONS ON EGG PRODUCTION None Indigenous Hybrid Exotic Laying hens Egg production per hen per clutch Average number of days per clutch Total number of clutches during the reference period QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations 62  |  Investing in the Livestock Sector: Why Good Numbers Matter TABLE 7. N  IGER NATIONAL SURVEY OF HOUSEHOLD LIVING CONDITIONS 2011: QUESTIONS ON MEAT PRODUCTION What was the Over those months, average live what was the Livestock How many [animals] did you slaughter in the past 12 months? weight (in kg) of average quantity type animals that you of meat that you slaughtered? produced? Number of animals slaughtered Kg Kg 1 2 3 4 5 6 7 8 9 10 11 12 INDIGENOUS Cattle       Small rumin.     Camels     Pigs     Poultry   Guinea fowl CROSS/EXOTIC Cattle Small rumin. Pigs    Poultry QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations PART II. METHODS TO IMPROVE THE QUANTITY AND QUALITY OF LIVESTOCK DATA   |  63 TABLE 8. T  ANZANIA ADMINISTRATIVE RECORDS: DATA ENTRIES ON LIVESTOCK SLAUGHTERED AND MEAT PRODUCTION Total number slaughtered Total carcass weight (kg) Type of Livestock This quarter Cumulative to date This quarter Cumulative to date Cattle         Sheep         Goat         Pig         Chicken (local)         Chicken (improved)         Others (specify)       LIVESTOCK TECHNICAL CONVERSION nationally representative production and productivity statistics for the livestock sector. FACTORS In order to measure the level of production of livestock Technical conversion factors are coefficients that convert a products and by-products, three different levels of technical measured quantity to a different unit of measure. Examples conversion factors are typically used. First level technical of livestock technical conversion factors are: conversion factors allow calculating the amount of meat, offals, fat and fresh hides from every slaughtered animal; or ●● ‘Meat per slaughtered animal’, which allows calculating the amount of manure and milk from every animal/milking total meat production when multiplied by the number of animal. Second level technical conversion factors are used to animals slaughtered over a certain period in a certain area; decompose, say, meat in boneless flesh, butcher fat, salted ●● ‘Off take rate’, which allows arriving at an estimation of meat, sausage, and other. At the third level, technical coeffi- the number of animals slaughtered from total livestock cients are used to convert, say, cattle butcher fat into animal population data over the reference period; oil, tallow and other (FAO, 2000). ●● ‘Milk production per cow/day’, which allows estimating In a developing country context, where self-consumption of the level of milk production by counting the number of livestock products is common and processing limited, first milking cows over a given period/area; level technical conversion factors are of foremost importance and widely used to generate national livestock statistics. For ●● ‘Dung per adult cattle’, which allows calculating the level example, in the Tanzania National Accounts, beef production of production for one of the major by-products of large ru- is calculated by multiplying the total number of beef cattle minants, manure, by counting the adult cattle population slaughtered by 125, which is the technical conversion factor over the reference period; used to convert beef carcasses into kg of meat. ●● ‘Eggs per hen’; ‘dry matter intake/day per animal’; ‘weight The ‘meat conversion factors’ for goats, pigs and indigenous gain per kg of dry matter intake’; etc. are other technical chickens are 12, 45 and 2 kilos respectively; as for cow milk, conversion factors that, if available, are useful to generate the technical coefficient used is 1 litre of fresh milk/day per cow. The problem with Tanzania, and with most developing QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations 64  |  Investing in the Livestock Sector: Why Good Numbers Matter countries, is that the adopted technical conversion factors are CALCULATING LIVESTOCK TECHNICAL often obsolete; calculated using data from non-representative or biased samples; taken from neighbouring countries; and/ CONVERSION FACTORS or rarely updated. The consequences for decision makers can be serious, as Figure 6 shows. The data needed to calculate livestock technical conversion factors, as explained above, cannot be obtained with statisti- Figure 6 depicts the number of beef cattle slaughtered and cal precision through surveys or visual observation, and some the volume of beef production in Tanzania from first quarter direct, physical measurement is recommended. This can occur 2001 to fourth quarter 2011, as reported in the National at different points along the value chains but, for the purpose Accounts. Note that the slope of the two curves, and hence of calculating first level technical conversion factors, two are the distance between them, is constant over the reference pe- the appropriate sampling units: riod. This is so as, for the entire period, a constant technical conversion factor has been attached to carcasses to estimate ●● Farms, or households keeping livestock; beef production. ●● Abattoirs and/or slaughterhouses. The implication is that increase in production is all accounted At the farm level, data to calculate the following key conver- for by the increased number of animals slaughtered, and that sion factors can be collected accordingly (MLFD, 2012): likely improvements in animal productivity — which are in part reflected in the value of livestock technical conversion ●● Milk production/day per milking animal factors — are not captured in official statistics, which thus miscalculate the contribution of livestock to the gross do- Graduated transparent high-quality plastic containers can mestic product. From another perspective, all policies and be provided to farmers, who are then required to record investments implemented by the Ministry responsible for milk production at each milking, usually in the morning animal resources aimed to increase beef cattle productivity, and the evening. Farmers are also to be given a record such as wider vaccination coverage and better feeding, are card. This is a standard methodology to estimate (partial) unappreciated in official statistics. And the latter influence milk productivity. the way public resources are allocated across sectors and ●● Manure production/day per large and small ruminants between Ministries. There are three methodologies available to measure daily manure production from large and small ruminants. FIGURE 6. B  EEF CATTLE SLAUGHTERED AND The first consists of attaching a faecal bag to the animal BEEF PRODUCTION IN TANZANIA, and weighing the collected faeces at the end of the day. 2001–2011 This method has been often used in research stations 9 and mainly in stall-fed systems; in traditional systems, however, it is likely to influence animal ‘behavior’ and 8 hence to generate biased results. The second method consists of weighing for a few days the faeces of some 7 animal and then asking the farmers to count the number 6 of times that the sampled animals defecate each day. The third method, which is the most labor-intensive, 5 consists of following a sample of animals for a number of days and weighing their faeces as they defecate. The 4 latter is possibly the most accurate method to quantify Q1 2001 Q3 2001 Q1 2002 Q3 2002 Q1 2003 Q3 2003 Q1 2004 Q3 2004 Q1 2005 Q3 2005 Q1 2006 Q3 2006 Q1 2007 Q3 2007 Q1 2008 Q3 2008 Q1 2009 Q3 2009 Q1 2010 Q3 2010 Q1 2011 Q3 2011 manure production per animal/day in traditional production systems. No. of ind. cattle slaughtered (log) Beef, kg (log) Source: Tanzania National Bureau of Statistics, unpublished data QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations PART II. METHODS TO IMPROVE THE QUANTITY AND QUALITY OF LIVESTOCK DATA   |  65 ●● Eggs/laying bird per clutching period supplemental feed for animals) and be provided at the end of the data collection exercise to avoid biased results. Basic A simple record card can be given to farmers to record the equipment such as disinfectants, raincoats, knives and boots number of eggs produced by each laying bird, provided are appropriate incentives to ensure good data collection in that she is in her clutching period. This methodology slaughterhouses/abattoirs. is straightforward, but farmers need also to provide information on the length of the clutching period, a Finally, while one-off investments to update livestock conver- pre-condition to arrive at quarterly/annual estimates of sion factors are valuable, country governments should make egg production. all efforts to ensure that livestock technical coefficients be regularly updated, a pre-condition for the efficient allocation In abattoirs/slaughterhouses, data to calculate the following of public resources. Updated technical conversion factors technical conversion factors can be collected: also reduce the need to collect data on livestock production through surveys or administrative records, thereby reducing ●● Live weight and carcass weight of slaughtered animals; the financial and human resources needed for implementing and meat, offals and fat content of carcasses. agricultural/livestock surveys and routine data collection There are tools and equipment — such as scales and (administrative records). carcass weighers — that slaughterhouses use to measure live weight, carcass weight and the meat, offals and fat content of the carcass. Many slaughterhouse/abattoirs are already equipped with effective measurement tools and, in these premises, slaughterhouse managers should be easily able to record, if required, selected production parameters on a daily basis. The above methodologies are not complex, but their implementation is challenging. First, to be meaningful for statistical, policy and investment purposes, technical conversion factors should be representative for the country as a whole and, possibly, for its major agro-ecological zones. In addition, seasonality should be captured. This has impli- cations for both the sample size and the time length of data collection, making it expensive the estimation of statistically accurate livestock technical conversion factors (ILCA, 1990; Thomson, 2012). Second, farmers in particular, but also abattoir/slaugh- terhouse managers, should be trained to properly collect the data needed to estimate livestock technical conversion factors, and be provided with equipment/tools for measuring and recording production parameters, such as a graduated plastic containers for quantifying milk production. Third, some incentives should be given to farmers and slaughterhouse/abattoir managers for proper data collection. ©FAO/Giulio Napolitano As a general rule, cash incentives should be avoided, as they may jeopardize future data collection activities, and in-kind incentives are to be preferred. At the farm level, these should possibly target livestock production (e.g. balanced/ QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations 66  |  Investing in the Livestock Sector: Why Good Numbers Matter CONCLUSIONS given period/area. Technical conversion factors are best calculated by physically measuring the value of selected parameters at different points along the value chains, but Measuring livestock productivity, and understanding in most countries the value of technical coefficients is its determinants, is essential to design and implement obsolete or sourced from inappropriate datasets. investments that maximize the contribution of livestock to socio-economic development. Productivity relates in- This chapter presented methods to collect data to calculate puts to outputs, and the quality of productivity measures key livestock technical conversion factors, namely milk strongly depends on the quality of the data available to production/day per milking animal; manure production/ measure them. These data, when it comes to producing day per large and small ruminants; and eggs/laying bird nationally representative statistics, are often of poor per clutching period at the farm level; and to collect data quality. to quantify live weight and carcass weight of slaughtered animals; and meat, offals and fat content of carcass in Traditional methods of livestock data collection, including slaughterhouses and abattoirs. The methods presented are direct interviews and visual observation used in surveys straightforward, but appropriate sampling, incentives and and administrative records, are not the best methods to institutional arrangements are needed for proper data col- collect data on variables that are continuous and difficult lection and the ensuing calculation of technical conversion to measure in low-income settings, such as meat, milk factors. Livestock technical coefficients should be updated and manure production. In these circumstances, technical regularly to properly measure livestock production and conversion factors are used or should be used to produce productivity. This allows one to assess the effects of poli- accurate, nationally representative statistics. These are cies and programs on the ground and to properly estimate coefficients that convert a measured livestock variable to livestock value added, i.e. the contribution of livestock to a different unit of measure: for example, ‘milk yield per GDP, which influences the way public resources are allo- cow per day’ allows estimating the level of milk produc- cated for livestock developmental purposes. tion by only counting the number of milking cows over a QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations PART II. METHODS TO IMPROVE THE QUANTITY AND QUALITY OF LIVESTOCK DATA   |  67 2.4 INSTITUTIONAL CHANGES TO IMPROVE THE QUANTITY AND QUALITY OF ADMINISTRATIVE LIVESTOCK DATA KEY MESSAGES reason, they are widely used to design, implement and moni- tor livestock sector policies and investments. Good administrative records, also called routine data, are critical for policies and investments Routine livestock data also contributes to regional and design as they provide data at low administrative international livestock-related information systems and/or databases, such as the Livestock Information Management level. System (LIMS) of the Southern Africa Development Community (SADC), the Animal Resources Information Routine data are often considered of relatively System 2 (ARIS 2) of the Interafrican Bureau for Animal poor quality, as they are collected by extension Resources of the African Union (AU-IBAR), CountrySTAT and officers who are rarely, if ever, trained FAOSTAT of the Food and Agriculture Organization (FAO), and the World Animal Health Information System (WAHIS) statisticians or trained in data collection. of the World Organization for Animal Health (OIE). Indeed, international obligations require that African countries Routine data, on paper, are collated on a submit monthly, six-monthly and annual animal health/ complete enumeration basis, which make data disease reports to the World Organization for Animal Health (OIE) — the reference organization to WTO with respect to collection extremely demanding. A sampling trade-related trans-boundary animal diseases (TADs) — to approach is possibly a more effective way to the Africa Union-Interafrican Bureau for Animal Resources collect data at local level with some statistical (AU-IBAR); and to some Regional Economic Communities accuracy. (RECs). Despite governments’ and other regional and international Institutional experiments, whereby different institutions’ wide-ranging use of routine livestock data, ad- methods to organize data collection at local level ministrative records are often incomplete, out-of-date and are performed on a small scale and their efficacy unreliable. Insufficient resources, and limited skills in da- compared, are an effective way to improve the ta-handling and processing, are the two most-cited reasons system of routine livestock data collection. for the inadequacy of administrative records. Improvement is thus essential to promote evidence-based policy and in- vestment decisions and implementation. Notably, the Global Strategy to Improve Agricultural and Rural Statistics considers administrative records to be one component of the integrat- INTRODUCTION ed survey framework; it highlights that routine data are a key source of information for generating several indicators Most livestock data publicly available in sub-Saharan for agricultural statistics; and it includes administrative African countries are collected either by the National Office data as one of the priority research areas in its Action Plan of Statistics or by the Ministry responsible for livestock for Africa. development. The latter, often in cooperation with local government authorities, collects livestock-related data at a Efforts to improve administrative records in developing low administrative level during its routine operation. These countries, however, have to date been limited. But for few data, called routine data or administrative records, are, along exceptions, such as the JICA-sponsored improvement of the with census data, the only ones that provide information at agricultural routine data in Tanzania, national and interna- district/province or lower levels of disaggregation. For this tional investments have mostly targeted censuses and sample QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations 68  |  Investing in the Livestock Sector: Why Good Numbers Matter surveys. There are thus few experiences and methodologies livestock. The next sections describe this system and present available to allow assessment and improvement of routine and apply to Uganda a rapid assessment methodology for data systems. In turn, this further contributes to reduced livestock administrative records. A section follows that pro- investments in administrative records. poses actions for improvement. These proposals are intensive ‘field experiments’ or pilot approaches with control groups, This paper presents a methodology for undertaking a which represent significant institutional changes in Uganda. rapid assessment of routine livestock data systems and A last section presents conclusions and recommendations. identifies options for improvement. It has been developed by the Uganda Ministry of Agriculture, Animal Industry and Fisheries (MAAIF) and the Uganda Bureau of Statistics ROUTINE LIVESTOCK DATA (UBOS), in collaboration with the FAO-World Bank-ILRI- COLLECTION IN UGANDA AU-IBAR Livestock in Africa: Improving Data for Better Policies Project. Uganda, like several other developing countries, has The Directorate of Animal Resources within the Uganda a system of routine data collection that explicitly targets Ministry of Agriculture (MAAIF) is comprised of two Departments, namely the Department of Animal Production and Marketing and the Department of Livestock Health and Entomology. The Directorate of Animal Resources is man- dated to formulate and implement livestock sector policies, plans and programs, and to control and manage epidemic animal diseases. MAAIF makes use of census and survey data to fulfil its mandate, but its major source of information on livestock is administrative records. These represent the country’s only information regularly available at district and lower administrative level and, therefore, are of primary importance to MAAIF. The system of routine data collection in Uganda is structured as follows. Sub-county level Livestock/Veterinary officers are responsible for provision of extension services to rural households, and for collection of some livestock-related data during their routine work. These officers collect data according to a reporting form formulated at the district level: across districts there is no unique format used, as data are primarily collected to meet the differing information needs of District Authorities/Local Governments. On a monthly basis, the District Livestock/Veterinary Officer compiles and assembles the data gathered by extension officers in the var- ious sub-counties and submits a pre-designed livestock data reporting form to MAAIF, through his/her respective Chief Administrative Officer. It is notable that District Authorities are not legally obliged to report to MAAIF, as they are subor- dinated to the Ministry of Local Government. The livestock data report that districts compile on a monthly ©FAO/Gianluigi Guercia basis includes information under several headings: QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations PART II. METHODS TO IMPROVE THE QUANTITY AND QUALITY OF LIVESTOCK DATA   |  69 ●● ‘General information’, namely basic information on rain- ●● ‘Animal quarantine and other restrictions’, including fall pattern; water availability and grazing conditions; number of counties/sub-counties quarantined; number of livestock markets closed; control measure taken; etc.; ●● ‘Outbreaks of contagious diseases’, including outbreaks of any of 28 major diseases, numbers of animals affected and ●● ‘Animal production’, which refers to number of live ani- at risk, and action taken to control/manage any outbreak; mals in the district by species; ●● ‘Rabies’ cases, including those in humans; ●● ‘Types of livestock farming systems in the district’, i.e. number of animals in pastoral/communal, semi-extensive, ●● ‘Vaccination’, which refers to the number and species of semi-intensive and intensive production systems; animals vaccinated against any of 8 major diseases (CBPP, FMD, LSD, Black Leg, Brucellosis, NCD, Rift Valley Fever, ●● ‘Livestock markets’, which collects information on CCPP); number of live animals offered and sold in the different markets and maximum, minimum and average price; ●● ‘Other clinical cases handled’, by species, which refers to first aid and surgical interventions, diarrhea, mastitis and ●● ‘Hides and Skins’, including salted and non-salted and others; kilograms produced; ●● ‘Tick control’, including number of cattle dipped; num- ●● ‘Staff disposition and vehicle strength’, namely grade of ber of dip tanks available by ownership (communal or staff and level of education; number of vehicles by type private); (e.g. trucks; 4WD; motorbikes; etc); and other equipment available, such as computers, GPS, refrigerators and ●● ‘Dip wash testing’, which reports on acaricide type, num- generators. ber of samples tested and the results of tests. The routine data that MAAIF collects largely target animal ●● ‘Laboratory activities’, i.e. results of analyses of blood/ health and diseases, with some limited information on the lymph node smears; faeces and serum. livestock population (production) and on livestock markets. Indeed, almost 60 percent of the 2011/12 MAAIF budget for ●● ‘Vaccine stocks’, with details on doses available and date ‘animal agriculture’, excluding fishery, is allocated to ‘vector of expiry; and disease control measures’, which basically means animal ●● ‘Internal animal movements in relation to animal laws’, vaccination. Note that not all information in the livestock including from/to other districts and means of movement reporting format can be regularly sent by District Authorities (e.g. foot; truck/train; or air); to MAAIF: for example, new outbreaks of animal diseases do not occur every month, nor in all districts is there a function- ●● ‘Artificial insemination’ for four major dairy cattle breeds al laboratory or a quarantine station. In any case, the amount (Friesian, Ayreshire, Guernsey and Jersey); of information that districts should produce on a monthly basis is significant and should suffice to formulate and moni- ●● ‘Veterinary regulatory activities’, i.e. information on dis- tor the implementation of animal health-related policies and semination and sensitization meetings on animal-health investments. related issues; ●● ‘Meat inspection’, namely pre- and post-mortem inspec- tion activities and results by species; QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations 70  |  Investing in the Livestock Sector: Why Good Numbers Matter BOX 6. ROUTINE LIVESTOCK DATA COLLECTION IN ZANZIBAR R outine livestock data, or administrative record data, are regularly collected by the Ministry of Livestock and Fish- eries (MLF) of the Revolutionary Government of Zanzibar. gender and animals owned, including cattle (indigenous and improved), goats (indigenous and improved), indigenous poultry, and layers and broilers; (b) number of farmer groups MLF staff work in the Central Government, the Districts and by animal species and membership; (c) animals owned by the Shehias. The first step of data collection is performed species by government farms, including multiplication units at Shehia level, where, as one of their tasks, so-called Live- for dairy cattle and dairy goats; (d) number of animals sold, stock Production Assistants and Para-veterinarians collect both within Zanzibar and between Zanzibar, Tanzania main- livestock-related data from livestock keepers. These data are land and other countries; (e) number of animals slaughtered, sent every month to the District Authority, where the District yield (lit / kg) and production of cow and goat milk, beef, Livestock Officer and the District Veterinary Officer prepare goat, chicken and eggs; (f) types of extension services pro- monthly reports and send them to MLF HQs. In particular, vided (e.g. dairy husbandry practices; pasture management; every month District Officers submit to MLF HQs: (a) Animal animal welfare, etc.) and number of beneficiaries, as well Health Reports; (b) Animal Production Reports. MLF then as farmer field schools organized; (g) revenue collection, compiles monthly Animal Health and Animal Production Re- primarily from sales of pasture seeds and feed for animals; ports, which cover the whole of Zanzibar. These reports are (h) number of staff available by gender and participation in neither submitted to AU-IBAR nor to the World Organization training. of Animal Health (OiE). MLF’s objective is clearly to ensure regular and good quality In some circumstances, Shehia and District Officers also information on the livestock sector in Zanzibar, with a focus obtain data from Community Animal Health Workers, even on animal health and production. However, the quantity and though the latter are not MLF staff. Another source of data quality of available livestock data is often unsatisfactory, for are the so-called Animal Health and Production Centres of a number of reasons. (a) officers in Districts and Shehias are MLF. There are about 20 such Centers in Zanzibar, which not trained in data collection/analysis, which is one of their are located in the higher livestock concentration areas and many tasks, and not among their top priorities; (b) Livestock provide livestock keepers with clinical, diagnostic, treatment Production Officers and Para-vets in Shehias collect data and extension services. Finally, when there are disease from the farmers they visit, which may differ from month outbreaks that risk spreading throughout the islands, MLF to month; (c) while there is a common data format for MLF provides human and financial resources to Local Govern- District staff to compile the monthly reports, at Shehia ments to control the disease. Additional data are collected in level, there is no common template, with extension officers these circumstances, which can enter the monthly reports. collecting and reporting data as they prefer; (d) at local level, resources are often scarce and, therefore, Districts do not The Monthly Animal Health Report targets a variety of infor- always send with regularity their Animal Health and Produc- mation, including: (a) disease outbreaks by type of disease tion Reports to MLF HQs. and animal species (cattle, sheep, goats, donkeys, chicken, ducks, cats and dogs); (b) number of animals by species MLF has plans to improve the quantity and quality of affected, treated (by type of treatment) and dead (by type routine livestock data, including recruiting more staff and of disease); (c) number of vaccinations, disease control and conducting staff training to establish benchmark data, and warm control practices by animal species and practice; (d) information systems. It recognizes the major challenges activities in quarantine stations (at ports and the airport), inherent in the generation of good quality production statis- and related to meat inspections and laboratory investiga- tics, including information on off-take, carcass weight and tions; (e) revenue collection, primarily generated by service milk yield per animal. Virtually all efforts to control and erad- fees (e.g. for AI or dipping) and movement permit; (f) num- icate animal diseases have as an objective the improvement ber of staff available by gender and participation in training. of livestock productivity. The challenge is to measure these productivity gains, and, ultimately, to contribute to improved The Monthly Animal Production Report contains the fol- livelihoods for livestock farmers. • lowing information: (a) number of livestock keepers by QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations PART II. METHODS TO IMPROVE THE QUANTITY AND QUALITY OF LIVESTOCK DATA   |  71 ©FAO/Simon Maina AN ASSESSMENT OF THE UGANDA to the Ministry of Agriculture/Livestock, including the proportion of sections filled. This ratio provides an indi- ROUTINE DATA SYSTEM cation of the capacity of local staff/authorities to report on specific data items. Indeed, while information on some Routine livestock data are a critical piece of information for variables can be easily captured — number of vaccines ad- the Ministry responsible for animal resources and, if properly ministrated by extension officers — other is more difficult collected, it could become an integral part of the statistical to gather, such as average market prices for live animals. system. So far, however, despite ample criticism of admin- istrative records, there have been few if any attempts to ●● Qualitative assessment — Semi-structured interviews comprehensively assess routine data systems. In most cases, with expert informants, including not only those directly evaluations target specific issues of routine data systems in involved in data collection and analysis, but also staff in industrialized economies, such as the use of administrative the National Bureau of Statistics, who can provide a sta- records to identify undercounted population in the human tistical perspective on data systems usually managed by census; or to update the survey framework by, for example, agricultural/livestock experts. providing updated information on the dynamics of private and public sector businesses (Sheppard et al., 2013). Number of reports This section first presents a low-cost methodology to assess Figure 7 displays the number of livestock data reports sub- routine livestock data and then applies it to Uganda. The mitted by the 112 Uganda Districts to MAAIF from January proposed methodology builds on both quantitative and quali- to December 2012. Figure 8 summarizes the frequency of dis- tative information and employs three measures: trict reporting: the histogram shows a U-shape distribution as out of 112 districts, only 31, or 27 percent, regularly sub- ●● Number of data reports — A quantitative assessment of mitted their monthly livestock data report to MAAIF in 2012; the number of statistical reports submitted by local staff on the other hand, another 16 districts, or 14 percent, never and/or local authorities to the Ministry of Agriculture/ reported to MAAIF that year. The remaining 66 districts Livestock versus the number of reports due. Although reported to MAAIF in a number of months between 1 and 11 simple, this ratio is a good indicator of the effectiveness of in 2012. The overall reporting rate stands at 62 percent, i.e. the prevailing institutional architecture, including mecha- of 112 reports expected each month — one per district — 70 nisms of data collection and reporting. were received by MAAIF in 2012. An immediate conclusion is that the current institutional architecture of data collection ●● Completeness of data reports — A quantitative assess- and reporting does not properly work. ment of the completeness of the information in the different sections of the statistical reports submitted QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations 72  |  Investing in the Livestock Sector: Why Good Numbers Matter FIGURE 7. U  GANDA: LIVESTOCK DATA REPORTS SUBMITTED BY DISTRICTS BY MONTH, JANUARY–DECEMBER 2012 DISTRICT Bukomansimbi Kaberamaido Kiryandongo Bundibugyo Kapchorwa Kamwenge Butambala Bulambuli Kalangala Adjumani Bushenyi Amolatar Kabarole Buhweju Kanungu Alebtong Buyende Bukedea Kampala Kaabong Kiruhura Kayunga Buvuma Butaleja Kalungu Bududa Amudat Katakwi Budaka Gomba Isingiro Kibaale Buikwe Amuria Dokolo Ibanda Kamuli Kibuku Kasese Kiboga Kabale Amuru Bukwa Buliisa Hoima Iganga Kisoro Agago Bugiri Kaliro Busia Abim Gulu Apac Arua Jinja Month Jan 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 Feb 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 Mar 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 Apr 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 May 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 Jun 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 Jul 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 Aug 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 Sep 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 Oct 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 Nov 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 Dec Total 10 0 11 7 3 12 11 2 12 9 0 6 0 12 12 11 11 0 12 11 0 6 8 12 0 7 8 9 12 6 12 10 4 10 9 6 12 8 6 6 11 10 0 6 7 9 4 12 0 5 12 11 5 10 0 2 Nakasongola Namutumba Nakapiripirit Kyankwanzi Sembabule Namayingo Ntungamo Lyantonde Mubende Kyegegwa Nakaseke Rukungiri Manafwa Mitooma Maracha Mbarara Kyenjojo Mukono Sheema Masindi Ntoroko Mityana Mayuge Masaka Lwengo TOTAL Sironko Koboko Moroto Luwero Rubirizi Lamwo Wakiso Kitgum Yumbe Zombo Nwoya Kween Tororo Serere Mbale Napak Kotido Pallisa Nebbi Ngora Luuka Soroti Otuke Pader Oyam Mpigi Moyo Rakai Kumi Kole Lira Month Jan 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 84 Feb 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 85 Mar 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 80 Apr 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 73 May 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 70 Jun 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 67 Jul 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 73 Aug 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 70 Sep 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 67 Oct 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 66 Nov 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 56 Dec Total 12 7 10 10 0 3 10 8 12 6 12 9 5 12 12 8 12 1 12 12 4 0 12 10 2 12 9 0 8 12 12 12 4 3 12 8 12 12 3 7 12 0 0 0 12 11 11 9 3 8 3 12 7 11 0 12 845 QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations PART II. METHODS TO IMPROVE THE QUANTITY AND QUALITY OF LIVESTOCK DATA   |  73 total number of reports that should have been submitted FIGURE 8. U  GANDA: FREQUENCY OF DISTRICT (Figure 9), and over the number of actual reports submitted REPORTING (Figure 10). In other words, Figure 9 shows the probability 35 31 for MAAIF of getting the information for the data item at 30 hand, while Figure 10 shows the probability of getting that NUMBER OF DISTRICTS 25 same information conditional on selecting one of the reports 20 submitted to MAAIF by the district authorities. 16 15 9 10 Figures 9 and 10 substantiate the evidence that the current 10 8 8 6 6 7 system of routine data collection and reporting is somewhat 5 3 4 3 1 inadequate: not only are relatively few reports regularly 0 0 1 2 3 4 5 6 7 8 9 10 11 12 submitted, but those submitted are often incomplete. The NUMBER OF REPORTS SUBMITTED most reported item is ‘general information’ which, as said, comprises basic information on rainfall pattern, water avail- ability and grazing conditions: this is reported in 35 percent Completeness of reports of expected cases, and present in 56 percent of the submitted The second step for assessing routine data systems is to look reports. In other words, there is a probability of 33 percent at the completeness of the reports received by MAAIF. As of getting ‘general information’ from any district and a prob- noted, the required information can be difficult to gather ability of 56 percent of finding that information among the and assemble for data collectors and authorities at the local available reports, with ‘general information’ being the most and national level. Figures 9 and 10 display the number of reported data item. livestock data reports, by section, as a proportion of the FIGURE 9. U  GANDA: DISTRICT OVERALL FIGURE 10.  UGANDA: DISTRICT CONDITIONAL REPORTING RATE REPORTING RATE Dip wash testing Dip wash testing Vehicles/equipment Vehicles/equipment Staff disposition Staff disposition Animal production Animal production Internal animal movements Internal animal movements Artificial insemination Artificial insemination Vaccine stocks Vaccine stocks Hides and skins Hides and skins Regulatory activities Regulatory activities Rabies Rabies Farming systems Farming systems Ante-mortem inspection Ante-mortem inspection Livestock markets Livestock markets Vaccinations Vaccinations Tick control Tick control Clinical cases Clinical cases Post-mortem inspection Post-mortem inspection General information General information 0% 20% 40% 60% 80% 0% 20% 40% 60% 80% QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations 74  |  Investing in the Livestock Sector: Why Good Numbers Matter Qualitative assessment “Extension officers lament that A team from the Ministry of Agriculture, Animal Industry data collection — and other and Fisheries and the Uganda Bureau of Statistics conducted activities they must perform — semi-structured interviews with expert informants to assess the system of routine data collection. The team travelled to involves significant movement three selected districts — namely Lira, Nakasongola and for which they have insufficient Soroti — which submitted all reports to MAAIF in 2012 and are located in the so-called cattle corridor, an area stretching resources, such as motorbikes, from northeast, through central to southwest Uganda and computers and fuel.” with a high animal population density. Semi-structured interviews were conducted with extension officers, who are responsible for data collection at sub-county level, and with some animals. This means that an extension officer, while the district veterinary officers, who are tasked with assembly performing his many other activities, should interview of the data gathered by extension officers and compilation about 100 households per day — assuming he/she works of reports for MAAIF. Then discussions were held with staff 24 days a month — in addition to gathering information from the College of Veterinary Medicine and Biosecurity, the from other sources, such as in livestock markets and National Agricultural Research Organization, the College of abattoirs. Agricultural and Environmental Science, the Animal Genetic ●● Extension officers are not trained in data collection and Resource Centre and Data Bank, the Dairy Development handling, and gather their information during their Board and the National Drug Authority. The conclusions daily activities. They do not follow specific rules and were: procedures, nor do they administer survey questionnaires ●● District authorities contend that livestock data are critical to households that have livestock and other relevant for management and planning, primarily for animal dis- stakeholders such as market authorities. Scattered direct ease control and management. Indeed, in all districts data observations are the norm. collection prioritizes animal vaccination and animal treat- ●● The livestock statistical report that District authorities ment, though some information is also collected on other submit to MAAIF includes data items that are not con- tasks performed by extension officers and the veterinary sistently defined. Some data reflect the routine work officers, such as artificial insemination and post-mortem undertaken by extension officers, such as the number of inspection of carcasses. Only Nakasongola district au- animals vaccinated; other data are based on ad hoc data thorities mentioned animal population as a key indicator collection, such as data on market prices for live animals for management and planning. Only in Soroti district are and on the livestock population; and data focus on both data stored electronically; in Lira and Nakasongola paper relatively static and highly dynamic items, such as number forms are used. of staff and vehicles available in the district office and out- ●● Extension officers lament that data collection — and breaks of animal diseases. This inconsistency makes data other activities they must perform — involves significant compilation and reporting difficult. movement for which they have insufficient resources, ●● The College of Veterinary Medicine and Biosecurity, the such as motorbikes, computers and fuel. Indeed, pa- National Agricultural Research Organization, the College per-based data collection should be done on a complete of Agricultural and Environmental Science, the Animal enumeration basis, but this is rarely, if ever the case. Genetic Resource Centre and Data Bank, and the National ●● Even if extension officers had enough resources to visit all Drug Authority collect their own data, such as on breeds, households that keep livestock in each sub-county, this breeding practices and reproductive performance. These would still pose a major challenge. According to UBOS data would represent a valuable input into policy design data, in a typical sub-county there are about 4000 house- and implementation if complemented by those collected holds, of which about 2400 or 60 percent on average keep by District authorities on a monthly basis. QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations PART II. METHODS TO IMPROVE THE QUANTITY AND QUALITY OF LIVESTOCK DATA   |  75 OPTIONS TO IMPROVE THE related to animal disease management and control. This information should not be used to generate official sta- LIVESTOCK ROUTINE DATA SYSTEM tistics. The quarterly report will target only information on the livestock population and market prices for live The MAAIF-UBOS assessment of the routine data system animals and hides and skins. This information, if prop- in Uganda revealed major weaknesses, which need to be ad- erly collated, can be used to generate official statistics. dressed to ensure proper management of the livestock sector. The annual report contains only information on major MAAIF and UBOS duly established a small team to identify livestock-related physical and human resources available options for improvement of the routine livestock data collec- in the district, such as slaughterhouses, market facilities, tion system. This team based its work on four assumptions. and staff by grade. It could also contain summary tables First, any improvement in the routine data system should derived from the monthly and quarterly reports. start from the set of core livestock indicators, as identi- fied and endorsed by the National Agricultural Statistical 2. Extension officers in all sub-counties should use a com- Committee. These are indicators needed by MAAIF and UBOS mon collection and reporting format. In particular, one on a regular basis and collected using their recurrent budget. form should target the monthly information and the oth- They are the core indicators presented in Chapter 1.2. er the quarterly information that districts are supposed to send to MAAIF. While extension officers can collect Second, routine data, if collected according to sound statis- data for the monthly report during their routine work, the tical principles, could also be used by the National Statistical information in the quarterly report requires some target- Authority, thereby facilitating data integration and improv- ed data collection activity. Extension officers should be ing the overall efficiency of the agricultural statistical system. trained to administer questionnaires to collect these data. As far as possible, therefore, statistical principles should be adopted by the routine livestock data collection system. 3. Four pilots are suggested to implement sound statistical principles in gathering routine livestock data which are Third, the budget allocated to extension and data collection is collected on a quarterly basis. The pilots build on the limited and, most likely, will remain limited. Options to im- evidence that, as shown, data collection on a complete prove routine data, therefore, should attempt to simplify the enumeration basis is not achievable with current human current system and involve little or no increase in the current resources and, therefore, a sampling approach is needed. budget. Indeed, there will be transaction costs to move to an Sub-counties will be subdivided into enumeration areas improved data collection system, but these are one-off, or una (EAs) — a list of EAs is already available and, in most cas- tantum, investment costs. es, one EA corresponds to one village. In each sub-county Finally, various institutional reforms can be devised to the extension officer will travel either in all, or a sample improve the routine livestock data collection system. A of, EAs for data collection; in the sampled EAs s/he will priori, however, it is difficult to identify the most appropriate interview a sample of households and, depending on the and efficient reforms. Pilot implementation of alternative case, s/he will be given an incentive for data collection, institutional reforms to identify the most promising options such as some free fuel. The four approaches, which are is widely appreciated as an effective way of promoting signifi- summarized in Table 9, vary because of different sampling cant improvements. Based on these assumptions, and on the and resources provided to extension officers for data rapid assessment of the routine livestock data system, the collection. Note that in two cases the current budget following is recommended: should suffice to implement the proposed new systems of data collection at the country level, while in the other 1. District authorities should produce monthly, quarterly two some additional budgetary allocation is anticipated. and annual statistical reports to be shared with MAAIF, To identify which of the different pilots provides better constructed so as to recognize demands on the time of estimates of the livestock population in the country, a the extension officers and the District Veterinary Officers. livestock census will be conducted in the pilot sub-coun- The monthly report will target only data related to animal ties, which will also allow building an updated frame for diseases, including information on disease outbreaks, selecting the sampled households. Results will be com- on vaccination and treatment, and other core activities pared with those from two control sub-counties, in which QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations 76  |  Investing in the Livestock Sector: Why Good Numbers Matter the current monthly reporting systems will remain in help to identify the most appropriate institutional reform place. Implementation of the pilots will be joint responsi- for improved routine livestock data collection. The proposed bility of MAAIF, UBOS and Local Government Authorities. pilots target only data collection and do not include any ac- tivity related to data transfer and analysis. Finally, it is worth The implementation of the proposed pilots will provide evi- noting that independent of the implementation of any pilot, dence on whether or not statistical principles can be brought MAAIF can request Districts to adopt the proposed monthly, into the routine livestock data collection system. It will also quarterly and annual livestock statistical reporting formats. TABLE 9. U  GANDA: PROPOSED PILOTS TO IMPROVE THE ROUTINE SYSTEM OF LIVESTOCK DATA COLLECTION Pilot 1 Pilot 2 Pilot 3 Pilot 4 Sub-county 1 Sub-county 2 Sub-county 3 Sub-county 4 EAs All All Sample Sample Households Sample Sample Sample Sample Training for extension Yes Yes Yes Yes officers Resources to extension No Yes No Yes officers Benchmark Livestock Census Livestock Census Livestock Census Livestock Census ©FAO/Simon Maina QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations PART II. METHODS TO IMPROVE THE QUANTITY AND QUALITY OF LIVESTOCK DATA   |  77 CONCLUSIONS livestock data collection is inadequate because of missing information and poor quality of the data. The paper pro- poses to streamline the current livestock-data reporting The Ministry responsible for livestock development, often form, by suggesting that MAAF should request District in cooperation with local government authorities, collects authorities to report on different items on a monthly, livestock-related data on a regular basis in the course of quarterly and annual basis. It then sketches four possible its routine operation. These data, called routine data or pilots to identify the first best institutional reform for administrative records, are compiled at relatively low cost an improved system of routine livestock data collection. and collected at ground level. They represent a critical The pilots contain three innovative elements. First, two input into policy and investment design, implementation, of the proposed pilots are budget neutral, i.e. they could monitoring and evaluation, and the management of the be implemented with a one-off investment and without livestock resources more generally. the need to increase the recurrent expenditure budget. There is scattered evidence that in developing countries Second, they introduce sound statistical principles to ad- routine livestock data are inadequate, and no standard ministrative records by proposing a sampling approach for methodology is available to assess their quality. This the routine data collection. Third, the pilots are designed paper presented a methodology for a rapid assessment of to tests the relative efficiency of alternative institutional the routine livestock data system, which builds both on arrangements underpinning routine livestock data quantitative and qualitative information. The quantitative collection. information targets the number of available statistical While designing and testing alternative pilots to improve reports and their completeness; the qualitative infor- the routine livestock data collection system in Uganda is mation includes semi-structured interviews with expert recommended, the adoption of improved monthly, quar- informants. terly and annual livestock statistical reports — which is a The methodology to assess the routine livestock data sys- no-cost action — is also expected to enhance the quality tem was applied to Uganda. The current system of routine of routine livestock data. QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations 78  |  Investing in the Livestock Sector: Why Good Numbers Matter PART III. LIVESTOCK DATA FOR DECISION MAKING: EVIDENCE AND EXAMPLES 3.1 ESTIMATING LIVESTOCK NUMBERS: EXAMPLES FROM COUNTING ANIMALS IN WEST AFRICA KEY MESSAGES A priority core indicator of relevance to Agricultural/livestock censuses are not governments and livestock practitioners are undertaken regularly. In the interim, models statistically sound — both nationally and locally could be used to update the estimates of the — livestock numbers. livestock population. The agricultural/livestock census or agricultural/ FAOSTAT data suggests that livestock population livestock surveys are potentially effective survey estimates in West African countries are tools to collect data on the livestock population. somewhat inaccurate. Both are undertaken on a sample basis, however, which leads to biased estimates of the livestock population when the sampling units are rural households or farm households, as is often the case. QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations PART III. LIVESTOCK DATA FOR DECISION MAKING: EVIDENCE AND EXAMPLES   |  79 ©FAO/Giulio Napolitano INTRODUCTION time. The quality of these models strongly depends on the availability of reliable and timely data to estimate some key parameters, such as calving rate and pre-weaning mortality. Statistically sound livestock numbers are a critical core These data, however, are often lacking and many countries, statistical indicator (see chapter 1.2) needed to formulate, therefore, just apply a constant rate of growth, such as 3 per- implement and monitor livestock sector investments, cent, to available census data to generate livestock population both in the public and private sector. They also feed into estimates over years. The growth rate is adjusted, in some the generation of other key sector statistics, including the cases, to reflect weather variability, the availability of pasture calculation of ‘livestock value added’ as an input into the and water, and on occasion, disease outbreaks. gross domestic product (GDP). Agricultural and/or livestock censuses and surveys are the first best source of data to esti- This chapter provides evidence on how West African coun- mate the livestock population in a country. However national tries estimate the livestock population. First, it reviews governments rarely undertake, with regularity, agricultural agricultural/livestock censuses and surveys undertaken in or livestock censuses and, in many cases, agricultural sample West Africa since 2000, including two country case studies. surveys do not generate accurate estimates of the livestock It then reviews the structure of herd growth models and de- population, mainly because of sampling issues, as revealed in scribes how country governments have been estimating the chapter 1.4. livestock population between censuses and surveys. The final session summarizes the main evidence and provides some In the absence of readily available statistics, statistical recommendations for improving the agricultural statistical agencies and livestock departments could, building on survey system in a way that produces more reliable livestock popula- data, use demographic herd models to simulate the future tion estimates. evolution of the livestock population and its structure over QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations 80  |  Investing in the Livestock Sector: Why Good Numbers Matter BOX 7. LIVESTOCK POPULATION: A CRITICAL STATISTICS B etween January and February 2012 the Livestock in Africa: Improving Data for Better Policies Project administered a global online survey among livestock stakeholders (Pica- 8. Labor force devoted to livestock; 9. Animal power, which primarily includes data/indicators Ciamarra et al., 2012). The primary objective was to identify on the use of animals for draught power and for hauling and rank core livestock domains/areas for which livestock services; data/indicators are demanded. The survey targeted 10. Meat production; livestock-related data and indicators along the value chain. These include information on livestock inventories; inputs 11. Milk production; and husbandry practices; production; and consumption of livestock products, i.e. data/indicators that measure and 12. Egg production; provide information on livestock market opportunities, 13. Production and use of dung, including but not only as production and marketing-related constraints. A total of 641 manure; respondents filled in the survey questionnaire. Respondents were asked to rank in the importance data/indicators in 14. Hides & skins production; 15 livestock domains. Ranking is based on a 5 level rating scale (most important; important; useful; partly useful; 15. Consumption of animal source foods. marginally useful), while the livestock domains are: Under each domain quantity and price data can be collected 1. Livestock inventory; to generate various indicators, including value indicators (quantity × price). A specific question on the importance of 2. Change in livestock stock, which includes data/indica- getting price information was added, given price data’s rele- tors on births, deaths, slaughters, marketing, etc.; vance to formulating economically sustainable investments. Over 83 percent stakeholders consider getting price data as 3. Animal health and disease; most important or important. 4. Livestock breeds; Respondents identified six core livestock domains, which are 5. Water for livestock; considered as most important or important by at least 80 percent of the sample. Beyond prices, these include data/ 6. Feed for livestock; indicators on animal health and disease; meat production; livestock population; feed; milk production; and consump- 7. Housing for livestock; tion of animal foods. Ranking in domains is similar across all groups of stakeholders. • QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations PART III. LIVESTOCK DATA FOR DECISION MAKING: EVIDENCE AND EXAMPLES   |  81 CORE LIVESTOCK DOMAINS / INDICATORS Animal health / disease Meat production Livestock population Feed for livestock Milk production Consumption of animal food Change in livestock stock (incl. marketing) Water for livestock Egg production Livestock breeds Labour devoted to livestock Production / use of dung Housing for livestock Hides & skins production Animal power (draught, transport, etc.) 0 0.25 0.5 0.75 1 Most important Important % OF RESPONDENTS QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations 82  |  Investing in the Livestock Sector: Why Good Numbers Matter AGRICULTURAL AND LIVESTOCK Table 10 lists the agricultural/livestock censuses and surveys implemented in West Africa since 2000.3 Since the year 2000, CENSUSES AND SURVEYS IN WEST agricultural/livestock censuses and surveys have been imple- AFRICA mented in 7 out of the 16 West African countries, including Burkina Faso, Cape Verde, The Gambia, Guinea, Ivory Coast, Two main methods are used in developing countries to collect Mali and Niger. At the same time, two countries plan to data on the number of animals and estimate livestock popu- annually undertake sample agricultural/livestock surveys, lations. These include, as detailed in chapter 1.4, agricultural notably Burkina Faso and The Gambia, though these surveys and/or livestock censuses and nationally representative are not administered with regularity. In virtually all cases, agricultural/sample surveys. Due to budget constraints, data collection was implemented on a sample basis. however, country governments often undertake agricultural and/or livestock censuses on a sample basis, which reduces the difference between censuses and surveys to the sample 3 Sources of information are the FAO World Census of Agriculture, both from 2000 and 2010, and the International Household Survey Network size — larger in the case of the census — and to the length of (IHSN), which maintains the most comprehensive catalogue of household the questionnaire — longer in the case of sample surveys. surveys undertaken in developing countries since the late 1800s. TABLE 10. AGRICULTURAL/LIVESTOCK CENSUSES IN WEST AFRICA: 2000–2012 Country Year Type of survey Sample size Livestock data collected between January 2008 and January 2009 from 7,500 Burkina Faso 2006/10 General Census of Agriculture households. Data were collected from May to July 2004. Complete enumeration of all holding Cape Verde 2004 General Census of Agriculture was carried out. Gambia 2002 Agricultural Census Data were collected from July to September 2002 from a sample of 666 dabadas.* Guinea 2000/01 Agricultural Census Data were collected from January to December 2001 on a sample basis. Data collected from January to August 2002; sampling method to collect informa- Ivory Coast 2001 National Census of Agriculture tion from stallholder farmers; large farms were fully enumerated. Data were collected from June 2004 to March 2005 from 10,000 smallholder Mali 2004/05 General Census of Agriculture farmers; modern holdings were fully enumerated. General Census of Agriculture Data on livestock were collected from a sample of 10,500 agro-pastoralists; water Niger 2005/07 and Livestock pointes were samples to count transhumant and nomadic livestock. Burkina Faso regularly Permanent Agricultural Survey In 2007, data were collected from 5,648 households, from July to December. National Agricultural Sample In 2005/06 data were collected from a sample of households between May 2005 Gambia regularly Survey and August 2006. * Group of persons who pool their agricultural resources together, usually headed by one person who takes management decisions. Sources: FAO, World Census of Agriculture 2000 and 2010 rounds (www.fao.org) and International Household Survey Network (www.ihsn.org) QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations PART III. LIVESTOCK DATA FOR DECISION MAKING: EVIDENCE AND EXAMPLES   |  83 Table 10 implies that estimates of livestock numbers in West (within the country) and external (cross-border trans- Africa countries are not updated regularly, nor are they nec- humance, usually towards Benin, Burkina and Nigeria). essarily reliable. In all cases, estimates are biased not only by Along the trekking routes there are permanent wells and non-sampling errors but also by sampling errors, because the ponds where livestock are taken to water. Enumerators, household — the ultimate sampling unit — might keep or positioned at a sample of water points, were responsible not keep animals. to directly count the animals and, to avoid double count- ing or omissions, they also issued a certificate of census to the livestock herder. Country case study: Niger ●● Counting nomadic livestock, whose movement is largely In 1974, the Niger Government, in an effort to increase unpredictable. However, given that animals are taken immunization coverage and improve livestock availability to water points regularly, these were used as sampling during vaccination campaigns, abolished the tax on livestock points. In particular, water points were classified in three and made vaccination free and compulsory. To identify layers — including bore holes, wells and surface water — vaccinated animals, part of the ear of each vaccinated cattle and a sample of 1,223 were selected to which enumerators was cut, which also allowed for a better estimation of live- were posted for three to five days to directly count the stock number in the country and facilitated the estimation animals. To avoid double counting, the livestock herder of yearly changes in herd structure. The veterinary services was issued a certificate of census. estimated that about 90 percent of cattle were vaccinated during any vaccination campaign conducted between 1974 Different questionnaires were drafted to collect information and 1994. This estimate presumably generated a fairly accu- on sedentary, transhumant and nomadic livestock, including rate overview of the animal population in the country. Since one specifically targeting camelids. 1995, however, with the withdrawal of the state in providing free vaccinations, the vaccination rate has dropped drastically from 90 to 12 percent, making it impossible to estimate cat- Country case study: Burkina Faso tle numbers using this method. The Government of Burkina Faso undertook the General In 2007/2008 the Government of Niger, assisted by the Census of Agriculture between 2006 and 2010. The previous international community, undertook the General Census of one was administered in 1993. The Census aimed to fully Agriculture and Livestock, which covered all eight regions, 36 measure agriculture; generate a sampling frame for subse- departments and the three communes of Niamey. This cov- quent agricultural surveys; and favor the establishment of a erage provided data at three levels of government (national, permanent agricultural statistical data collection system, also regional and district) including for three types of livestock targeting livestock. Data from the Census are expected to im- systems; i.e. sedentary, transhumance and nomadic livestock prove the quality of the Burkina Faso Agricultural Permanent (Republique du Niger, 2007b). Survey (Enquête Permanente Agricole, EPA), which produces estimates of the agricultural production on an annual basis, ●● Counting sedentary livestock. The sedentary livestock including forecasts by province and post-harvest estimates. census was conducted on the basis of a primary sample The ultimate objective of the EPA is to provide policy makers consisting of 700 enumeration areas (EAs), in which two with key information on the food security situation in the types of livestock keepers were identified: agro-pastoral- country. The first EPAs were implemented in the early 1990s ists and livestock-only producers. The latter were mainly and the survey still remains a major source of agricultural located in peri-urban areas. A sample of 15 households information for the country (MAHRH, 2009). in each EA were randomly selected, for a total of 10,500 households. Enumerators conducted face-to-face inter- ●● The EPA 2007/08 sample consisted of over 5,648 house- views to collect information on livestock. holds located in 706 villages in 45 provinces throughout the country. The number of villages selected in each prov- ●● Counting transhumant livestock, which are animals — ince was proportional to the population of the province mainly large and small ruminants — seasonally taken to at hand. Within each village, eight farm households were pastures following standard trekking routes, both internal randomly selected, independent of the size of the village. QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations 84  |  Investing in the Livestock Sector: Why Good Numbers Matter Data were collected by 706 enumerators, supervised by 72 livestock-related equipment owned by the households, local statisticians, 12 regional supervisors, and a coordi- such as animal-drawn carts. nation team at central level. ●● The results of the EPA are aggregated at provincial level ●● The EPA comprises a core fixed module, which is a ques- and published in an annual publication whose priority tionnaire focused on collecting basic information on a focus is more on agricultural production for food security regular basis on current and anticipated harvests for than on agricultural/livestock statistics per se. Even if major crops. It also includes rotational modules, which livestock statistics were to be generated using the EPA are implemented depending on the circumstances. These data, these might not be accurate, as seminomadic and modules target information on agricultural production; nomadic animals are not well accounted for in the survey. extension services; livestock populations; agricultural inputs; prices, etc. THE LIVESTOCK POPULATION IN ●● The 2007/08 livestock module of the EPA included 18 BETWEEN CENSUSES AND SURVEYS questions. Questions are asked on livestock ownership, by animal species and sex. Species included are cattle, One of the major constraints to generating accurate esti- sheep, goats, pigs, mules, horses, chicken and other mates of livestock populations in West Africa is the lack of animals, such as ducks and guinea fowl. Information is regularity in undertaking agricultural/livestock censuses then collected on change in stock over the last season due and surveys. This requires statistical authorities, and the to births, deaths, sale and other (e.g. given away as gift). Ministry responsible for livestock, to estimate the livestock The earnings from animal sales are quantified, including a population, based on most recent census/survey data, using question on their use. Finally, questions are asked about set rate increases for different animal species. Figure 11, elab- orated from Lesnoff et al. (2011), shows the basic parameters which are, in principle, needed to estimate with accuracy the changes in the livestock population, starting from the same FIGURE 11. ANIMAL LIFE CYCLE AND BASIC base year. DEMOGRAPHIC PARAMETERS There are three major methods that can be used to estimate Reproductive all, or part, of the above demographic parameters, and hence females estimate the livestock population in between censuses and surveys. These are the method of ‘tracking the herd’; the Abortion Parturition method of ‘follow the animals’; and retrospective surveys. rate Abortions Parturitions rate ●● Method of ‘tracking the herd.’ This is a simple form of monitoring, whereby over one or more years, investi- Prolificacy gators monitor change in a randomly selected sample Births rate of herds. Investigators regularly visit the herds (e.g. fortnightly or monthly) and document all critical changes in herd structure between two successive visits, including Stillbirth changes in calving, mortality, livestock use and any pur- Born alive Stillborn rate chases of new animals. ●● Method of ‘follow the animals’. This method targets the animals (not the herds) and is the reference method for Natural Survivals Offtake deaths demographic data collection in the tropics. An investiga- tor identifies all animals kept by a sample of households, Abortion Parturition most often using ear tags or microchip injections at the rate rate base of the neck. Investigators then visit the households QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations PART III. LIVESTOCK DATA FOR DECISION MAKING: EVIDENCE AND EXAMPLES   |  85 regularly and document all critical changes in key demo- Evidence graphic parameters, such as changes in calving, mortality, livestock use and any purchases of new animals. Country governments seldom make use of statistical meth- ods to estimate herd demographic parameters. First, the ●● Retrospective surveys are based on the memory recall of methods of ‘tracking the herd’ and ‘follow the animals’ are selected livestock raisers. Under this method, the enu- costly to implement on a regular basis. Second, retrospective merator’s role is to count the animals in the herd at the questions are infrequently included in survey questionnaires time of the survey and then to ask questions on all demo- and, when they are, they are rarely, if ever, analyzed to gener- graphic events (births, natural deaths, slaughtering, loans, ate the coefficients needed to model herd growth. In practice, purchases, etc.) that have occurred over the reference national governments simply apply some given growth rate period. Depending on the animals at hand, the reference to the livestock population, which is adjusted as new agricul- period might differ. This method is similar to the progeny tural census/survey data become available. history technique in which, with reference to each adult female animal sampled, the producer is asked how it en- Growth rates of the livestock population are, in the best cas- tered the herd, then about the offspring to which it gave es, derived from estimates of the livestock population at two birth. Information on the sex and disposition is solicited different points in times, such as two consecutive censuses. about each offspring in turn. Recall methods often lead When information on the livestock population is available to approximate results — particularly when questions are only for one year, information on growth rate is taken from asked on short-cycle animals and using a long recall peri- neighbouring countries and expert informants. In both cases, od — and, as such, country are always advised to regularly estimates of the livestock population are rarely accurate, undertake agricultural/livestock censuses and surveys. particularly when governments do not regularly update ©FAO/Pius Ekpei QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations 86  |  Investing in the Livestock Sector: Why Good Numbers Matter population estimates or review the elements influencing changes in the animal population, defined as those of over population growth rates. 10 percent on a year-to-year basis. These type of events occurred 15 times for large ruminants and 16 times for small Table 11 and 12 review year-to-year growth rates in the large ruminants. However, it should be emphasized that the ability ruminant and small ruminant numbers from 1990 to 2010 as of livestock professionals to estimate the livestock population obtained from FAOSTAT for all West African countries, with at the time ‘t +1’ remains one of the major challenges for the the exception of Liberia, Sierra Leone and Saint Helena. In statistical services in West Africa, even when relatively good the tables, two elements are highlighted. The light grey cells data are available. identify instances of three or longer-year period in which the large ruminant/small ruminant population was estimated to Overall, the two tables are illustrative of the weak capacity grow at exactly the same rate: this occurred in 13 instances in of governments in West Africa to regularly monitor changes the case of cattle, and 15 in the case of small ruminants. The in the livestock population. It is highly unlikely that between dark grey cells report instances of major positive or negative 1990 and 2003, the cattle population of Niger grew at a constant rate of 3.0 percent per year; or that the cattle pop- ulation of Guinea grew at 6.7 percent per year from 2000 to 2010. Similarly, it defies credibility that in Cape Verde the large ruminant stock increased by 23, 19, 16 and 16 percent in the four years spanning from 2004 to 2008. Some of the growth rates estimated for the small ruminant population seem likewise unreliable: in Nigeria the sheep and goat pop- ulation increased by 2.5 percent per year in every year from 2004 to 2009, and in Ghana at 4.2 percent per year from 2006 to 2010. In The Gambia, the small ruminant population is revealed to have increased by 43, 14 and 23 percent from 2000/01 to 2002/03, which would imply a doubling of the sheep and goat population over a four year period. ©FAO/Issouf Sanogo QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations PART III. LIVESTOCK DATA FOR DECISION MAKING: EVIDENCE AND EXAMPLES   |  87 TABLE 11. YEAR TO YEAR CATTLE POPULATION GROWTH RATE IN WEST AFRICAN COUNTRIES, 1990 TO 2010 90/91 91/92 92/93 93/94 94/95 95/96 96/97 97/98 98/99 99/00 00/01 01/02 02/03 03/04 04/05 05/06 06/07 07/08 08/09 09/10 00/11 Benin 0.7 4.9 -0.1 12.9 -15.5 19.6 3.5 1.9 4.9 7.1 3.8 2.5 2.5 2.5 2.4 2.7 2.6 2.8 2.4 2.6 2.6 Burkina Faso 4.7 4.7 4.7 4.7 4.7 4.7 4.7 4.7 4.7 -21.2 2.0 2.0 46.5 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0 Cape Verde -15.3 7.9 1.8 1.8 1.8 14.5 0.1 5.4 -1.8 -2.3 0.0 2.3 -0.8 2.2 23.6 19.0 16.2 16.1 1.7 2.2 1.1 Côte d’Ivoire 3.3 3.1 2.1 2.2 2.2 2.2 2.3 -2.7 2.2 0.0 2.2 2.2 2.0 2.0 2.0 2.0 2.0 2.0 2.3 0.5 0.1 Gambia 4.1 0.7 0.7 0.8 0.7 0.8 0.7 0.7 0.7 0.8 -11.2 1.0 21.3 3.0 0.5 0.7 0.5 1.2 2.9 -1.6 -6.2 Ghana 4.4 -2.9 0.8 1.6 2.5 2.6 1.0 1.0 1.2 1.1 1.0 1.1 1.1 1.1 1.0 -1.0 1.0 1.4 3.3 1.1 3.0 Guinea 8.4 8.4 8.4 8.4 8.4 5.2 5.2 5.2 5.2 6.7 5.5 5.5 5.5 5.5 5.5 5.5 5.5 5.5 5.5 5.5 -4.8 Guinea B. 0.0 2.5 2.5 2.5 2.5 2.5 2.5 2.5 2.5 2.5 0.6 0.0 1.0 0.0 1.9 3.8 4.4 4.4 3.5 3.5 1.3 Mali 1.9 0.7 1.0 1.2 1.4 1.7 2.0 2.2 2.5 2.8 3.1 3.3 3.7 4.0 4.3 4.6 5.5 10.1 3.0 3.0 3.0 Mauritania 3.7 -14.3 0.0 -8.3 1.0 1.0 20.6 3.0 5.8 3.0 3.0 -0.1 2.3 3.1 2.5 0.5 0.0 -2.7 1.4 0.1 1.2 Niger 6.0 6.0 6.0 6.0 6.0 6.0 6.0 6.0 6.0 6.0 6.0 6.0 6.0 6.1 5.9 6.0 6.0 6.0 6.0 6.0 -2.7 Nigeria 0.5 0.5 5.1 0.5 0.8 0.3 0.2 0.1 0.1 0.1 0.1 0.1 0.1 3.5 1.1 0.9 0.9 0.9 0.9 -2.6 17.8 Senegal 3.0 2.5 3.5 2.5 1.4 2.5 1.0 0.5 0.5 2.0 2.5 -2.1 0.7 0.7 1.7 1.5 0.8 1.5 1.6 1.6 1.0 Togo -2.1 -1.6 -1.5 -1.5 -10.9 7.4 24.9 0.7 2.5 -1.5 1.0 2.1 0.2 1.8 3.4 0.8 0.1 -0.1 1.7 0.6 0.6 TABLE 12. Y  EAR TO YEAR SHEEP/GOAT POPULATION GROWTH RATE IN WEST AFRICAN COUNTRIES, 1990 TO 2010 90/91 91/92 92/93 93/94 94/95 95/96 96/97 97/98 98/99 99/00 00/01 01/02 02/03 03/04 04/05 05/06 06/07 07/08 08/09 09/10 00/11 Benin 3.4 -9.1 -1.0 13.2 -3.0 4.9 1.6 0.7 5.5 4.3 2.0 3.0 2.4 2.8 1.0 2.9 1.7 4.6 -0.7 4.3 2.2 Burkina Faso 2.9 2.9 2.9 2.9 2.9 2.9 2.9 2.9 2.9 2.9 2.9 2.9 2.9 3.0 3.0 3.0 3.0 3.0 3.0 3.0 3.0 Cape Verde 13.0 8.2 8.2 -5.9 -15.0 -3.0 0.7 5.1 -3.1 -2.1 0.0 1.7 0.9 30.4 9.3 8.7 7.8 7.4 7.4 7.5 1.5 Côte d'Ivoire 2.3 2.5 2.5 2.6 2.5 2.5 2.5 1.6 2.0 0.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.2 1.3 0.5 Gambia 19.2 -3.9 -3.9 -3.9 -3.9 -3.9 -3.9 -3.9 -3.9 -3.9 42.7 14.0 22.7 -0.8 3.0 2.9 5.3 3.6 3.7 1.3 -8.1 Ghana 2.7 -1.7 1.6 1.4 -4.4 12.9 7.2 3.0 6.3 4.1 2.6 3.0 6.9 2.0 6.4 2.5 4.2 4.2 4.2 4.2 4.8 Guinea 5.0 5.1 5.2 14.0 7.3 6.4 6.4 6.4 6.4 7.8 6.7 6.7 6.7 6.7 6.7 6.7 6.7 6.7 6.7 6.7 -10.9 Guinea B. 3.3 7.5 5.0 2.9 1.9 2.7 2.7 2.6 2.5 -0.8 0.8 0.0 1.6 0.0 30.8 7.6 7.1 7.1 6.9 6.9 4.3 Mali -10.5 0.5 0.8 1.2 6.3 2.9 3.1 9.4 9.5 7.9 8.1 5.0 5.0 5.0 0.0 5.4 8.5 8.1 7.1 5.0 5.0 Mauritania 3.5 -3.4 3.5 0.0 0.2 17.2 1.6 8.5 10.2 4.5 4.5 4.5 0.5 0.3 0.0 0.0 0.0 -9.0 5.0 4.0 0.6 Niger 3.8 3.8 3.8 3.8 3.8 3.8 3.8 3.8 3.8 4.3 3.8 3.2 3.8 3.8 3.8 3.8 3.8 3.8 3.8 3.8 -5.4 Nigeria 2.0 2.7 4.0 9.0 8.2 7.6 10.1 8.3 8.5 7.0 8.0 2.5 2.4 2.4 2.5 2.5 2.5 2.5 2.5 4.6 1.4 Senegal 5.0 4.0 4.5 4.5 2.1 4.2 3.9 3.5 3.5 1.1 3.0 -2.7 1.7 2.1 2.8 2.8 2.2 2.8 2.6 3.5 0.7 Togo -19.6 -25.0 -9.4 -8.0 -20.6 46.9 23.2 7.9 8.0 8.1 1.8 3.6 3.5 -0.9 9.5 3.4 1.6 1.4 3.7 2.5 1.2 QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations 88  |  Investing in the Livestock Sector: Why Good Numbers Matter All of these recommendations, many of which have been proposed over the past two decades, make little sense if resources are limited or not available at all, which is often the case for countries in West Africa and other developing regions. A practical recommendation is therefore proposed for the National Statistical Authorities and the Ministry responsible for livestock to look at systematically integrating livestock data generated by existing nationally coordinated surveys. CONCLUSIONS livestock populations. Apart from not having an adequate baseline (nationally representative statistics on livestock numbers), countries have no frameworks for estimating Estimates of livestock numbers represent one of the most herd performance, e.g. the evolution of herds, because critical core indicators for stakeholders, both in the public of gaps in accurate and periodically monitored livestock and private sector. Indeed, accurate information on the population-related parameters. number of animals in the country are necessary for the Ministry responsible for livestock to formulate, imple- Several recommendations can be proposed to improve ment and monitor sector policies and for the National countries’ quantity and quality of data on livestock num- Statistical Authority to estimate livestock value added, a bers. These include the regular undertaking of agricultural key component of the GDP. At the same time, the private censuses with some sampling adjustments to reduce er- sector is interested in investment in the sector because rors when the objective is to estimate livestock numbers; demand for livestock products is anticipated to dramati- and the periodic implementation of specialized livestock cally increase on the continent in the coming decades. surveys, including in settled, semi-nomadic and nomadic areas, which require different survey tools. Additionally, A cursory review of how the livestock population is the routine data collection system — which includes the estimated in West African countries illustrates that data collected by government officials in their routine op- there are serious gaps. First, there are no countries in erations — could be enhanced, as proposed in chapter 2.4 the region which have regularly undertaken agricultural for Uganda. Better demographic parameters are needed to censuses over the past two decades. This is clearly the estimate changes in the livestock population starting from ‘gold standard’, namely the best option to estimate live- a base year; this could be facilitated through long term stock numbers. Furthermore, when agricultural censuses linkages between governments and research institutions are implemented, these are sample surveys which might which carry out animal based monitoring over several generate inaccurate statistics on the livestock population, years in selected areas. particularly when the distributions of animals and that of the farming population over the space are markedly All of these recommendations, many of which have been different. Second, according to available information, proposed over the past two decades, make little sense if only 2 out of 16 countries in West Africa plan to regularly resources are limited or not available at all, which is often undertake sample agricultural surveys which can also the case for countries in West Africa and other devel- be used to estimate livestock numbers. Finally, in the oping regions. A practical recommendation is therefore absence of a regular flow of livestock numbers data, gov- proposed for the National Statistical Authorities and ernments tend to apply a constant rate of growth that is the Ministry responsible for livestock to look at system- calibrated on a baseline year to update their estimates of atically integrating livestock data generated by existing QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations PART III. LIVESTOCK DATA FOR DECISION MAKING: EVIDENCE AND EXAMPLES   |  89 nationally coordinated surveys. The National Statistical ●● identify how and if the various surveys can generate Authority routinely undertakes a variety of surveys that useful information to estimate the livestock popula- often target agriculture and, within agriculture, livestock. tion, and on other key livestock-related variables; Examples include Household Budget Surveys and Living Standards Measurement Surveys which, as chapter 1.4 ●● attempt to improve the current estimates of the illustrates, also contain information on livestock. The livestock population using available data, while also National Statistical Authority also updates on a quarterly identifying low-cost options for improvements, such basis estimates of the gross domestic product, and the as adding or rephrasing a question in the survey livestock value added therein. Generating livestock value questionnaire; added necessitates information on livestock populations ●● establish consistency between the survey question- and its change over the previous quarter; on the level of naires, e.g. by ensuring that questions are formulated production and use of inputs. The Ministry responsible in the same way in different surveys; generating com- for livestock is the major livestock data stakeholder in the plementarity between different surveys, e.g. by using country, with significant incentives to access and utilize the same sampling unit; and other. available livestock-related data. The Ministry also collects livestock data in the course of its routine operations, e.g. It is believed that low-cost marginal changes in the when it implements a vaccination campaign. current system of agricultural data collection, if jointly supported by the National Statistical Authority and the It is recommended that the National Statistical Authority Ministry responsible for livestock, can on their own gen- and the Ministry responsible for livestock: erate improvements in the current livestock population ●● examine the questionnaires of all surveys undertaken estimates. That said, agricultural/livestock censuses and in the country over the last 15 years that include tar- surveys remain the first-best option to collect data to geted questions on farm animals; accurately estimate the livestock population. QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations 90  |  Investing in the Livestock Sector: Why Good Numbers Matter 3.2 PEOPLE AND LIVESTOCK: LIVELIHOOD ANALYSIS USING THE LIVESTOCK MODULE FOR INTEGRATED HOUSEHOLD SURVEYS KEY MESSAGES INTRODUCTION Livestock contribute in multiple ways to An absence of and inadequate data on the contribution of households’ livelihoods, including through livestock to national economies and to household livelihoods the provision of cash income, food, manure, contribute to the sector’s marginalization by policy makers. draft power and hauling services, savings and Even when data are available, these are often underutilized insurance, and social status. either because they are inaccessible; disseminated in an untimely fashion; unavailable in appropriate formats; or because they cannot be usefully linked to other data sources Living Standards Measurement Studies, that would deepen their analytical potential. A lack of especially those with a comprehensive module investment focused on improving the quantity and quality on livestock, are the best source of information of livestock statistics hampers the allocation of productive for quantifying the contribution of livestock resources towards the sector, which leaves its potential un- to household livelihoods, including both its tapped to reduce poverty and contribute to economic growth. monetary and non-monetary value. This chapter reveals that data collected through implemen- tation of the livestock module for multi-topic or integrated Accurate measures of livestock’s contribution to household surveys, presented in chapter 2.1, provide an households’ livelihoods are nevertheless difficult unprecedented opportunity to enhance understanding of to achieve, both because of the difficulties of livestock’s role in the household, in particular its contribu- tion to livelihoods. The livestock module for multi-topic, or properly measuring and valuing some inputs (e.g. integrated household surveys, consists of a set of livestock feed from road hedges) and some outputs (e.g. questions which can be included in the survey questionnaires draught power). of living standards measurement studies, typically adminis- tered to a nationally representative sample of households, as illustrated in chapter 1.4. Integrated household surveys cap- ture information on household characteristics and on a range of production and consumption activities. This generates a portrait of household characteristics and behavior and facili- tates an analysis of the relationships and causalities between livestock and livelihoods, as measured by different indicators, such as poverty, education, resilience, health and other (Davis et al., 2010: Zezza et al., 2009). The following sections illustrate how strategic indicators of key relevance to the sector can be derived through an analysis of the livestock module for integrated household surveys. ©FAO/Giulio Napolitano A review of these indicators improves our understanding of the role of livestock in the household economy and facilitates sector development through strategic interventions, either through policy or investment. First, appropriate measures of QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations PART III. LIVESTOCK DATA FOR DECISION MAKING: EVIDENCE AND EXAMPLES   |  91 livelihoods linked to livestock are identified; then categories status, therefore, are related to the value of the animal, a of livestock keepers and their husbandry practices are charac- question which is asked in the livestock module. terized by specific indicators; followed by a review of the role of gender in livestock keeping. The two final sections provide ●● Changes in the embedded value of the animals, as the some suggestions for data analysis and highlight the useful- module collects information on variances in the herd ness of this analysis in the conclusions. structure over the reference period. However, the data only allow capturing value changes associated to the mat- uration of animals (a heifer that becomes a cow) and not IMPROVED MEASURES OF weight gains/losses of each animal in the herd over the LIVELIHOODS reference period. A critical development issue is to properly measure the con- tribution of livestock to household livelihoods. Answering CATEGORIES OF LIVESTOCK KEEPERS this question gives an appreciation of how much the different The role of livestock in households and its contribution to types of households, including the poor, benefit from their poverty reduction needs to be reviewed within the context animals, and to what extent livestock represent a pathway of the households themselves; consequently categories of out of poverty for the less well-off. households have to be generated. Data from the livestock The contribution of livestock to household livelihoods cannot module embedded within integrated household surveys can be derived from traditional LSMS data. This is because survey be used to produce several indicators — such as income, questionnaires often do not include information on livestock expenditure or an asset-index — that allow differentiating inputs, but only ask questions on livestock outputs, thereby households by their livelihood level and clustering them in overestimating livestock income. They also do not collect different groups. Income and expenditure terciles/quintiles information on livestock by-products, such as manure, or the are often used to cluster households, but one can also non-monetary services provided by livestock, such as hauling differentiate households between poor and non-poor, with services and draught power, thereby underestimating the poverty defined according to national or international pov- contribution of livestock to household livelihoods (see chapter erty lines. In general, it is useful to generate a criterion (or 1.4). The newly developed livestock module for multi-topic a set of criteria) to categorize households into more or less household surveys includes detailed questions on assets, homogeneous groups (in some way akin to a typology) that inputs and outputs and is, thereby, anticipated to improve the can assist in looking beyond the indicators’ averages and into way the contribution of livestock to household livelihoods is the heterogeneity across households. The following are some assessed. In particular, the data can be used to measure: possible household typologies that can be generated using the available data: ●● The net recurrent household livestock-derived income for the reference period, which is the difference between ●● Livestock owners. These are defined as those households the value of livestock production and the value of inputs that own and raise their own animals, which is the most used for maintaining the animals. Outputs also include common situation in smallholder settled farming systems. non-monetary services, such as draught power and haul- ●● Livestock keepers. These are defined as those households ing services. Depending on the objective of the analysis, that own livestock and/or raise livestock on behalf of the value of food for self-consumption and the value of some other households. Indeed, there are circumstances family labor can be incorporated into the analysis. in which the manager of the herd is not necessarily the ●● The insurance, credit and social value of livestock, which owner of the animals. result from the potential of being able to sell the animals ●● Livestock managers. These are defined as those house- when there is a need (e.g. drought in case of insurance; holds that only keep animals on behalf of some other investment in case of credit; weddings in case of social households. This is, however, an uncommon practice. status). The benefits of insurance and/or credit and social QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations 92  |  Investing in the Livestock Sector: Why Good Numbers Matter Beyond differentiating households on livestock ownership, policy implementation can only be successful and have a e.g. whether they own/raise animals, the data can be used to development impact if the incentives provided correspond to generate categories based on herd and flock size (number of household priorities. large and small ruminants and number of birds) and on herd composition (sex and age of animals). To facilitate analysis, livestock numbers are aggregated, using a Livestock Unit INPUTS AND OUTPUTS (LU), which corresponds to an agreed upon live weight. In the tropics, the Tropical Livestock Unit (TLU), the equivalent Traditional agricultural surveys and living standards to 250 kg live weight, is used to standardize live animals by measurement studies include limited information on species mean live weight. LU conversions factors notably livestock-related inputs and outputs and usually target a have some drawbacks: they aggregate household animals by small number of households, with the consequence that the weights and not value, and therefore have limited market results are not nationally representative of the smallholder relevance; and they assume that there is little heterogeneity livestock sector. The implementation of the livestock module within animal species, disregarding differences in breed, sex, for multi-topic household surveys can partly fill this gap, as age and health status of animals. However, the approach it collects information on breeding practices, type of animal provides a convenient method for quantifying a wide range housing, feeding practices and water access, access to a of different livestock types and sizes in a standardized variety of animal health services — such as vaccination, de- manner, and it is widely used in the literature. To quantify worming and curative treatment — use on family and hired herd composition, some diversity index could be constructed, labor, and on major livestock products and by-products, such which takes into account the number and the composition of as meat, milk, manure and hauling services. species in the herd. ●● First, the data allow a broader perspective of households’ The livestock module data also allows the grouping of major husbandry practices, for example by calculating households according to their market-orientation, which is a the number and share of households that purchase feed, critical piece of information for the formulation of livestock maintain shelters for their animals, have access to veteri- sector policies and investment. Below, two possible ways of nary services, etc. grouping farmers according to these criteria are presented: ●● Second, the data facilitate a more detailed analysis of ●● Subsistence-oriented livestock farmers: these are house- household access to natural resources. For example, holds that do not regularly sell surplus meat/milk/egg information is collected on the main sources of water for production and, therefore, derive a marginal share of their animals: borehole, dam, well, river, spring, stream, con- agricultural/total income from livestock. structed water point, rainwater harvesting, and other; and on major feeding practices: only grazing, mainly grazing ●● Market-oriented livestock farmers or livestock with some feeding, mainly feeding with some grazing, and specializers. These are households that — contrary to sub- only feeding. sistence-oriented livestock farmers — regularly sell some surplus production and derive a large, if not the largest, ●● Third, the data allows for the quantification of some, share of their agricultural/total income from livestock. but not all, of the inputs used. For instance, the module includes questions on the quantity and value of the feed Finally, the livestock module also includes a question on the purchased; on the payment for different types of veteri- household rationale for owning/keeping animals, including nary services and the costs incurred for breeding animals. sale of adult/young animals; sale of livestock products; food for the family; a risk mechanism for coping with Documenting husbandry practices of individual households unexpected events (such as drought, crop failures, family is important, but the quantification of corresponding outputs emergencies); draught power; manure; transport; wealth assists in a better appreciation of potential development sup- status; savings; breeding, etc. The information generated port. The livestock module for multi-topic household surveys from this open question could be used to construct addi- generates information on: tional categories of households since targeted investments/ ●● The number and value of the live animals sold; QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations PART III. LIVESTOCK DATA FOR DECISION MAKING: EVIDENCE AND EXAMPLES   |  93 ●● The quantity of meat, milk, eggs and other major products livestock module. Information is requested from respondents generated by the household over the reference period; on where they sell their animals, in which kind of outlets (at the farm gate; at buyer’s house; on the road to market; ●● The quantity of livestock products sold and in small local markets or large markets; at the abattoir and self-consumed; other). In addition, they are questioned as to whom they sold their animals/livestock products (e.g. to relatives; local con- ●● The use and sale of animal dung and the use and sale of sumers; private traders; a marketing organization; butcher animal power, including for draught power and transport. or other). This information is useful in formulating policies, This information, complemented with data on inputs, poten- as it provides indications on the extent of livestock holders’ tially generates an empirically based and targeted estimate market integration and, hence, on their likely response to of the benefits derived by households keeping animals. These market-related policies. benefits are both monetary and non-monetary. While some, such as the value of livestock sales, are easily quantifiable, others, such as improved nutrition level due to increased in- WOMEN AND CHILDREN take of animal source foods by household members, or higher Gender division of labor in livestock systems varies according crop yields due to increased manure availability, are more to country, culture, religion and socio-economic variables. difficult to measure, but equally important for the livelihoods But women generally play an important role in the livestock of households. economy and in the household. This is revealed through The role of marketing and access to marketing channels for questions focused on the care and management or transfor- livelihoods can also be analyzed using data from the new mation and marketing of certain livestock products. There ©FAO/Giulio Napolitano QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations 94  |  Investing in the Livestock Sector: Why Good Numbers Matter is evidence, for instance, that both men and women harvest MOVING FROM DATA TO ANALYSIS and transport feed, chaff fodder, water, etc. In general, milking, cleaning of sheds and the processing and sale of The enhanced data available from the revised livestock mod- milk is mainly done by women. Children are also involved ule can be analyzed from a variety of perspectives, dependent in husbandry practices, such as in grazing animals, fetching on the interest of the user. However, the unique value of this feed and water, and milk collection and processing. Analysis improved data is to better estimate the contribution of live- of household data also confirms that boys and girls have stock to livelihoods, including household income; the implied different roles in tending livestock, with girls generally more ‘capital asset’ value of animals (including insurance, credit involved in general livestock care than in herding. and social value); and livestock production. Second, the data Available household datasets allow differentiating the house- can be used to generate a picture of the smallholder livestock hold on the basis of the gender of the household head (male/ farming system. In particular, livestock-keeping households female) and detailing household composition. The livestock could be grouped according to one or more criteria and typol- module presents an opportunity to deeper investigate the ogies of households established. Then the various dimensions role of women and children (and men) in livestock rising. of livestock ownership, husbandry practices and outputs can be reviewed to better understand whether they differ by ●● The section on ownership includes questions on who owns typology of livestock-keeping households. For instance, for and who keeps the various animals: respondents are asked each typology of household one can tabulate: to identify members of the household responsible for each task at hand, such as milking or selling animals. ●● Livestock ownership, i.e. herd size and composition; ●● In the section on water and feed, questions target the ●● Use of different livestock inputs, including quantities and responsibilities of the various household members for values, e.g. access to basic inputs and services, such as feeding, watering, and herding the animals. In the milk animal vaccination; production section, focus is placed on understanding the ●● Production level of different livestock products, including role of household members in milking the animals. The sales; module data should facilitate a rough quantification of the man-month devoted to different tasks. ●● Use of animal products, including for self-consumption and sale; ●● Finally, questions are asked on household decision mak- ing, in particular for selling animals/animal products and ●● Use of animal by-products, such as draught power and for using the earnings. hauling services. The additional detail provided by the data from the livestock Third, for the different typologies of households potential module can facilitate a better appreciation of the role of dif- correlations can be hypothesized and tested between house- ferent household members — and in particular women and hold-related and livestock-related variables. For example, children — in livestock farming and can also provide some comparisons can be made with non-livestock-keeping rough indications on the man-month/hour-day spent on households to determine whether livestock ownership could tending animals by different household members. This could influence other variables which have broader development presumably better inform investments which target labor implications. Examples include: saving technologies/innovations on a household level. ●● Gender of head of household and herd size/composition; ●● Household composition, including women and children, and herd composition, hypothesizing that women and children play a key role in livestock raising; QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations PART III. LIVESTOCK DATA FOR DECISION MAKING: EVIDENCE AND EXAMPLES   |  95 ●● Livestock ownership, by species, and land ownership, based on the assumption that keeping land facilitates access to feed for the animals; ●● Livestock ownership and credit access, contending that livestock can be used as collaterals for loans; ●● Livestock ownership and nutrition, assuming that house- holds keeping animals can have some direct access to the protein and micronutrient available in animal source foods; ●● Livestock ownership and children education/health condi- tions of family members, as animals are known as a source of cash in time of need; ●● Livestock ownership and access to market, positing that livestock are used as means of transport and surplus live- stock products cannot be easily stored. Finally, analysis of the data can be undertaken with the objective of identifying the causal relationships between dif- ferent variables. Data collected in the context of multi-topic household surveys are appropriate to better understand the determinants of household poverty and well-being. The data can also be used to investigate the determinants of livestock productivity. Examples of questions that the data can possi- bly answer are: ●● Do livestock significantly contribute to household livelihoods? ●● Which households are more likely to escape poverty from ©FAO/Giulio Napolitano investment in livestock-keeping? ●● What are the major determinants of livestock keeping? ●● Are there significant differences in livestock keeping be- tween male-headed and female-headed households? Given relatively small sample sizes, data from these surveys ●● Does household composition affect herd size and are not suitable for generating nationally representative sta- composition? tistics on certain indicators such as livestock herds. However, ●● Does livestock ownership/production contribute to food they allow an in-depth look at certain aspects of the impor- security through increased intake of animal protein? tance of the livestock within households and its contribution to rural livelihoods. It offers empirically derived insights into ●● Does livestock ownership facilitate access to formal/ smallholder livestock production systems. informal credit? QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations 96  |  Investing in the Livestock Sector: Why Good Numbers Matter BOX 8. LIVESTOCK AND LIVELIHOODS IN TANZANIA T he Tanzania National Panel Survey (NPS) is a unique, and as yet largely underutilized, source of knowledge and in- formation on rural Tanzania’s economy and living standards. 35 30 Share of livestock owned (right axis) Share of income from livestock (left axis) 90 80 % HOUSEHOLD INCOME FROM LIVESTOCK It is a nationally representative survey regularly conducted 70 % OF TOLAL LIVESTOCK HOLDINGS 25 by the National Bureau of Statistics (NBS). Consequently it 60 is much richer in data on the rural economy than previous 20 50 living standard surveys carried out in Tanzania, thus allowing a much more detailed snapshot of households compared to 15 40 what has been possible to date. Its first round, on which this 30 text-box is based, was carried out in 2008–09. Since then, 10 the survey has been implemented every two years (2010–11 20 and 2012–13). Analysis of the 2008–09 NPS shows that sixty 5 10 percent of rural households in Tanzania engage in livestock 0 0 keeping, earning an average of over 20 percent of their 1 2 3 4 5 income from livestock, while also benefitting from other QUINTILES OF LIVESTOCK OWNERSHIP livestock uses (e.g. traction, manure). In aggregate, large ruminants dominate, accounting for over 80 percent of total livestock holdings when measured in Tropical Livestock Units of relatively larger livestock owners who are substantially (TLUs). Cattle ownership is, however, less common and more different from the rest. This is confirmed by the fact that clearly linked to wealth than ownership of smaller livestock. households in the top quintile earn about a third of their in- Conversely, poor goat herders have flocks of similar size, or come from livestock, as opposed to 10–14 percent of income larger, than those of rich ones. Meanwhile, poultry ownership in the other quintiles. is very common place. From a household livelihood perspec- tive, the importance of poultry emerges clearly alongside Results show that women are relatively disadvantaged in that of cattle: the average livestock-keeping household terms of livestock ownership, particularly for cattle: this holds 44 percent of the total poultry birds in the country. In effect is strongest among poorer households. Where women particular, the poorest 40 percent of rural households rely es- do own livestock, they appear to be as market oriented as sentially on small numbers of poultry, with goats becoming are men, if not more so, due to their role in the marketing of more important among the somewhat better-off house- milk and milk products. holds, and cattle dominating among the richest 20 percent The NPS data allow going beyond livestock production to of rural households. look into patterns of consumption of products of animal One issue emerging from the analysis is the high degree of origin. The picture that emerges is one of substantial dis- concentration in livestock holdings, with the top 20 percent parities in livestock product consumption between rural and of livestock keepers holding over 80 percent of livestock as- urban areas and between different income groups. Overall, sets (as measured by animal numbers in TLU). one can argue that that as average incomes in Tanzania con- tinue to increase, the demand for livestock products on the Interestingly, levels of per capita expenditures do not change domestic market will expand, offering good opportunities significantly across quintiles of livestock ownership, whereas for livestock producers to increase incomes (Covaburrias et herd size and structure does, with a particularly steep gradi- al., 2012). • ent in the top quintile, suggesting that there is a small core QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations PART III. LIVESTOCK DATA FOR DECISION MAKING: EVIDENCE AND EXAMPLES   |  97 CONCLUSIONS and livelihoods-related variables; and to understand some of the determinants of livestock production and productivity. Living standards measurement surveys provide an up- to-date portrait of living standards and livelihoods in a To facilitate the availability and further analysis of basic country. Where they provide the most insights, however, livestock statistics, a livestock module has been developed is in their ability to move beyond national averages to and included in the ADePT software platform of the World focus on how households’ income sources, productive Bank4. This improved data availability will strengthen activities, access to basic services, market participation, analyses which identify the heterogeneity across house- access to assets, and a host of other socioeconomic vari- holds, thus moving beyond the broad brush stereotypes ables vary across households. When sufficient attention is which are often used to characterize the livestock sector. given to livestock at the survey design stage, such national It should, however, be noted that national household sur- data can be very useful for assessing livestock’s role in veys, being based on population sampling frames, usually household livelihoods. fail to capture the large-scale intensive sector, which in some countries or for some species can form a consider- Use of the livestock module for multi-topic household able portion of the sector. Depending on the sampling size surveys, details of which are presented in chapter 2.1, is and strategy of the survey utilized, it is also necessary to anticipated to produce a more complete understanding of recognize that specific populations groups, which may be smallholder livestock production systems. In particular, in small in number relative to the national population but the collected data, as illustrated in the Tanzania example, hold a considerable share of the national herds, may not will provide an unprecedented opportunity to appreciate be adequately represented in the sample. if and how livestock contribute to livelihoods; to critically review the husbandry practices of different categories of livestock keepers, the typologies of which can be refined 4 ADePT uses micro-level data from various types of surveys, including multi-topic household surveys, to develop publically available sets of based on different criteria; to undertake analysis of tables and graphs for a particular area of economic research. Livestock is the correlations between a variety of livestock-related now included as one of the data sets. QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations 98  |  Investing in the Livestock Sector: Why Good Numbers Matter 3.3 DATA INTEGRATION TO MEASURE LIVESTOCK AND LIVELIHOODS IN UGANDA KEY MESSAGES utilizing data generated from different datasets, is a cost-ef- fective way of ensuring data availability that feeds national There are no datasets which, on their own, data systems into more informed livestock sector policy and suffice to generate all necessary information investment decisions. for effective livestock sector policies and The Global Strategy to Improve Agricultural and Rural investments. Statistics (World Bank, 1011) recommends that countries, to achieve data integration, develop a unique master sample Integrating data from different surveys is frame for agriculture; design and implement an integrated an effective way to generate information on survey framework; and make results available in a common livestock, which goes beyond the indicators data management system. A unique master sample frame ensures that the statistical units (e.g. the farm; the house- produced using data from individual surveys. hold) are the same for all surveys, so that data targeting different items originating from different surveys can be Critical for effective data integration is a jointly analyzed. common master sample frame for agriculture This chapter presents the use of Small Area Estimation (SAE) and the implementation of an integrated survey techniques as an effective tool to integrate data from differ- framework. ent sources, and in particular to combine livestock-related information from sample surveys, censuses and other data sources. SAE techniques have, in the past, been mainly used Integrating data from the Uganda Livestock to generate food consumption-related maps at high level of Census and the Uganda National Panel Survey disaggregation. SAE, however, can be also applied to livestock allows estimating per capita livestock income and mapping to provide policy makers with reliable and spatial- the share of income from livestock at sub-county ly-detailed information on livestock and livelihoods, given level. that small area estimates of poverty are being increasingly used to target anti-poverty programs (see Hentschel et al., 2000; Alderman et al., 2001; Simler and Nhate, 2005 among others). Beyond policy-decision support, the results of this chapter demonstrate how integration of different data sets can greatly enhance spatial analysis. INTRODUCTION This chapter generates estimates of household income in Uganda from livestock activities (and its share of total in- Evidence-based policies and investment decisions that come) at low level of disaggregation by integrating data from support an efficient and equitable development of the live- the 2009/2010 Uganda National Panel Survey and the 2008 stock sector cannot be based on one only source of data. As Uganda National Livestock Census. Maps are generated that chapter 1.3 illustrates, there are several steps that lead to the provide a finer spatial disaggregation of statistics than that formulation of policies and investments and, in many circum- obtained through the use of survey data alone. The following stances, more than one data source should be simultaneously section presents the methodology and the data used; results used to improve the quantity and quality of information un- are then presented, followed by concluding remarks. derpinning any decision. Data integration, which consists in QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations PART III. LIVESTOCK DATA FOR DECISION MAKING: EVIDENCE AND EXAMPLES   |  99 “Through the integration of FIGURE 12. STAGES FOR INTEGRATING CENSUS AND SURVEY DATA USING SAE survey and census data, decision makers could benefit Selection of comparable variables from both the survey and the census determined by means, standard deviations, from the detailed information Stage 0 and frequency distributions at the national level in the survey and the large Estimation of the model using survey data, where sample size of the census the dependent variable of interest is missing in the census data Stage 1 to analyze variables at a higher spatial disaggregation than would be Parameter estimates from survey data are applied to the census data possible with the survey alone.” the average of the full set Y predicted values provides Stage 2 the point estimate of the dependent variable for the spatial subgroups METHOD AND DATA Surveys usually collect detailed information from a sample of households: the sample size is usually sufficient to provide ac- curate statistics for the country as a whole, or some regions, Two datasets are used for this analysis. The 2009/2010 but not to yield statistically reliable estimates at lower levels Uganda National Panel Survey (UNPS) collected information of disaggregation. At the same time, census data have a large on 2,975 households from 322 Enumeration Areas (EAs). By enough sample size to generate accurate statistics at low level sampling design, the survey is representative at national lev- of disaggregation, but only provide basic information on the el, plus the strata of (i) Kampala City, (ii) Other Urban Areas, (sampled) households. Through the integration of survey and (iii) Central Rural, (iv) Eastern Rural, (v) Western Rural, and census data, decision makers could benefit from the detailed (vi) Northern Rural. Data were collected in two visits, one information in the survey and the large sample size of the for each cropping season, over a twelve month period. For census to analyze variables at a higher spatial disaggregation the purpose of the analysis, the sample is narrowed to 2,375 than would be possible with the survey alone. households, as 45 households reported incomplete informa- tion and 555 households had moved, of which 521 are urban. The Small Area Estimation (SAE) techniques integrate data from censuses and household surveys with the objective of The other dataset incorporated in the analysis, the 2008 producing reliable estimates of priority indicators for small ar- Uganda National Livestock Census (UNLC), collected data eas where that information is not available. The methodology from 964,690 rural holdings in all 80 districts of the country underpinning the concept of SAE is relatively straightforward during a single visit during the month of February, 2008. The and, in the case of livestock, could be undertaken using the UNLC is not a full enumeration census but a sample-based following process. First, comparable livestock-related variables one, and is representative at the district level, which is the need to be selected from both the survey and the census in level of interest in the SAE. Given that the average sample terms of different statistical measures. The objective is to se- size at the sub-county level is adequately large (around 1,000 lect a variable around which other data from the two surveys households), results are also reported at this lower geograph- can be harmonized. Second, an estimation model is fitted in ic administrative level. Nonetheless, the limited amount of the survey data, where the dependent variable is missing in information collected in the 2008 UNLC is a constraint on the census. Third, the estimated parameters are used to pre- the number of explanatory variables in the estimation model dict the missing livestock-related information in the census (see chapter 1.4 for content of different survey types). data which are available at local level. The steps are outlined in Figure 12. The method is explained in greater technical detail in Elbers et al. (2003). QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations 100  |  Investing in the Livestock Sector: Why Good Numbers Matter The predictors used include: land size (separately by agricul- RESULTS tural, pasture, and other land); number of livestock heads by type (disaggregated by indigenous and exotic bulls, cows Three models are estimated on the 2009/10 UNPS and fitted. and calves, poultry, small ruminants); average weekly egg In the first model, the densities of large ruminants at the and milk production; age and gender of the household head; sub-county level are predicted and then compared to actual the use of household-hired agricultural labor; area covered values in the census. This model is used to test the reliability by each agro-ecological zone and the Normalized Difference of the prediction method used. In the second model, the Vegetation Index (NDVI)5 at the sub-county level. dependent variable is the log of per capita livestock income Figure 13 shows the comparison of the share of households (expressed in 2005 international Purchasing Power Parity rearing livestock by region in the survey and the census. dollars); and, finally, the third dependent variable is the Within each region, the prevalence of livestock owners is share of total household income from livestock. The latter not statistically significantly different between the census two models are the core of the analysis, since they estimate and the survey. The Figure also highlights the importance of dimensions (livestock income) not captured in the census but livestock, as the prevalence of livestock owners in Uganda collected in the survey. is relatively high in all regions, with a national average of One of the main results of the analysis is that, by virtue of around 70 percent. survey-to-census prediction, it is possible to derive higher spatially-disaggregated maps than using the survey alone. Figure 14 displays the actual densities (no. of livestock/ FIGURE 13.  UGANDA: PERCENTAGE OF HOUSE- square kilometer) of large ruminants from the survey and HOLDS OWNING LIVESTOCK BY REGION: census, as well as the predicted density into the census. Some 2009/10 NPS and 2008 UNLC (with 95% important elements emerge: confidence interval) ●● First, what from the survey appear to be homogeneous re- SHARE OF LIVESTOCK OWNERS gions, once disaggregated to the sub-county level through 8 the census, becomes a more detailed and scattered picture. ●● Second, the density range is wider in the census than in 6 the survey, as in the latter the distribution is composed SHARE OF HOUSEHOLDS of four values — one for each region — as averages of sub-county values within each region. 4 ●● Third, and foremost from a policy perspective, the census map is more meaningful for targeting purposes. 2 The first model also tests the reliability of the methods used in conducting this analysis. Figure 14 reveals that the actual 0 and the predicted densities of large ruminants from the cen- Central Eastern Northern Western sus is very close to the predicted one using the SAE method. survey census This result offers an insight as to how SAE can be a viable and reliable method to estimate spatial distribution of missing information through prediction. While the density of large ruminants in the census resembles 5 It is an indicator assessing whether the observed area contains live closely the distribution from the survey, the model fitted on green vegetation or not. Negative values of NDVI (values approaching -1) the log of per capita livestock income in purchasing power correspond to water. Values close to zero (-0.1 to 0.1) generally correspond parity is less able to predict missing information into the cen- to barren areas of rock, sand or snow. Lastly, low, positive values represent shrub and grassland (approximately 0.2 to 0.4), while high values indicate sus. Figure 14 shows maps from the survey and the census temperate and tropical rainforests (values approaching 1). for the estimated model. QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations PART III. LIVESTOCK DATA FOR DECISION MAKING: EVIDENCE AND EXAMPLES   |  101 FIGURE 14.  UGANDA: DENSITY OF LARGE RUMINANTS ACTUAL FROM SURVEY (LEFT), ACTUAL FROM CENSUS (RIGHT), AND PREDICTED FROM CENSUS (BELOW) AT REGIONAL AND DISTRICT LEVEL DENSITY OF LARGE RUMINANTS DENSITY OF LARGE RUMINANTS (CENSUS) 19 41 20 – 43 42 44 – 51 43 – 46 52 – 53 47 – 65 DENSITY OF LARGE RUMINANTS BY DISTRICT (CENSUS) DENSITY OF LARGE RUMINANTS (CENSUS PREDICTED) 3 5 4 – 30 6 – 30 31 – 43 31 – 43 44 – 68 44 – 68 69 – 124 69 – 124 0 20 40 80 120 160 Miles QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations 102  |  Investing in the Livestock Sector: Why Good Numbers Matter FIGURE 15. UGANDA: PER CAPITA LIVESTOCK INCOME ACTUAL FROM SURVEY AND PREDICTED TO CENSUS PER-CAPITA LIVESTOCK INCOME PPP (ACTUAL) 9.7 9.8 – 11.7 11.8 – 12.3 12.4 – 19.1 0 0.45 0.9 1.8 Miles PER-CAPITA LIVESTOCK INCOME (PREDICTED USING SAE) CENSUS (DISTRICT) CENSUS (SUB-COUNTY) 3.9 4.0 – 11.0 1.2 – 3.0 11.1 – 14.0 3.1 – 7.0 14.1 – 21.0 7.1 – 15.0 21.1 – 178.0 15.1 – 338.0 0 25 50 100 Miles QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations PART III. LIVESTOCK DATA FOR DECISION MAKING: EVIDENCE AND EXAMPLES   |  103 FIGURE 16.  UGANDA: SHARE OF INCOME FROM LIVESTOCK ACTUAL FROM SURVEY AND PREDICTED TO CENSUS SHARE OF INCOME FROM LIVESTOCK (ACTUAL) 0.02 0.03 0.04 0.05 0 0.45 0.9 1.8 Miles SHARE OF INCOME FROM LIVESTOCK (PREDICTED USING SAE) 0.01 -0.09 – 0.01 0.02 – 0.03 0.02 – 0.03 0.04 0.04 0.05 – 0.06 0.05 – 0.06 0.07 – 0.13 0.07 – 0.53 0 20 40 80 120 160 Miles QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations 104  |  Investing in the Livestock Sector: Why Good Numbers Matter Finally, the analysis of the predicted income share from live- argument that it is the lack of timely, reliable, and compre- stock at the sub-county level yields interesting results hensive survey and census data which are key constraints to (Figure 16). The predicted spatial distribution looks consis- effective policy formulation targeting local levels, more than tent regardless of the method used, and this reinforces the the need for advancement in spatial methodology. CONCLUSIONS livestock policy. Indeed, integration between different data sources allows for finer spatial resolution: regional distributions looking homogeneous based on survey data The integrated use of multiple data sources, such as alone masks very diverse sub-county distributions emerg- household surveys and censuses, satellite imagery and ing from the integrated use of survey and census data. administrative data, combined with spatial analysis techniques such as SAE and spatial allocation models, can The results are internally and externally consistent with provide reliable, coherent and location-specific insights to the literature, strengthening reliability. The novelty of guide policy and investment. Cross-validation across pri- the proposed approach is that it relies on micro-data and mary and secondary data sources provides clearer insights the census, which is particularly important for policy into livestock-related farmer decision making and, in so targeting, as it would greatly enhance the local relevance doing, provides a better springboard for effective pover- of policy interventions. In fact, there is the need to com- ty-reduction policy action. plement survey data with census information to provide more spatially-specific findings. As to external relevance By fitting accurate prediction models, there is the concrete and viability, this approach can be easily scaled-out to possibility of combining multi-topic household surveys other countries with similar statistical data systems. with specialized databases to estimate the contribution However, it is only when a common master frame for of livestock to household livelihoods. Among the various agriculture and an integrated survey framework are econometric models tested, the SAE technique has been established and implemented that the ultimate value used for targeting poverty programs in many countries of the SAE technique in providing information for evi- worldwide, and this chapter provides evidence that it dence-based policies and investments can be fully tapped. could represent a potentially useful tool for informing QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations PART III. LIVESTOCK DATA FOR DECISION MAKING: EVIDENCE AND EXAMPLES   |  105 3.4 COMPLEMENTING SURVEY DATA ON QUANTITY WITH QUALITATIVE INFORMATION: THE MARKET FOR ANIMAL-SOURCE FOODS IN TANZANIA AND UGANDA KEY MESSAGES quantity and value, they are insufficiently disaggregated to offer insight into consumers’ preferences for quality and The statistical system provides information safety attributes. Hence, there is little guidance available to on the quantitative dimension of the market smallholder producers, to supporting distribution and service for animal-source foods, which is one piece of providers, or to governments supporting market-driven the information needed to appreciate market smallholder and food security initiatives, on the potential opportunities for livestock producers. for local livestock product markets to deliver benefits to the producer. Ad hoc data collection exercises are needed to National data on livestock products are often aggregated into appreciate the qualitative dimensions of the such broad categories as ‘meat’ or ‘meat and fish’, ‘dairy’ and ‘eggs’. Consideration of product quality and differentiation, market for livestock products and better design which motivates value addition by producers and others in livestock sector policies and investments. the value chain, is generally absent. For livestock products in developing counties, few studies of consumers’ willingness Collecting qualitative information on preferred to pay for specific attributes are available, although Jabbar et al. (2010) provides an exception. At the levels of product retail forms, retail outlets and safety and quality assembly, distribution and retailing, little beyond anecdotal attributes is relatively straightforward and not information emerges. Data on product form, retail outlet expensive. type, urban and rural market differences, and characteriza- tion of consumers by income levels are little known, and this Data integration is essential to provide a national represents a barrier to the identification and service of high value markets. level picture of the qualitative dimensions of the market for animal-source foods. This chapter presents a method for generation, synthesis and basic analysis of data to inform decisions about the retail markets for livestock products in developing countries. The results, for which an illustrative set are presented here, INTRODUCTION Growing developing-country demand for livestock products “National data on livestock potentially provides commercial opportunities for smallhold- er producers and the supporting service and distribution products are often aggregated providers. Exploiting such potential requires identification into broad categories… and use of data on the nature of consumer demand and retail Consideration of product quality practice. and differentiation, which motivates Developing countries’ national statistical agencies’ data on value addition by producers consumption, and associated dietary monitoring, capture the broad commodity level. Although they provide generally good and others in the value chain, evidence of trends in consumption and production, including is generally absent.” QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations 106  |  Investing in the Livestock Sector: Why Good Numbers Matter generate information guide policies that might support are growing in demand, and the extent to which demand is market-led development of the livestock sector. The method sensitive to price and income changes. Nationally representa- is designed to be inexpensive to implement, and to provide tive consumption surveys, particularly where supplemented results rapidly. It can be used to support the implementation by price information, offer estimations of key consumer of Pillar 2 of the CAADP. response parameters such as income and price elasticity. Although these are mostly cross-sectional in nature, a na- tionally representative sample generally provides sufficient BOX 9. CAADP PILLAR 2: MARKET ACCESS variation in prices and income that inference may be drawn about consumption patterns over time, as these variables P illar 2 of the Comprehensive Africa Agriculture Devel- opment Programme aims at increasing market access through improved rural infrastructure and other trade-re- grow. Illustrative examples of use of this information are employed in this chapter for the purpose of identifying high value products, although the details of the method are not lated interventions. The objectives of Pillar 2 are to: (i) presented. accelerate growth in the agricultural sector by raising the capacities of private entrepreneurs (including commer- Field level data cial and smallholder farmers) to meet the increasingly complex quality and logistical requirements of markets, A major challenge is the absence of quality- and income-dis- focusing on selected agricultural commodities that offer aggregated data at relevant points in the value chain the potential to raise rural (on- and off-farm) incomes; (including the retail and consumer levels). A common (ii) create the required regulatory and policy framework approach, applied in this chapter, is the use of expert advice. that would facilitate the emergence of regional economic In what follows, an expert informant interview is employed spaces that spur the expansion of regional trade and effectively to bridge a gap between the nationally representa- cross-country investments. These two objectives are best tive aggregate data and the market level reality of assembly, achieved when the market for agricultural products are distribution and retailing of products that are disaggregated well characterized, both from a quantitative and qual- across numerous forms, quality levels and consumer types. itative perspective. While quantitative information on This procedure distils information on commodities into a current and projected consumption of livestock products guide on product form and retail format. Sampling proce- is largely available for the African continent, there is dures then address locations. limited information on consumers’ preferred retail forms, retail outlets and safety and quality attributes, which in Individual observations on consumers’ and retailers’ char- some circumstances could make it challenging to effec- acteristics, choices and practices are required for a robust tively implement Pillar 2 of the CAADP. • analysis of products’ potential for profitable smallholder de- livery. Unlike farm households, with which many researchers and government agencies are familiar, such targets for survey work require interview experiences that are brief, deliver DATA quantitative results, and do not encourage strategic respons- es from any market actor. Robust inference requires proper Official data available at national level sampling and adequate sample numbers. Training of enumer- Notwithstanding their aggregate nature, household surveys ators is required, both for standardized procedures and to and other data from official sources can be used in market equip them to assess selected variables that are unsuitable for analysis. They provide information on quantities consumed, survey questions. price and income across expenditure categories and locations. These offer insight into which products (at an aggregate level) QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations PART III. LIVESTOCK DATA FOR DECISION MAKING: EVIDENCE AND EXAMPLES   |  107 ©FAO/Pius Ekpei METHOD these included chicken, beef, goat meat, pork, milk and eggs. Applications of the method may better suit a narrower range of commodities, perhaps identified as above. Commodity selection — estimation from nationally representative survey data Product identification — expert informants’ interviews From analysis of nationally representative data, livestock Meetings of expert informants were convened to generate commodities are identified as featuring higher expenditures a ‘consumer product matrix’ for each of the commodities per unit of volume in response to increases in income. In identified from aggregate data. Note that a standard coding is essence, the commodities are identified for which consumers used for each type of retail outlet. For each commodity (Table have been shown to pay higher prices as their incomes rise. 13 is for beef), the matrix is composed of collated informa- For a given commodity, this approach requires the assump- tion on: tion that higher price is an indicator of higher quality. ●● The main products purchased by consumers, and their The example presented here features livestock products forms; in Uganda and Tanzania. To fully test the method, a large number of livestock commodities and products (see below for ●● The retail formats selling to consumers. disaggregation methods) were examined. At commodity level, QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations 108  |  Investing in the Livestock Sector: Why Good Numbers Matter TABLE 13. TANZANIA: EXAMPLE OF A CONSUMER PRODUCT MATRIX (BEEF) MAIN RETAIL PRODUCT FORMS RETAIL OUTLET TYPE 1 Bone in large piece 1 Abbattoir 2 Steak, cooking, frying or roasting piece 2 Road side butcheries 3 Ground beef 3 Food markets 4 Mixed beef 4 Supermarkets 5 Offal To guide subsequent field work (particularly sampling and Surveys conducted the planning of study logistics) expert informants were also called upon to list locations (both urban and rural) known to Two surveys were conducted: one each for consumers and feature retail outlets selling the products identified. Similarly, retailers. Consumer surveys were conducted in retail prem- for the subsequent training and informing of enumerators, ises. Enumerators observed consumers purchasing products, the products and retail outlet types were fully described, pho- and immediately following a purchase of livestock products, tographed and summarized as shown in Figures A and B. approached the consumer according to sampling practice (e.g. every third purchaser). Five brief questions were posed and the enumerator then observed and recorded quality of the products purchased. Retailer surveys similarly entailed a small number of brief questions and an observation on quali- ty by the enumerator. Sampling Sampling draws on the expert informants’ list of retail out- lets locations. The sampling strategy to be pursued depends on the purpose and emphasis of the study. Sample stratifi- cation by sex of customer, rural/urban location, and type of retail outlet are all reasonable approaches. Examination of products from several commodities requires a substantial number of visits to shops, as not all shops sell all products or all commodities. Experience in Tanzania and Uganda was that, within each of the categories of retail outlet, outlets in urban areas and outlets in rural areas were randomly selected, for a total of 36 and 42 outlets respectively. Retailers were interviewed and, in each retail outlet, a minimum of 12 consumers were ©Getty Images/iStockphoto randomly selected — i.e. those that were purchasing some livestock products when the enumerator was in the retail shop — and also interviewed, for a total of 144 Tanzanian and 160 Ugandan consumers. QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations PART III. LIVESTOCK DATA FOR DECISION MAKING: EVIDENCE AND EXAMPLES   |  109 Identification and assessment of products’ quality attributes TABLE 14.  GANDA: EXAMPLE OF A U PRODUCTION QUALITY SCORING Information about the quality attributes that are important TABLE (MILK) to developing country consumers of livestock products was drawn from the compilation of studies presented by Jabbar Attribute Score = 1 Score = 0 et al. (2010). Although such a list might also be compiled by Freshness yes no expert informants, it is recommended that objective research results be used. For each commodity a list of five quality Fat content low high attributes was selected. An alternative is to use the expert informants to identify the quality attributes, as is reported Origin/breed Known unknown in Jabbar et al. (2010) in several settings. However, a key fea- ture of the economic analysis of product attributes is that it Cleanliness of premises/ Clean unclean provides evidence of willingness to pay and hence is of more absence of flies commercial relevance than opinion as regards ‘what consti- Packaging Present absent tutes quality’. It should be noted that many of the attributes identified are, unsurprisingly, indicative of food safety and hygiene, and measurable variables such as fat content in milk, rather than of observed attributes like color and texture. Characterization of consumers Once a set of quality attributes had been established, a scor- The livestock product being purchased by each consumer was ing system for products was used which was subsequently observed and recorded by the enumerator. Consumers were employed to generate overall quality ratings for the products; characterized by sex and income group. An income proxy was for the retail outlets in which they were sold; and for the bun- employed, requiring the assumption that the means of trans- dle of purchases made by consumers. Scoring is an exercise to port owned or used is correlated with income levels. Hence be carried out by enumerators — not by survey respondents. consumer surveys featured yes/no questions about such own- The simplest form of scoring (1 and 0, or presence and ership and use, and results were compiled to generate income absence respectively) was used and overall quality ratings classes. For convenience, such analysis can feature 5 classes were constructed by adding the scores across attributes for (quintiles) which are consistent with many aggregate level products, retail outlets, consumer bundles, etc. An example analyses including household surveys. Other classifications, of quality attributes used in such scoring is presented as such as upper, lower and medium (terciles) are also avail- Table 14. able. Further characterization of consumers was achieved by asking retailers to assess their customers’ income class, particularly in relation to individual product forms, amounts “A key feature of the economic purchased, or quality levels. All these income assessments can be used across product forms purchased, retail formats, analysis of product attributes rural/urban locations, sex of customer, quantities purchased, is that it provides evidence and statements of future intent. of willingness to pay and hence is Statements by consumers of more commercial relevance Consumers were asked questions about their reasons for than opinion as regards shopping at a particular location for the product, patterns of ‘what constitutes quality’.” expenditure over time, and projections of purchases in the event of income increases (see Table 15). QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations 110  |  Investing in the Livestock Sector: Why Good Numbers Matter FIGURE 17.  DEMAND ANALYSIS: QUESTIONS FIGURE 18.  DEMAND ANALYSIS: ENUMERATOR TO CONSUMERS REGARDING OBSERVATIONS ON RETAIL PURCHASING BEHAVIOR PRODUCTION (BEEF) Statements by retailers Enumerators then posed questions to retailers on assessment of customers’ incomes, perceptions of market growth and potential at the product level, and constraints faced. Characterization of retailers Enumerators recorded retail outlets’ type (by code) and location, and their observations on products sold. They also assigned quality scores as described above. ©Getty Images/iStockphoto QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations PART III. LIVESTOCK DATA FOR DECISION MAKING: EVIDENCE AND EXAMPLES   |  111 FIGURE 19. DEMAND ANALYSIS: QUESTIONS POSED TO RETAILERS RESULTS ●● Clear patterns of preference for retail outlet appeared, and these were found to be sensitive to income (Figure 20). The studies cited as an example provided several important results: ●● Quality scores differed across products, but rural/urban differences in quality offered were not large (Figure 21). ●● Across all income levels, consumers purchased approx- imately the same quality. This indicates that very high ●● Consumer income was found to be a strong determinant quality such as seen in supermarkets faces rather limited of the product forms purchased (Figure 22). demand. This is in turn indicates that a large market exists for low and medium quality product supplied to tradition- al retail outlets. Smallholder producers are well-placed to deliver such products. QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations 112  |  Investing in the Livestock Sector: Why Good Numbers Matter FIGURE 20.  CONSUMERS’ RETAIL OUTLET FIGURE 22.  CONSUMERS’ PREFERENCES FOR PREFERENCES PRODUCT TYPE 100 BEEF 100 80 TYPE OF CONSUMERS (%) 80 60 TYPE OF CONSUMERS (%) 60 40 20 40 0 Roadside Small retail Wet market Supermarket Butchery Milk kiosk/ 20 shop vendor Less well off Middle class Better off 0 Offals Mixed beef Steak Sausage POULTRY FIGURE 21.  QUALITY SCORED, BY RETAIL 100 OUTLET TYPE 5.0 80 4.5 TYPE OF CONSUMERS (%) 4.0 3.5 60 3.0 2.5 40 2.0 1.5 1.0 20 0.5 0.0 0 Beef Chicken Eggs Pork Dairy Goat Mixed pieces Live bird Dressed bird Urban Rural Less well off Middle class Better off QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations PART III. LIVESTOCK DATA FOR DECISION MAKING: EVIDENCE AND EXAMPLES   |  113 CONCLUSIONS while offering a profile of these variables for both urban and rural locations. It is notable that the method is pri- marily based on actual purchases and sales, rather than This chapter offers practitioners a method for identifying hypothetical statements about preferences. These are and collecting commercial information in developing supplemented by statements by retailers and consumers country retail contexts. The method was developed to about future intentions. target business opportunities for smallholder livestock producers with the potential to serve vibrant retail The examples presented here depict a range of qualities, markets. A role is identified for official data sources, and a generally good level of quality, of animal-sourced particularly historical series, but the focus is on a robust products on sale. Across all apparent income levels, procedure for private sector operators interested in in- consumers opt for a variety of quality. However, income vestment in markets with potential growth. levels do influence the choice of retail outlet and form of product consumed. These results indicate substantial The example presented proceeds from undifferentiated opportunities for smallholder producers, and for those livestock products through to identification of shop and quality preferences for a range of consumer classes, involved in commercial distribution to retailers. ©Getty Images/iStockphoto QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations 114  |  Investing in the Livestock Sector: Why Good Numbers Matter TABLE 15. SELECTED EXAMPLE OF RETAIL PRODUCTS Livestock product Retail form and description Photograph Bone in Large piece This is usually a thigh and a portion of the ribs. Chops for roasting or frying These are usually small pieces of meat that are cut from the large piece and can easily be cooked without further cutting. The comprise of any part of the animal that is fleshy (e.g. ribs, muscles, bones and fats). Beef Ground beef This is usually the muscle that is minced in a machine. It may be lean or may contain some fats. Offals These are the intestines and gastro enteric parts of a bovine which are edible. QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations PART III. LIVESTOCK DATA FOR DECISION MAKING: EVIDENCE AND EXAMPLES   |  115 TABLE 16. UGANDA: DESCRIPTION OF RETAIL OUTLETS Retail outlet Description Photograph Abattoir A fairly large place where animals are slaughtered and hang in large pieces. These are small outlets which specialize in selling meat products. The operators Roadside butchery of such places usually purchase large pieces from abattoirs then sell smaller cuts to consumers. Roadside outlet These are sheltered or unsheltered places along roads which sell food products mainly to passersby. Wet market These are specialized markets which sell live animals (mainly small ruminants). QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations 116  |  Investing in the Livestock Sector: Why Good Numbers Matter 3.5 CONSTRAINTS: COMBINING MICRO-DATA WITH FARMERS’ VIEWS KEY MESSAGES a continuous range of levels of key variables, rather than a situation where access or use is constrained. Hence, policy The statistical system provides information on or investment indications inevitably focus on symptomatic the constraints affecting livestock keepers (e.g. issues such as low productivity, rather than addressing causal animal diseases) but not on the root causes of the mechanisms such as specific diseases or nutrition shortages. constraints (why animal diseases are rampant), Second, in most if not all circumstances, surveys undertaken which should be the target for policies and by the national statistical authorities are based on relatively investments. small sample sizes. The consequence is that detailed informa- tion on some features of specific livestock sub-sectors — such Ad hoc data collection is needed to identify the as on smallholder sheep fattening or dairy production sys- tems — cannot be represented. root causes of constraints, which depend on the main objectives for keeping animals and Third, it is widely known that policies and investments are ultimately originate from lack or inadequate effective when they are consistent with the goals and aspira- availability of land, capital, labor, and knowledge tions of the targeted beneficiaries. These are straightforward and information. in developed countries’ production systems, being few in number and generally of a commercial nature. However, in traditional production systems such as those found in Combining household surveys with farmers’ developing countries, livestock play a variety of roles in the perception of constraints is essential to identify household economy and so goals and aspirations are diverse priority areas for livestock sector policies and and often non-commercial. Policy and investment decisions, investments. therefore, are more effective if based on agricultural/livestock household survey data complemented with some ad hoc data collection and communication with farmers that identifies both the nature of the household and the role played by live- stock within it. INTRODUCTION This chapter presents a tested method for the identification Official data generated from agricultural/livestock household of the most important constraints faced by smallholder surveys are essential to portray the smallholder livestock livestock producers which should be tackled by policies and production system, as chapter 3.2 illustrates, including investments. The method employs a hybrid approach to data constraints that prevent farmers from deriving full benefits collection, for which a tested procedure is described. Piloting from their livestock. This type of information, however, while of the method was carried out in Tanzania and Uganda. In necessary for decision makers to identify priority areas of Tanzania, this was achieved in partnership with the Ministry interventions is, on its own, insufficient to guide investment of Livestock and Fisheries Development and local authorities decisions, for three major reasons. in four locations. In Uganda, the partnership was provided by the Ministry of Agriculture, Animal Industry and Fisheries First, a descriptive analysis of the household survey data and its extension and veterinary officers in two locations. helps identify some of the potential constraints on efficiency The method could be used to support the implementation of in production and sale of animals, such as animal disease. Pillar 3 of the CAADP. Commonly, multivariate analysis is then used in identifying some of the determinants of the constraints by exploring associations between key households’ and production sys- tems’ characteristics. Such analysis, however, usually assumes QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations PART III. LIVESTOCK DATA FOR DECISION MAKING: EVIDENCE AND EXAMPLES   |  117 empirically-important attribute of constraints is that they BOX 10.  CAADP PILLAR 3: FOOD SUPPLY are not easily observed, and consequently are often confused AND HUNGER with their symptoms (e.g. ‘low productivity’) that are associ- P ated with performance. Performance may itself be complex illar 3 of the Comprehensive Africa Agriculture De- to measure, as it (i) may represent satisfaction of just a few of velopment Programme (CAADP) aims to increase the multiple objectives of smallholder systems, and (ii) its im- food supply and reduce hunger by raising smallholder provement requires easing of a number of constraints which productivity and improving responses to food emergen- may be sequentially associated with reduced performance cies. The objectives of Pillar 3 are to: (i) improve domestic (e.g. profits are a consequence of productivity, price forma- production and marketing; (ii) facilitate regional trade in food staples; and (iii) build household productivity and tion, market access and value addition, amongst others). assets. In particular, Pillar 3 is a deliberate attempt to Clarification of the linkages between constraints and pro- ensure that the agricultural growth agenda targets the ductivity is offered by reference to ‘domains’ of management poor and the vulnerable directly, rather than through (Salami et al., 2010) which capture key livestock husbandry indirect and hoped-for trickled down effects. The impli- and production issues. These domains are consistent with cation is that investments under Pillar 3 should directly this Sourcebook’s approach to household questionnaires (see target smallholder farmers, with the objective to remove chapter 2.1). or ease constraints to their productivity. Available data, Farmers’ identification and ranking of constraints from a list however, chiefly provides information on the symptoms of pre-identified constraints has been used by Meganathan et of the constraints rather than on their root causes, the al. (2010) and Devendra (2007). In preference to pre-defined identification of which requires ad hoc data collection lists, Salami et al. (2010) opt for fundamental categories of and stakeholder involvement. • ‘long term’ constraints listed as land, labor, capital, knowl- edge and information, access to markets, and the policy EXPLORING CONTRAINTS environment. This is a list recognizable to students and practitioners of economics as it includes classical factors of production and emphasizes the enabling environment that is Increasing livestock productivity is critical to promote the stressed so much in recent development advocacy. development of the livestock sector, both at micro and macro level. This involves identifying and tackling the constraints In the presence of detailed farm level data, linear program- which prevent farmers from deriving benefits from their ming has often been applied to identify binding constraints animals and tapping into existing market opportunities. In (Siegel and Alwang, 2005; Jansen and Wilton, 1984). As the context of smallholder livestock production systems, a above, this approach also requires that potential constraining constraint can be defined as any barrier that prevents live- factors be pre-identified and appropriately incorporated into stock keepers from achieving their goal of improving their the programming. Econometric methods to estimate agricul- livelihoods. The livestock module for multi-topic and agri- tural supply responses, using both household and country cultural household surveys, for example, includes questions level data, have also been used to identify productivity-en- on a list of potential constraints affecting farmer’s livestock hancing or hindering factors: essentially via opportunities enterprise, such availability of water and feed for animals and constraints (e.g. Heltberg and Tarp, 2002). Data envelope (see chapters 2.1 and 3.2). Owing to smallholders’ many and analysis (DEA) that combines farm efficiency analysis with diverse goals, and equally diverse ways and means of meeting statistical identification of the factors associated with low them, constraint analysis also requires communication with performance, has also been used as a two-step approach uti- individual smallholders and other market actors as outlined lizing elements of the above methods (e.g. Gelan and Murithi above. 2012; Stokes et al., 2007). Constraints occur in many different forms, and can be Few methods, however, are available that attempt to combine classified in different ways. They range from bio-physical, quantitative analyses based on household survey data with resource and technical constraints to those associated ad hoc data in forms that are understandable to a range of with socio-cultural factors, infrastructure and policy. An audiences and easily usable by decision makers. The method QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations 118  |  Investing in the Livestock Sector: Why Good Numbers Matter presented in this chapter was designed according to these ●● Contributions of the group approach include the estab- considerations, and to cost concerns and avoidance of lishment of shared understanding, and development of complexity. It targets constraints to productivity and access ownership of the data generation and analysis process. to markets, building on both survey data and targeted data Use of ‘management domains’ (animal health, feeding, collection activities on a small scale. breeding and markets) allows both convenience in pack- aging constraints and critical mass amongst producer participants. Four management domains were employed A METHOD TO IDENTIFY to generate both discussion and individual data on the CONSTRAINTS symptoms (again, following Salami et al. (2010) and consistent with Sourcebook methods of household data Cost and logistic considerations require a pragmatic approach collection): to application of available existing data, and collection of new data in ways that maximize both participatory stakeholder ■■ Animal feeds input and rigor in sampling and collection. In this respect, ■■ Animal breeding the method described here is hybrid in nature, and opportu- ■■ Animal health nities exist for its adaptation. ■■ Markets and inputs Household level survey data: demand and supply ●● Group activities surrounding constraint analysis offers an opportunity for explanation and examination of the National level household survey data on consumption are difference between a ‘stated’ (or symptomatic) constraint used, via estimates of elasticity, to identify products for and an ‘underlying’ (basic, or long term) constraint. Many which there is high demand or (via panel data) rapidly-grow- ing demand. The main contribution of such analysis to an understanding of constraints is in the identification of the products to be pursued in the constraint analysis, i.e. it is expected that by removing those constraints to productivity and marketing, farmer’s livelihoods will improve. National level household survey data are also used to esti- mate the influence on productivity of key household and production systems’ characteristics. Such analysis (typically regression) provides basic guidance on identification of constraints to productivity, but has limitations as outlined above. A further problem with household level survey data is that, in many countries, survey observations on rural house- holds that feature relevant production systems are both few in number and difficult to identify because sampling does not usually address individual systems or constraint sets. Ad hoc data collection Targeted ad hoc data collection is thus recommended to better appreciate constraints to productivity and market access, which requires that, beyond analyzing nationally representa- tive household surveys data, producers themselves nominate ©FAO/Simon Maina and assign importance to the constraints they face. This can be achieved in two ways (group discussion and individual surveys) which are used in combination here. QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations PART III. LIVESTOCK DATA FOR DECISION MAKING: EVIDENCE AND EXAMPLES   |  119 participants harbor individual concerns, and indeed hopes Household survey data analysis for specific forms of assistance, that are expressed as ‘stat- ed’ constraints such as low milk yield or large numbers of Identification of commodities with such characteristics can deaths amongst young animals. The method developed draw on an analysis of the National Panel Survey data. This here collects such information, but also insists on its as- used consumption and expenditure data to identify the live- signment to underlying causes (such lack of animal feed at stock commodities featuring increasing expenditures per unit certain times of the year). ‘Underlying’ constraints are few of volume in response to increases in income. Hence, com- in number, and are readily comparable across sites and modities are identified for which consumers pay higher prices commodity systems. as incomes rise. This approach maintains the assumption that commodity price is an indicator of quality. The pilots ●● Individual household data generated by interviews offers also used the results of the demand analysis described in statistical inference. Importantly, producers’ individual chapter 3.4 of this Sourcebook, and aggregate national data responses may be classified according to factors (e.g. on patterns of consumption. These analyses allowed identi- enterprise size and specialization, locality, market served) fication of pork and dairy in Uganda, and dairy in Tanzania, that may be hypothesized to influence both identifica- as commodity sectors offering substantial opportunities to tion of constraints and the severity of their influence. smallholder producers. Household interviews characterize each producer’s production systems, and assembled data in relation to five Sampling ‘underlying’ or basic constraints as identified by Salami et A group of 30–50 producers are selected from a locality of in- al. (2010): terest. Primarily, such interest is centered on localities known ■■ Land to feature poverty amongst small-scale livestock producers. ■■ Labor Participants should be representative of critical social, eco- ■■ Capital nomic and geographic distributions. ■■ Information and knowledge The sample size enables critical levels of degrees of statistical ■■ Other (infrastructure, policies, institutions, markets) freedom. Randomness can be achieved by compilation of a ●● Individual data collection also presents the opportunity list of all farm households and ordered selection. Additional to identify individual households’ objectives or purposes guidelines (such as prohibiting multiple participants from in keeping livestock, better to interpret the impact of singe households) can be imposed, and experience in Uganda constraints. and Tanzania encourages this. Key sample strata include administrative zones, type of farm production system, degree of engagement in marketing and trading of inputs and live- IMPLEMENTATION stock products, gender, age, and ownership of local and/or improved breeds. Stratified sampling is to be superimposed The above method was implemented in both Uganda and on the randomization procedures, and in practice in Tanzania Tanzania, where a sample of 35 farmers took part to the exer- and Uganda this was achieved by way of information shared cise, assisted by 5–7 research and support staff. In particular, by local extension authorities. pursuant to objectives of the analysis, questionnaires were prepared for the guidance of discussion groups and individual Ad hoc data collection data collection. Identification of commodities can be either The day’s activities are laid out in a single questionnaire/ purposive (e.g. for those with an interest in a commodity) or guidelines document. The sequence is shown in Figure 22. a consequence of study design (e.g. for those with an interest The questionnaire/guideline document is displayed continu- in commodities with characteristics that need defining as ously during the sessions. part of the study). The pilot of the method which is reported here fell into the latter category, with interest directed at ●● A principle facilitator conducts all sessions, except constraints to producers of commodities for which demand is round-robin ‘cafes’ and focus group domain sessions. high and/or rapidly growing. QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations 120  |  Investing in the Livestock Sector: Why Good Numbers Matter FIGURE 23. FLOW CHART REPRESENTATION OF CONSTRAINT ANALYSIS METHODOLOGY BACKGROUND INFORMATION Geography Demography Socio-Economics, etc. SAMPLING INTRODUCTIONS Personal Introductions Opening Speeches About the LDIP Project Objectives of Constraint Analysis Exercise PERSONAL DATA Name Farming System Land & Water Information, Round Technology Robins Cafes Labour & Innovation Capital, Cash & Credit FEEDING BREEDING INDIVIDUAL FGDs — DOMAINS CONSTRAINTS CONSTRAINTS ANIMAL HEALTH RATINGs MARKETS AND VALUE CHAINS QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations PART III. LIVESTOCK DATA FOR DECISION MAKING: EVIDENCE AND EXAMPLES   |  121 ●● The participants attend all sessions, except the domain quality control variables and provides for discussion of the focus group discussions (see below). day. This also assists in adjustments to procedures for the following days’ work. ●● The ‘introductions’, ‘personal data’ and ‘farming systems’ sessions are conducted in a plenary style. The round Introductory sessions robin ‘cafes’ require separation (generally random, but see below) into four groups, each one involving a ‘café’ The plenary introductions session features both participatory basic constraint topic (land, labor, capital, knowledge and and individual sections. Basic information on size and nature information). of production systems is interspersed with derivation of local knowledge (see excerpts in Figure 24). A key (individual) ●● At the end of the round robin cafes, all participants will component is the identification and rankings of ‘main reason’ have completed all basic constraint sessions and complet- for keeping the animal species in question: this provides ed these sections of the questionnaire. much context for the examination of constraints. The milk marketing question in Figure 24 is an example of assessment ●● Following departure of the participants at the end of of individual conditions: specifically the presence of quality each day, an informal team meeting is held, chaired by incentives. the principal facilitator. This addresses and assesses key FIGURE 24. CONSTRAINT ANALYSIS: ELICITATION OF LOCAL KNOWLEDGE Group discussion of rainfall pattern Individual questions on milk marketing and quality premia (cattle, Tanzania) Identification of main reasons for keeping livestock species (cattle, Tanzania) QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations 122  |  Investing in the Livestock Sector: Why Good Numbers Matter Round robin cafes resources such as land and water (see example in Figure 25’s top left panel) and examination of how the resources are Round robin cafes (addressing land and water, labor, capital used (Figure 25’s right panel examines intra-household labor and information and knowledge) are individual data collec- allocation). Other examples in Figure 25 include the gender tion exercises, each of which focuses on a basic or underlying distribution of income from various sources and the use of constraint. Questions address both the quantification of credit. FIGURE 25. CONSTRAINT ANALYSIS: IDENTIFICATION OF UNDERLYING CONSTRAINTS Individual questions on land access (cattle, Tanzania) Individual questions on household labor use, and gender allocation of tasks (pigs, Uganda) Individual questions on receipt and control of income, and on use of credit (cattle, Tanzania) QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations PART III. LIVESTOCK DATA FOR DECISION MAKING: EVIDENCE AND EXAMPLES   |  123 Domain sessions and described according to their underlying basic constraint (land, labor, capital, knowledge and information, as well as Domain sessions provide the opportunity for groups to ‘other’). Prior to the specification of constraints, domain define key constraints. The management domains (feeds, sessions first compile sets of information about the produc- breeding, animal health and markets and inputs) provide a tion and marketing system that inform later analysis of the focus for discussion of constraints, and the use of self-se- individually-collected data. Examples in Figure 26 include lected groups encourages the concentration of expertise in identification of feed sources and systems, seasonal feed the appropriate domain. Each participant appears in just one availability (left panel) and basic epidemiological information domain discussion, at which constraints (limited to four from (right panel). each domain session) relevant to that domain are nominated FIGURE 26. CONSTRAINT ANALYSIS: EXCERPTS FROM DOMAIN SESSION CHECKLISTS Excerpt from “Feeds” domain session checklist (pigs, Uganda) Excerpt from “Animal Health” domain session checklist (cattle, Tanzania) Individual rating of constraints ●● Indicate his/her main purpose of keeping the livestock species in question (available from his/her response to the In the final plenary session, a representative of each domain main questionnaire); session’s focus group discussion summarizes the group’s work and presents and explains the selection of constraints ●● Rank, on the A4 page, the three most important con- and their attribution to basic constraints. At the conclusion straint/basic constraint combinations (by circling a cell on of these presentations, each participant is asked to do two the table on the A4 sheet). things with the A4 page (see example, Figure 19) listing the identified constraints: QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations 124  |  Investing in the Livestock Sector: Why Good Numbers Matter TABLE 17. EXAMPLE LIST OF NOMINATED CONSTRAINTS (MILK, WAKISO DISTRICT, UGANDA). KNOWLEDGE & CONSTRAINT SCORE LAND LABOUR CAPITAL INFORMATION OTHER MARK-INP Lack of access to high quality cows MARK-INP Lack of access to loans for expansion and increased productivity MARK-INP Slow growth of group action/co-operatives MARK-INP Lack of good technical help and service ANBREED Lack of knowledge in use and mixing of feeds, making silage ANBREED Poor quality and high cost of concentrated feeds ANBREED Lack of appropriate feed processing machines ANBREED Inadequate feed quantity (esp. in dry season) ANHEALTH High cost of drugs ANHEALTH Low level of husbandry ANHEALTH Poor veterinary services ANHEALTH Ineffective drugs FEED Lack of available replacement animals FEED Inefficient AI services (delivery and information) FEED Limited breeding-related information FEED Lack of communication with farmers for feedback and learning RESULTS FIGURE 27.  BASIC CONSTRAINTS IDENTIFIED IN Key results delivered from Tanzania and Uganda depict first, TANZANIA AND UGANDA the substantial difference in basic constraint identification BASIC CONSTRAINTS: TANZANIAN PRODUCERS % of producers identifying basic constraint 60 between the two countries (Figure 27). Land dominates the 50 lists of constraints in Tanzania, while capital and knowledge 40 do so in Uganda. 30 20 ●● Producers nominated a range of (‘stated’) constraints in 10 both countries (see Figure 28 for Tanzania). A notable 0 feature of the results is that the nominated constraints Capital Knowledge Labour Land Other dwell on resources (e.g. land, seasonal feed fluctuations, and information water). Land tenure (a policy consideration) is also iden- tified by many Tanzanian participants. In both Tanzania BASIC CONSTRAINTS: UGANDAN PRODUCERS % of producers identifying basic constraint 60 and Uganda, notable results included a general reluctance 50 to nominate animal health as a constraint, and the small 40 proportion of participants nominating soft infrastructure 30 such as market information and extension services. 20 10 0 Capital Knowledge Labour Land Other and information Appearing in top 3 constraints Identified as single most important constraint QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations PART III. LIVESTOCK DATA FOR DECISION MAKING: EVIDENCE AND EXAMPLES   |  125 ●● In both Uganda and Tanzania, cross-tabulation of produc- ers’ nominated constraints with the other information generated revealed: ■■ Locality is a strong determinant of constraints identified; ■■ Little evidence of linkages between main reasons for keeping the animals and the constraints identified; ■■ Stage of development of a household’s production and marketing system was a strong determinant of constraints identified; ■■ The type of knowledge and skills that producers’ saw as lacking were strongly related to the constraints they faced. FIGURE 28. TANZANIA: CONSTRAINTS NOMINATED BY PRODUCERS NOMINATED CONSTRAINTS: TANZANIAN PRODUCERS Poor roads, bridges and infrastructure Low incomes from product sales High costs of inputs and services Lack of information Lack of advisory services Lack of training or skills Land shortage or tenure insecurity Inappropriate breeds Difficulties in managing improved breeds Lack of good quality animals Lack of capital Poor or uncertain quality of veterinary drugs Animal disease Poor quality of feed Lack of feed Water shortage — quality and quantity Seasonal feed variation Lack of product storage Poor organization of marketing and input supply Long distance for product sales or input purchase Absence of product standards Absence of input providers or product buyers Poor product quality 0 5 10 15 20 25 % of producers nominating each constraint Appearing in top 3 constraints Identified as single most important constraint QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations 126  |  Investing in the Livestock Sector: Why Good Numbers Matter CONCLUSIONS The results obtained offer some important messages to agencies interested in the easing of constraints faced by smallholder livestock producers. First, smallholders’ basic This chapter puts forth a method for the identification, constraints are closely linked to resources (land and water, prioritization and explanation of the constraints faced but also capital) and the extent to which this applies is by smallholder livestock producers. The results of pilot dependent on locality. Second, little evidence suggests studies conducted in Tanzania (for dairy) and Uganda (for that smallholders’ objectives influence their definition pigs and dairy) are presented as examples, with a discus- of constraints. Hence, interventions to ease constraints sion of analysis and use. The method employs a hybrid, should target localities and production systems rather opportunistic approach to data collection, and is designed than management categories. However, a third result is to overcome several limitations of existing methods for that constraints (both nominated and basic) identified constraint analysis. Chief among these methodological are closely related to the stage of development of the advances is the demarcation between basic or underlying household with regard to size, productivity and market constraints, and nominated constraints which are symp- utilization. tomatic of the basic constraints. The method also allows for compilation of both forms of constraint. The constraint ‘knowledge and information’ occupied a surprisingly high ranking amongst basic and nominated The method is applicable across commodity sectors, and constraints in both pilot countries. The form taken by the several potential approaches to selection of commodity constraint was able to be linked both to commodity sector are identified. The pilot studies targeted high-growth live- and to stages of development of household production stock sectors, and so used a demand-related commodity and marketing. This provides substantial insight into selection mechanism. An improvement offered by the research and extension needs for smallholder-oriented method is that individual households’ intentions or pur- development. poses of keeping a species is fully recorded, and used in the definition and interpretation of constraints. QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations RECOMMENDATIONS | 127 RECOMMENDATIONS D ialogue and interaction with livestock policy makers and 7. Include livestock in Living Standards Measurement stakeholders in Africa have resulted in the following Surveys, which is essential to appreciate how livestock recommendations which, if promptly implemented, contribute to household livelihoods. would be the first steps in improving livestock data systems in Africa. 8. Implement, at regular intervals, different types of spe- cialized livestock surveys as recommended under the integrated survey framework of the Global Strategy to To National Governments: Improve Agricultural and Rural Statistics. The objective of these surveys address priority areas for investment to 1. Ensure dialogue between the Bureau of Statistics with increase livestock production and productivity. These other data stakeholders to integrate livestock data into surveys could target herd composition and dynamics, feed the National Statistical Plan, which would include design, availabilities, breeding, meat, milk and manure produc- financing and implementation of surveys with the aim of tion or other specific issues. generating adequate information on the sector. 9. Commit to undertaking ad hoc surveys through the 2. Provide for the adequate inclusion of livestock in the Ministry responsible for Livestock to generate critical live- integrated survey framework as recommended by the stock information when considering alternative policies Global Strategy to Improve Agricultural and Rural Statistics. and investments. This will guarantee that the different survey instruments jointly generate comprehensive and timely information on 10. Ensure that livestock is adequately represented in a na- livestock, provided that adequate financial resources are tional data platform for the dissemination of agricultural allocated for the implementation of the various surveys. data and statistics, and in so doing enable easy access to national and sub-regional survey results and other rele- 3. Adopt agreed-upon international standards and clas- vant data. sifications for the collection of livestock data and the generation of livestock statistics so as to ensure the generation of accurate statistics at country, regional and To Regional, Pan-African Institutions and the continental level. This harmonization should be discussed International Community: and agreed upon at the sub-regional level. 1. Encourage national governments to include animal health 4. Include animal health- and disease-related data among the and disease-related data among the core data on agricul- core data on agriculture identified by the Global Strategy to ture identified by the Global Strategy. Improve Agricultural and Rural Statistics. 2. Facilitate the adoption of common methodologies to 5. Update on a regular basis livestock technical conversion estimate technical conversion factors, so as to allow factors and reproductive parameters, the estimation of cross-country comparison of livestock data. which is critical to generate accurate livestock statistics, including livestock population, production levels and 3. Create a common data platform at the regional and livestock value added. pan-African level to follow and leverage the trends and dynamics of the livestock sector. 6. Improve the quality of administrative record livestock data, which are key for the Ministry responsible for ani- 4. Develop methodologies to improve the quantity and qual- mal resources to deliver public goods. ity of available livestock data, both from a statistical and institutional perspective. QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations 128  |  Investing in the Livestock Sector: Why Good Numbers Matter 5. Facilitate the sharing of best practices in survey design and implementation among African countries in order to adequately include livestock in the integrated survey framework recommended by the Global Strategy to Improve Agricultural and Rural Statistics. 6. Provide financial and technical assistance to countries to undertake ad hoc surveys to generate to generate critical livestock information when considering alternative poli- cies and investments. QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations REFERENCES | 129 REFERENCES Achoja F.O., P.C. Ike and P.O. Akporhuarcho (2010) Economics of Benin S., J. Thurlow, X. Diao, C. McCool and F. Simtowe (2008) Veterinary Services Delivery among Commercial Poultry Farmers Agricultural Growth and Investment Options for Poverty Reduction in in a Market-Driven Economy: Evidence from Delta State, Nigeria. Malawi. IFPRI Discussion Paper 794. Washington D.C.: IFPRI. International Journal of Poultry Science, 912:1140–1145. Bocoum I., S. Yahaya, A. Issa and A. Zezza (2013) L’élevage et les Ahuja V., D. Umali-Deininger and C. de Haan (2003) Market structure conditions de vie des ménages au Niger. Une analyse descriptive de and the demand for veterinary services in India. Agricultural Economics, l’enquête sur les conditions de vie des ménages et l’agriculture. Niamey: 29: 27–42. Ministère de l’Elevage et Institut National de la Statistique. Akter S., M.A. Jabbarand and S.K. Ehui (2003) Competitiveness and Bogale A., K. Hagedorn and B. Korf (2005) Determinants of poverty efficiency in poultry and pig production in Vietnam. Socio-economics in rural Ethiopia. Quarterly Journal of International Agriculture, 44: and Policy Research Working Paper 57. Nairobi: ILRI. 101–120. Alderman, H., M. Babita, G. Demombynes, N. Makhatha, and B. (2002). Chambers R.G. (1988) Applied Production Analysis. A Dual Approach. How Low Can You Go? Combining Census and Survey Data for Mapping Cambridge University Press: Cambridge, UK. Poverty in South Africa. Journal of African Economies, 11: 169–200. Costales A., N.T. Son, M.L. Lapar and M. Tiongco (2008) Determinants ALRMT, Arid Land Resource Management Project (2007) Livestock of Participation in Contract Farming in Pig Production in Northern Survey in the Arid Land Districts of Kenya. Nairobi: Otpimax for the Vietnam. Report 08–04, FAO Pro-poor Livestock Policy Initiative. Office of the President. Rome: FAO. Ashagidigbi W.M., S.A. Sulaimanandand and A. Adesiyan (2011) Covarrubias K., L. Nsiima and A. Zezza (2012) Livestock and livelihoods Technical and Allocative Efficiency of Poultry Egg Producers in Nigeria. in rural Tanzania. A descriptive analysis of the 2009 National Panel Agricultural Journal, 6: 124–130. Survey. A paper of the World Bank-FAO-ILRI Livestock Data Innovation in Africa Project. Rome: FAO; Washington D.C.: World Bank; Nairobi: AU-IBAR (2013) A Tool for the Inclusion of Livestock in the CAADP AU-IBAR and ILRI. Compacts and Investment Plans. Nairobi: AU-IBAR. CSA (2012) Agricultural Sample Survey 2011/12. Report on Livestock and Bahta S.T. and S. Bauer (2007) Analysis of the determinants of market Livestock Characteristics. Addis Ababa: Central Statistical Agency. participation within the South-African small-scale livestock sector. Paper presented at Tropentag 2011, Utilisation of diversity in land use CSO (2005) The Living Conditions Monitoring Survey IV Report. Lusaka: systems: Sustainable and organic approaches to meet human needs. Central Statistical Office. Witzenhausen. CSO (2008) Livestock Force Survey. Gaborone: Central Statistics Office. Bandiera O., R. Burgess, N.C. Das, S. Gulesci, I. Rasul, R. Shams and M. Sulaiman (2012) Asset Transfer Programme for the Ultra Poor: A Davis, B., P. Winter, G. Carletto, K. Covarroubias, E.J. Quiñones,E.J. Randomized Control Trial Evaluation. BRAC Challenging the Frontier A. Zezza, K. Stamoulis, C. Azzarri and S. Di Giuseppe (2010) A Cross- of Poverty Reduction/Targeting Ultra Poor (CFPR). Working Paper 22, Country Comparison of Rural Income Generating Activities. World Dhaka: BRAC. Development 38: 48–63. Bardasi, E., K. Beegle, A. Dillon and P. Serneels (2010) Do Labor Deininger, K., C. Carletto, S. Savastano, and J. Muwonge (2012) Can Statistics Depend on How and to Whom the Questions Are Asked? diaries help in improving agricultural production statistics? Evidence Results from a Survey Experiment in Tanzania. Policy Research Working from Uganda. Journal of Development Economics. 98: 42–50. Paper, No. 5192. Washington, D.C.: World Bank. Delgado, C., M. Rosegrant, H. Steinfeld, S. Ehui and C. Courbois (1999) Bashir M.K., S. Schilizzi and R. Pandit (2012) Livestock and Rural Livestock to 2020: the next food revolution. Food, Agriculture, and the Household Food Security: The Case of Small Farmers of the Punjab, Environment Discussion Paper 28. Washington, D.C.: IFPRI. Pakistan. Working Paper 1207.School of Agricultural and Resource Economics. Crawley: University of Western Australia, Australia. de Mel, S., D.J. McKenzie and C. Woodruff (2009) Measuring microenterprise profits: Must we ask how the sausage is made? Journal Beegle, K., C. Carletto and K. Himelein (2011) Reliability of Recall of Development Economics, 88: 19–31. in Agricultural Data. Policy Research Working Paper, No. 5671. Washington D.C.: World Bank. Devendra, C. (2007) Constraint analysis to improve integrated dairy production systems in developing countries: The importance of Beegle, K., J. De Weerdt, J. Friedman, and J. Gibson (2012) Methods of participatory rural appraisal. Tropical Animal Health and Production, household consumption measurement through surveys: Experimental 39:549–556. results from Tanzania. Journal of Development Economics, 98: 3–18. QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations 130  |  Investing in the Livestock Sector: Why Good Numbers Matter Dillon, A., E. Bardasi, K. Beegle, and P. Serneels (2012) Explaining Ishaq M., A. Farooq and U. Farooq (2007) Economies of scale in small variation in child labor statistics. Journal of Development Economics, 98: ruminants farming in Southern North West Frontier Province-Pakistan. 136–147. Livestock Research for Rural Development, 19(2). DPS (2004) Rapport de synthèse de la deuxième Enquête Sénégalaise Jabbar M.A., S.K. Ehui and R. Von Kaufmann (2002) Supply and Auprès des Ménages. Dakar: Direction de la Prévision et de la Statistique, Demand for Livestock Credit in Sub-Saharan Africa: Lessons for Institute Nationale de la Statistique. Designing New Credit Schemes. World Development, 30: 1029–1042. EC (2006) Rural Development 2007–2013. Handbook on Common Jabbar M.A., D. Baker and M.L. Fadiga (eds,) (2010) Demand for Monitoring and Evaluation Framework. Directorate General for livestock products in developing countries with a focus on quality and Agriculture and Rural Development. Bruxelles: European Commission. safety attributes: Evidence from Asia and Africa. Research Report No. 24. Nairobi: ILRI. Elbers, C., J.O. Lanjouw and P. Lanjouw, P. (2003) Micro-Level Estimation of Poverty and Inequality. Econometrica, 71:355–364. Jansen, G. B. and J. W. Wilton (1984) Linear programming in selection of livestock. Journal of Dairy Sciences 67:897–901. FAO (2000) Technical conversion factors for agricultural commodities. Rome: FAO. Joliffe, D. (2001) Measuring absolute and relative poverty: The sensitivity of estimated household consumption in survey design. FAO (2009) The State of Food and Agriculture 2009. Livestock in the Journal of Economic and Social Measurement, 27: 1–23. Balance. Rome: FAO. Kawagoe T., Y. Hayami and V.W. Ruttan (1985) The intercountry FAOSTAT (2013) faostat.fao.org accessed, 12 September 2013. agricultural production function and productivity differences among countries. Journal of Development Economics, 19:113–132. FAO and UNFPA (2012) Guidelines for Linking Population and Housing Censuses with Agricultural Censuses. Rome: FAO. LBS (2008) 2002/03 Household Budget Survey Report. Maseru: Lesotho Bureau of Statistics. Gelan A. and B. Muriithi (2012) Measuring and Explaining Economic Efficiency of Dairy Farms: A Case Study of Smallholder Farms in East LDIP (2010a) Collecting Livestock Data: What, Who, Who Pays? Brief. Africa. Agrekon: Agricultural Economics Research, Policy and Practice in FAO-World Bank-ILRI-AU-IBAR Livestock Data for Better Policies Southern Africa, 51: 53–74. Project. Rome: FAO; Washington D.C.: World Bank; Nairobi: AU-IBAR and ILRI. Grace D., T. Randolph, O. Diall and P.H. Clausen (2008) Training farmers in rational drug-use improves their management of cattle LDIP (2010b) Routine Livestock Data Collection in Uganda. Brief. FAO- trypanosomiasis: A cluster-randomised trial in south Mali. Preventive World Bank-ILRI-AU-IBAR Livestock Data for Better Policies Project. Veterinary Medicine, 83:83–97. Rome: FAO; Washington D.C.: World Bank; Nairobi: AU-IBAR and ILRI. GSS (2008) Ghana Living Standards Survey 5. 2005/2006. Household LDIP (2010c) Routine Livestock Data in sub-Saharan Africa: Collection Questionnaire, Part B. Accra: Ghana Statistical Service. and Use. Brief. FAO-World Bank-ILRI-AU-IBAR Livestock Data for Better Policies Project. Rome: FAO; Washington D.C.: World Bank; Heltberg, R. and F. Tarp (2002) Agricultural supply response and Nairobi: AU-IBAR and ILRI. poverty in Mozambique. Food Policy 27: 103–124. LDIP (2011a) Livestock Data: What Do Tanzania Stakeholders Say. Henning J., J. Morton, R. Pym, T. Hla and J. Meers (2009) Evaluation of Brief. FAO-World Bank-ILRI-AU-IBAR Livestock Data for Better Policies strategies to improve village chicken production-controlled field trials to Project. Rome: FAO; Washington D.C.: World Bank; Nairobi: AU-IBAR assess effects of Newcastle disease vaccination and altered chick rearing and ILRI. in Myanmar. Preventive Veterinary Medicine, 90: 17:30. LDIP (2011b) Livestock Data Survey of Participants in the 12th Hentschel, J., J.O. Lanjouw and P. Lanjouw (2000) Combining census Livestock Donor Group Meeting. Brief. FAO-World Bank-ILRI-AU-IBAR and survey data to trace the spatial dimensions of poverty. The World Livestock Data for Better Policies Project. Rome: FAO; Washington D.C.: Bank Economic Review, 14: 147–165. World Bank; Nairobi: AU-IBAR and ILRI. Hurley R. (1957) Livestock data problems in the Census of Agriculture. LDIP (2011c) Routine Livestock Data Collection in Tanzania. Brief. FAO- Journal of Farm Economics, 39: 1420–1426. World Bank-ILRI- U-IBAR Livestock Data for Better Policies Project. Rome: FAO; Washington D.C.: World Bank; Nairobi: AU-IBAR and ILRI. Iarossi G. (2006) The Power of Survey Design: A User’s Guide for Managing Surveys, Interpreting Results, and Influencing Respondents. Washington LDIP (2011d) Livestock Market Data: Collection, Dissemination and D.C.: World Bank. Use in Tanzania. Brief. FAO-World Bank-ILRI-AU-IBAR Livestock Data for Better Policies Project. Rome: FAO; Washington D.C.: World Bank; ILO (2004) Consumer Price Index Manual. Theory and Practice. Nairobi: AU-IBAR and ILRI. Geneva: ILO. QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations REFERENCES | 131 LDIP (2011e) Livestock Data: What Do Uganda Stakeholders Say. Brief. MLFD (2012) Data Collection to Estimate Livestock Technical FAO-World Bank-ILRI-AU-IBAR Livestock Data for Better Policies Conversion Factors in Tanzania. Proceedings of a Workshop held at Project. Rome: FAO; Washington D.C.: World Bank; Nairobi: AU-IBAR MLFD on 8 November 2012. Dar es Salaam: Ministry of Livestock and and ILRI. Fisheries Development (MLFD). LDIP (2012a) A Critical Indicator: Livestock’s Contribution to Gross MLFD and LDIP (2011) New Perspectives on Livestock Data. Domestic Product. Brief. FAO-World Bank-ILRI-AU-IBAR Livestock Data Proceedings of a workshop on Production, Marketing and Consumption for Better Policies Project. Rome: FAO; Washington D.C.: World Bank; Statistics. Dar Es Salaam: Tanzania Ministry of Livestock and Fisheries Nairobi: AU-IBAR and ILRI. Development and FAO-World Bank-ILRI-AU IBAR Livestock Data for Better Policies Project: Rome: FAO; Washington D.C.: World Bank; LDIP (2012b) Routine Livestock Data Collection in Zanzibar. Brief. FAO- Nairobi: AU-IBAR and ILRI. World Bank-ILRI-AU-IBAR Livestock Data for Better Policies Project. Rome: FAO; Washington D.C.: World Bank; Nairobi: AU-IBAR and ILRI. Murphy S.P., C. Gewa, L.J. Liang, M. Grillenberger, N.O. Bwiboz and C. G. Neuman (2003) School Snacks Containing Animal Source Foods Lesnoff, M., S. Messad, and X. Juanes. (2010) 12MO: A cross-sectional Improve Dietary Quality for Children in Rural Kenya. Journal of retrospective method for estimating livestock demographic parameters in Nutrition, 113:3950S–3056S. tropical small-holder farming systems. Montpellier: CIRAD. NBS (2010) General Household Survey-Panel. Agriculture Questionnaires — Lesnoff, M., , R. Lancelot, C.H. Moulin, S. Messad, X. Juanès, and C. Post-Planting Visit. Abuja: National Bureau of Statistics. Sahut (2011) Calculation of demographic parameters in tropical livestock herds – A discrete time approach with LASER animal-based monitoring data. NBS (2012a) National Panel Survey 2012/2013.Livestock & Fishery Montpellier: CIRAD. Questionnaire. Dar es Salaam: National Bureau of Statistics. LID (Livestock in Development) (1999) Livestock in Poverty-Focused NBS (2012b) 2012 Population and Housing Census. Household Development. Crewkerne: Livestock in Development questionnaire. Dar es Salaam: National Bureau of Statistics. Loos, T. and A. Zezza (2013) Collecting data on pastoralist livelihoods Nin A., S. Ehui and S. Benin (2007) Livestock Sector Productivity in in the context of LSMS-ISA surveys. Designing and piloting a module Developing Countries: An Assessment. In R. Evenson and P. Pingalu on pastoralist households. Livestock Data Innovation in Africa Project. (eds.) Handbook of Agricultural Economics. Amsterdam: Elsevier. Working Paper. FAO-World Bank-ILRI-AU-IBAR Livestock Data for Better Policies Project. Rome: FAO; Washington D.C.: World Bank; NISR, (2010) National Agricultural Survey 2008 – Report of National Data Nairobi: AU-IBAR and ILRI. Analysis. Kigali: National Institute of Statistics of Rwanda. MAAIF, Uganda Ministry of Agriculture, Animal Industry and Fisheries, Nouala S., U. Pica-Ciamarra, J. Otte and A. N’guetta (2011) Investing in Uganda Bureau of Statistics, FAO, ILRI and World Resources Institute Livestock to Drive Economic Growth in Africa: Rationales and Priorities. (2010) Mapping a Better Future: Spatial Analysis and Pro-Poor Livestock Bulletin of Animal Health and Production in Africa, 59: 383–391. Strategies in Uganda. Washington D.C.:World Resources Institute. NSO (2010) Integrated Household Panel Survey 2010. Agriculture MAAIF (no date) District Livestock Data Reporting Format. MAAIF, Questionnaire. Lilongwe: National Statistics Office. Entebbe. OiE (2011) Guidelines — FAO/OiE/WHO Annual Report — Year 2011. MAAIF and UBOS (2009) The National Livestock Census Report 2008. Paris: OiE. Entebbe: Uganda Ministry of Agriculture, Animal Industry and Fisheries (MAAIF) and Kampala: Uganda Bureau of Statistics (UBOS). Okello, P., F.K. Blum, U. Pica-Ciamarra, D. Baker and G. Simbila (2013) Administrative records for official statistics: food for thought MAHRH (2009) Méthodologie de l’Enquête Permanente sur l’Agriculture. on Uganda’s livestock data. Poster delivered at the VI International Ouagadougou: Ministère de l’Agriculture de l’Hydraulique et des Conference on Agricultural Statistics, Rio de Janeiro, October. Ressources Halieutiques du Burkina Faso. Otieno D.J., L. Hubbard and E. Ruto (2012) Determinants of Technical Meganathan, N., K.N. Selvakumar, M. Prabu, A.S.S. Pandian and G.S. Efficiency in Beef Production in Kenya. Paper presented at the Kumar (2010) Constraint analysis of tribal livestock farming in Tamil International Association of Agricultural Economists (IAAE) Triennial Nadu, Tamilnadu. J. Veterinary & Animal Sciences 6: 12–18. Conference, Foz do Iguacu, August. MLFD (2010a) Medium-term Expenditure Framework 2010–11 — Otte, J., Costales, A., Djikman, J., Pica-Ciamarra, U., Robinson, T., Ahuja, 2012–13. Vote 99. Dar es Salaam: Ministry of Livestock and Fisheries V., Ly, C. and Roland-Holst, D. (2012) Livestock Sector Development for Development. Poverty Reduction. An Economic and Policy Perspective. Rome: FAO. MLFD (2010b) Livestock Data in Tanzania. Unpublished document. Dar Perry B. and K. Sones (2009) Improving the quality and availability of es Salaam: Ministry of Livestock and Fisheries Development. data to support livestock for development initiatives in sub Saharan Africa: Strategic Recommendations to the Bill and Melinda Gates QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations 132  |  Investing in the Livestock Sector: Why Good Numbers Matter Foundation. Background document of the Livestock in Africa: Improving Sheppard, D., T. Stewart, C. Rothhaas, F. Lestina, E. Compton, J. Data for Better Policies. Machowski and D. Smith (2013) 2010 Census Administrative Records Use for Coverage Problems Evaluation. Washington D.C. United States Census Pica G., U. Pica-Ciamarra and J. Otte (2008) The Livestock Sector in the Bureau. World Development Report 2008. Re-assessing the Policy Priorities. PPLPI Research Report 08–07. Rome: FAO. Siegel, P. B. and J. Alwang (2005) Poverty reducing potential of smallholder agriculture in Zambia: opportunities and constraints. World Pica-Ciamarra U., J. Otte and C. Martini (2010) Livestock sector policies Bank Working Paper Series No. 85. Washington D.C.: World Bank. and programmes in developing countries. A Menu for practitioners. Rome: FAO. Simler, K. and V. Nhate (2005) Poverty, Inequality, and Geographic Targeting: Evidence from Small-Area Estimates in Mozambique. FCND Pica-Ciamarra U. and D. Baker (2011) Supply and Demand of Livestock Discussion Paper 192. Washington D.C.: World Bank. Data in sub-Saharan Africa: Evidence from a Stakeholder Survey. African Statistical Newsletter, 5: 32–35. Somda J., K. Keita, M. Kamuanga and M.B. Diallo (2004) Diagnostic des systèmes d’élevage peri-urbain en Moyenne Guinée: Analyse socio- Pica-Ciamarra, U., N. Morgan and D. Baker (2012) Core Livestock Data économique des exploitations en production laitière dans la commune and Indicators: Results of a Stakeholder Survey. Working paper of urbaine de Labé Socio-economic. Working Paper No 3. Banjul: ITC the FAO-World Bank-ILRI-AU IBAR Livestock Data for Better Policies (International Trypanotolerance Centre). Project: Rome: FAO; Washington D.C.: World Bank; Nairobi: AU-IBAR and ILRI. Stokes, J.R., P.R. Tozer and J. Hyde (2007) Identifying efficient dairy producers using data envelopment analysis. Journal of Dairy Science, 90: Pica-Ciamarra U., D. Baker, N. Morgan, C. Ly and S. Nouala (2013) 2555–2562. Investing in African Livestock: Business Opportunities in 2030–2050. Paper presented at the African Livestock Conference and Exhibition, Thompson, S.K. (2012) Sampling. Chichester: Wiley. Nairobi, June. UBOS (2009) The Uganda National Household Survey 2009/2010. Socio- Presser, S., M.P. Couper, J.T. Lessler, E. Martin, J. Martin, J.M. Rothgeb economic Survey Questionnaire. Kampala: Uganda Bureau of Statistics. and E. Singer (2004a) Methods for Testing and Evaluating Survey Questions. Public Opinion Quarterly. 68: 109–130. UBOS (2011) The Uganda National Panel Survey 2010/2011. Agriculture Questionnaire. Kampala: Uganda Bureau of Statistics Presser, S., M.P. Couper, J.T. Lessler, E. Martin, J. Martin, J.M. Rothgeb and E. Singer (eds.) (2004b) Methods for Testing and Evaluating Survey UNDP (2009) Handbook on Planning, Monitoring and Evaluation for Questions. Chichester: Wiley. Development Results. New York: UNDP. République du Mali (2007) Recensement General de l’Agriculture (RGA) URT (2010) National Sample Census of Agriculture 2007/2008. Preliminary – Campagne Agricole 2004–2005. Principaux Resultats Definitifs, Vol. Report. Dar es Salaam: National Bureau of Statistics. 1. Bamako: Ministère de l’Agriculture, Ministère de l’Elevage e de la Pèche, Ministère du Plan et de l’Aménagement du Territoire. WAHIS (2013) World Animal Health Information System. www.oie.int/ wahis_2/public/index.php/home République du Niger (2007a) Enquête Nationale sur le Budget et la Consommation des Menages 2007.Questionnaire Ménage. Niamey: Zezza A., P. Winters, B. Davis, G.. Carletto, K. Covarrubias, L. and Ministère de l’Economie et des Finances et Institut National de la E. Quiñones (2011) Rural Household Access to Assets and Markets: Statistique. A Cross-Country Comparison. The European Journal of Development Research, 23: 569–597. République du Niger (2007b) Recensement General de l’Agriculture e du Cheptel. Volume II. Resultats Definitifs (Volet Chepte). Niamey: Ministère Zezza A., G. Federighi and G. Carletto (2013) Milking the data: de l’Economie et des Finances et Institut National de la Statistique. Measuring household milk production in extensive livestock systems. Experimental evidence from Niger. Paper presented at the 59th ISI République du Niger (2010) Enquête Nationale sur les Conditions de World Statistics Congress. Hong Kong, August. Vie des Ménages 2010–2011. Questionnaire sur l’Agriculture et l’Elevage. Deuxième Passage. Niamey: Ministère de l’Economie et des Finances et Institut National de la Statistique. Salami, A., A. B. Kamara and Z. Brixiova (2010) Smallholder agriculture in East Africa: trends, constraints and opportunities. African Development Bank Group Working Paper Series No. 105. Tunis: African Development Bank. Scholtz M.M., J. Bester, J.M. Mamabolo and K.A. Ramsay (2008) Results of the national cattle survey undertaken in South Africa, with emphasis on beef. Applied Animal Husbandry and Rural Development, 1:1–9. QUICK JUMP TO • Contents • Part II • Introduction • Part III • Part I • Recommendations ©FAO/Simon Maina Investing in the Livestock Sector: Why Good Numbers Matter is a sourcebook for decision makers on how to improve livestock data. Funded by the Gates Foundation, it is the output of a joint collaborative initiative, drawing together the World Bank, the FAO, ILRI, AU-IBAR and the national governments of Uganda, Tanzania and Niger. WORLD BANK REPORT NUMBER 85732-GLB