56719 ECONOMIC AND SECTOR WORK GLOBAL STRATEGY TO IMPROVE AGRICULTURAL AND RURAL STATISTICS REPORT NUMBER 56719-GLB SEPTEMBER 2010 United Nations ECONOMIC AND SECTOR WORK GLOBAL STRATEGY TO IMPROVE AGRICULTURAL AND RURAL STATISTICS R epor t No. 56719-GLB United Nations © 2011 The International Bank for Reconstruction and Development / The World Bank 1818 H Street, NW Washington, DC 20433 Telephone 202-473-1000 Internet www.worldbank.org/rural E-mail ard@worldbank.org All rights reserved. This volume is a product of the staff of the International Bank for Reconstruction and Development / The World Bank. The �ndings, interpretations, and conclusions expressed in this paper do not necessarily reflect the views of the Executive Directors of The World Bank 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. Rights and Permissions The material in this publication is copyrighted. Copying and/or transmitting portions or all of this work without permission may be a violation of applicable law. The International Bank for Reconstruction and Development / The World Bank encourages dissemination of its work and will normally grant permission to reproduce portions of the work promptly. For permission to photocopy or reprint any part of this work, please send a request with complete information to the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, USA, telephone 978-750-8400, fax 978-750-4470, http://www.copyright.com/. All other queries on rights and licenses, including subsidiary rights, should be addressed to the Of�ce of the Publisher, The World Bank, 1818 H Street NW, Washington, DC 20433, USA, fax 202-522-2422, e-mail pubrights@worldbank.org. Cover Photo: Eric C.M. Fernandes. A C R O N Y M S AND ABBRE VIAT IONS III ACRONYMS AND ABBREVIATIONS AQUASTAT FAO Information System on Water and Aquaculture COFOG Classi�cation of Functions of Government CPC Central Product Classi�cation Eurostat Statistical Of�ce of the European Union FAO Food and Agricultural Organization of the United Nations GDP Gross Domestic Product GPS Global Positioning System ICAS-V Fifth International Conference on Agricultural Statistics IDA International Development Association IMF International Monetary Fund ISI International Statistical Institute ISIC International Standard Industrial Classi�cation of Economic Activities LCCS Land Cover Classi�cation System MDG Millennium Development Goals MPPS Multiple Probability Proportional to Size NASS National Agricultural Statistics Service NSDS National Strategy for the Development of Statistics OECD Organization of Economic Cooperation and Development PARIS21 Partnership in Statistics for Development in the 21st Century PDA Personal Digital Assistant PPPs Purchasing Power Parities SEEA System of Integrated Environmental and Economic Accounting SNA System of National Accounts TFSCB Trust Fund for Statistical Capacity Building UNEP United Nations Environment Programme UNSC United Nations Statistical Commission UNSD United Nations Statistics Division USDA United States Department of Agriculture E C O N O M IC AND S E CT OR WORK C O N T E N TS V CONTENTS Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix Executive Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xi Chapter 1: Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 The Availability and Quality of Agricultural Statistics Has Declined, Just at the Wrong time . . . . . . . . . . . 1 Filling the Void: The Global Strategy to Improve Agricultural and Rural Statistics . . . . . . . . . . . . . . . . 3 Chapter 2: A Conceptual Framework for the Collection of Agricultural Statistics . . . . . . . . . . 5 Dimensions of Data Demand . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 The Conceptual Framework . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 The Economic Dimension . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 The Social Dimension . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 The Environmental Dimension . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 Agricultural Statistics: Scope and Coverage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 Chapter 3: The First Pillar—Identifying a Minimum Set of Core Data and Determining National Priorities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 Set of Core Items and Associated Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 Determining National Priorities: Content, Scope, and Frequency . . . . . . . . . . . . . . . . . . . . . . . . . 15 Chapter 4: The Second Pillar—Integrating Agriculture into National Statistical Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 Framework to Develop a Master Sample Frame for Agriculture . . . . . . . . . . . . . . . . . . . . . . . . . . 20 Vision for the Integrated Survey Framework . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 Steps to Implement an Integrated Survey Framework . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24 The Data Management System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26 Chapter 5: The Third Pillar—The Sustainability of Agricultural Statistics through Governance and Statistical Capacity Building . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27 Governance. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27 Statistical Capacity Building . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28 Chapter 6: Summary of Recommendations and the Way Forward . . . . . . . . . . . . . . . . . . . . 31 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31 The Way Forward . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31 Annex A. Menu of Indicators for Agricultural Statistics . . . . . . . . . . . . . . . . . . . . . . . . . . 33 Annex B. Examples of Sample Frames Used for Agricultural Statistics . . . . . . . . . . . . . . . . 37 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39 E C O N O M IC AND S E CT OR WORK VI C ONTENTS BOXES Box 1. Indicators, variables, and data items.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 Box 2. Reminder . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 Box 3. The Brazilian Institute of Geography and Statistics’ integration of the agricultural census with the population counting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21 Box 4. Master sample frame . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 Box 5. China’s integrated statistical system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 FIGURES Figure 1. Country response to the FAO, 2007 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 Figure 2. The conceptual framework for agricultural statistics. . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 Figure 3. The integrated survey framework . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25 TABLES Table 1. Minimum set of core data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16 Table 2. Frequency of coverage by geographic and structural detail. . . . . . . . . . . . . . . . . . . . . . . . 17 Table 3. Example of a replicated survey design with the use of an annual core questionnaire and rotating sets of supplemental questionnaires . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24 GLOB A L STR ATEGY TO IMPR OVE A GR ICULTURA L A ND R UR A L S TATIS TIC S P R E FA C E V II PREFACE The Global Strategy to Improve Agricultural and Rural Statistics presented in this document is based on input from a large number of stakeholders, including national statistical institutes and ministries of agriculture, and a number of regional and international organizations. One of the outcomes of the 2007 International Statistical Institute Conference on Agricultural Statistics was a consensus regarding the challenges of applying statistics to issues in agricultural development. Not only is there a lack of direction regarding agricultural data requirements posed by the Millennium Development Goals (MDG) and other emerging issues such as the use of food for biofuels, and the environment and food security; there is also a general decline in the overall quality and availability of agricultural statistics. These concerns were discussed during the 2008 meeting of the United Nations Statistical Commission (UNSC). The discus- sion led to the formation of a working group assigned to draft a strategic plan to improve agricultural statistics. The working group, under the guidance of the United Nations Statistics Division (UNSD), included the World Bank, the United Nations Food and Agriculture Organization (FAO), Statistical Of�ce of the European Union (Eurostat), the United States Department of Agriculture (USDA), and the International Statistical Institute (ISI). Using input from the working group and other stakeholders, the World Bank prepared a paper entitled “Framework to Develop a Strategic Plan to Improve National and International Agricultural Statistics.� The paper was the basis for the Expert Meeting on Agricultural Statistics held in Washington, DC, on October 22–23, 2008. The Expert Meeting was attended by heads and representatives of national statistical of�ces and ministries of agriculture from 27 countries. The FAO, the World Bank, International Monetary Fund (IMF), Eurostat, Organization of Economic Cooperation and Development (OECD), and the USDA also attended. The outcomes of the Meeting formed the basis for a paper discussed at the 2009 meeting of the UNSC, which concluded that a global strategy was needed to improve agricultural statistics. The UNSC recommended that a Friends of Chair Working Group be formed to develop the global strategy for review and approval at the 41st Meeting of the Commission in 2010. The Working Group is led by Brazil (Mr. Eduardo Pereira Nunes) and includes Australia, Brazil, China, Cuba, Ethiopia, Italy, Morocco, the Philippines, the Russian Federation, Trinidad and Tobago, Uganda, the United States, the FAO, and the UNSD, serving both as observer and secretariat, and Eurostat and the World Bank as observers. With input from the Friends of Chair Working Group and other stakeholders, the World Bank developed a draft “Global Strategy to Improve Agricultural Statistics� in collaboration with the FAO. The draft provided the basis for the International Statistical Institute Satellite Meeting on Agricultural Statistics that took place in Maputo, Mozambique, in August 2009. The Meeting was organized around the chapters of the draft Global Strategy, and was attended by more than 200 participants from over 45 countries as well as from regional and international organizations. Funding agencies such as the Bill & Melinda Gates Foundation also showed their interest by sending delegates to the meeting to discuss possibilities and modalities for participating in this global initiative. Based on discussions at the Maputo Meeting, the Friends of Chair formed four working groups to provide more details about components of the paper through consultations within the network of statisticians. E C O N O M IC AND S E CT OR WORK VIII PR EFA C E At the International Statistical Institute Conference in Durban, South Africa, that followed the Maputo Meeting later that month, a wide variety of papers on agricultural and rural statistics was presented and discussed. The papers covered topics relating to the Global Strategy such as censuses of agriculture, survey methods, and economic-environmental accounting for agriculture. A review of agricultural statistics by the United Nations Economic Commission for Europe again underscored the need for improved statistics in developing countries. The FAO included the Global Strategy as a main item on the agenda of its Biannual Conference in November of 2009—an event at which ministers of agriculture of all member countries were gathered. The Strategy was also discussed at sessions of the Regional Commissions on Agricultural Statistics attended by national directors of agriculture statistics. The PARIS21 Consortium meeting in Dakar in November 2009 provided another opportunity to further discuss the Global Strategy with a variety of stakeholders, donors, governments, private businesses, intermediate organizations, and statisti- cians. A seminar on the Global Strategy attracted around 100 participants and contributed signi�cantly to the further recogni- tion of its importance. A peer review by experts from the World Bank, International Food Policy Research Institute (IFPRI), and USDA also provided input to the Global Strategy. Efforts to expand access to the development of the Global Strategy to all ministries of agriculture and national statistical of�ces included the development of a Wikipedia Web page: wiki.asfoc.ibge.gov.br. This global consultation helped the Friends of Chair Working Group to improve the document and to widely publicize the initiative. The technical content and strategic directions of the Global Strategy were endorsed by the 41st session of the UNSC. It is the result of a wide process of consultation with national and international statistical organisations as well as with agricultural ministries and other governmental organisations represented in FAO governing bodies. GLOB A L STR ATEGY TO IMPR OVE A GR ICULTURA L A ND R UR A L S TATIS TIC S A C K N O W L E DGE ME NT S IX ACKNOWLEDGMENTS The Global Strategy to Improve Agricultural and Rural Statistics was prepared by the World Bank in collaboration with the FAO and Friends of the Chair working groups of the UNSC, and through extensive consultations with stakeholders. The World Bank team was led by Sanjiva Cooke and included Nwanze Okidegbe and Fred Vogel (principal author). The team is grateful for comments received from the following peer reviewers: Gero Carletto, Robert Townsend (World Bank), Gerald Nelson (International Food Policy Research Institute), and Mary Ahearn (United States Department of Agriculture). The team thanks Juergen Voegele, Mark Cackler, and Misha Belkindas for their support and inputs. The FAO team was led by Pietro Gennari and included Naman Kieta, Hiek Som, and Greg Chong. The Friends of the Chair on Agricultural Statistics was led by Eduardo Pereira Nunes (Brazil) and included representatives from Australia, China, Cuba, Ethiopia, Italy, Morocco, Philippines, Russian Federation, Trinidad and Tobago, Uganda, United States, FAO, the United Nations Statistics Division (UNSD), the Statistical Of�ce of the European Union (Eurostat), and the World Bank. World Bank staff Gunnar Larson edited and Sonia Madhvani managed the production and design of the publication. The National Statistical Institute in Mozambique, which hosted the International Statistical Institute Satellite Meeting on Agricultural Statistics, warrants special recognition. The meeting was organized in cooperation with the African Development Bank, Eurostat, the FAO, the Partnership in Statistics for Development in the 21st Century (PARIS21), the United Nations Statistics Division, the United States Department of Agriculture, and the World Bank. The African Development Bank, the World Bank, and the U. S. Department of Agriculture are also recognized for the �nancial support provided during the nearly two year period to develop the Global Strategy. The International Statistical Institute’s support of the global initiative to improve agricultural statistics has been greatly appreciated. The preparation of the Global Strategy was supported by the Trust Fund for Statistical Capacity Building (TFSCB), a multi- donor trust fund �nanced by Canada, the Netherlands, the United Kingdom, and administered by the Development Data Group of the World Bank. E C O N O M IC AND S E CT OR WORK E X E C U T I V E S UMMARY XI EXECUTIVE SUMMARY Policy makers and development practitioners who are responsible for developing investment strategies to promote economic growth �nd many challenges in the changing face of agriculture in the twenty-�rst century. In addition to its productive role of providing food, clothing, fuel, and housing for a growing world population, agriculture assumes other roles, the importance of which has more recently been recognized. In addition to its essential role in food security, agricultural development is now seen as a vital and high-impact source of poverty reduction. It is also seen as a source of environmental problems and a contributor to global warming, water scarcity and pollution, and land degradation. At the same time its potential as a source of environmental services needs to be de�ned, monitored, and evaluated. Many of the issues facing the sector transcend national boundaries. The Global Strategy is the result of an extensive consultation process with national and international statistical organizations as well as with agriculture ministries and other governmental institutions represented in FAO governing bodies. Considerable input came from the United Nations Statistical Commission Friends of Chair working group and the 2009 meetings of the International Statistical Institute in Maputo and Durban. Other collaboration involved the FAO Biannual Conference and dis- cussions at the Regional Commissions on Agricultural Statistics attended by national directors of agricultural statistics, the World Bank peer review process, and the development of a Wikipedia Web page to collect inputs from the statistical com- munity (wiki.asfoc.ibge.gov.br). The purpose of the Global Strategy is to provide a framework for national and international statistical systems that enables them to produce and to apply the basic data and information needed to guide decision making in the twenty-�rst century. This Strategy is based on three pillars. The �rst pillar is the establishment of a minimum set of core data that countries will collect to meet current and emerging demands. The second pillar is the integration of agriculture into national statistical systems in order to satisfy the demands of policy makers and other users who rely on comparable data across locations and over time. The integration will be achieved by implementing a set of methodologies that includes the development of a master sample frame for agriculture, the implementation of an integrated survey framework, and with results available in a data management system. The third pillar is the foundation that will provide the sustainability of the agricultural statistics system through governance and statistical capacity building. The Strategy is based on an assessment of the data that users need and that are currently available. The assessment, which is described in chapter 1, not only found a serious decline in the quantity and quality of agricultural statistics, but one that is occurring at the same time that many new data requirements are emerging. Among these emerging data requirements are those relating to global warming, land and water use, and the increasing use of food and feed commodities to produce biofuels—in addition to a number of requirements that relate to poverty and food security. The assessment of data that users need led to the formulation of a conceptual framework that relates the economic, social, and environmental dimensions of agriculture. This framework incorporates forestry, �sheries, and land and water use in E C O N O M IC AND S E CT OR WORK XII EX EC UTIV E S UM M A RY addition to the narrower, more conventional treatment of agricultural production. It recognizes linkages between rural house- holds, agricultural holdings, and the land and other natural resources that they use and that they impact. Applying this conceptual framework, an evaluation of national agricultural statistical systems points to an urgent need to improve their capacity to systematically collect and report reliable data. The evaluation also found a need to improve the coordination between national statistical organizations and the other national agencies that produce agricultural statistics. In 2008, the Global Donor Platform for Rural Development, with support from the United Nations Food and Agriculture Organization (FAO) and the World Bank, published a sourcebook of indicators for monitoring and evaluating results in agri- culture and rural development. This set of indicators was used as the starting point to develop a full menu of indicators that meets both current and emerging information requirements. From this menu of indicators, a set of core data or statistics is de�ned that will provide the input to estimate the indicators. The minimum set of core data is intended to be used as a start- ing point in building agricultural statistics systems for the twenty-�rst century. A strategy to determine the content, coverage, and frequency of the national system that goes beyond the core set of data is also provided. The emerging data requirements, the conceptual framework, the assessment of the national agricultural statistics systems, and the choice of a core set of indicators all point to the need to integrate agriculture into the national statistical systems. The Strategy identi�es the main elements with which the integration will be achieved. The integration of agriculture into a country’s national statistical system will begin with the development of a master sample frame for agriculture. This will be the foundation for all data collection based on sample surveys or censuses. The master sample frame is to be constructed based on the requirements to include both households and farms as statistical units. It provides a link between the census framework and land use. An integrated survey framework will be established to provide data measured consistently across time and comparable across countries using an annual survey of selected core items and periodic data from a set of rotating panels covering economic and environmental issues. The concept of a master sample frame will be extended to include a data management system for all of�cial statistics related to agriculture. All data collection is to be based on sample units selected from the master sample frame and integrated into the survey framework. The survey framework also takes into account the additional data sources that need to be included in the inte- grated statistical system, including administrative data, agribusiness and market information systems, community surveys, remote sensing, and consistent input from expert data collections. The of�cial statistics that are gathered are to reside in a data management system. These are the basic principles of the Strategy. Their implementation will require improved gover- nance across the national statistical system. The integration of agriculture into national statistical systems will also affect the roles and the divisions of responsibility between national statistical of�ces, ministries of agriculture, and institutions that govern other sectors. The Strategy suggests that each country establish a national statistical council to coordinate the integration of agriculture when the country designs its National Strategy for the Development of Statistics (NSDS). However, the Strategy leaves the respective roles of the organizations to the countries themselves to decide. The steps to implement the strategic plan will depend upon the statistical capacity of each country. Those needing to reform their statistical systems will begin with the core data items and build the rest over time. In countries in which national strate- gies for the development of statistics are being undertaken, they should be reviewed in light of the Global Strategy and revised accordingly. Many countries that have already developed statistical systems, but that have not integrated agricultural statistics into those systems will need to develop a master sample frame for agriculture and an integrated database. The Strategy is a long-term effort, with its implementation proceeding in stages that depend on each country’s initial statistical capacity. Given the dynamic nature of agriculture and its accompanying issues, the Strategy should be considered a living document to be updated when needed to reflect current situations. It will be followed by an implementation plan based on GLOB A L STR ATEGY TO IMPR OVE A GR ICULTURA L A ND R UR A L S TATIS TIC S E X E C U T I V E S UMMARY X III input from the national and international partners as well as additional contributors. The implementation plans will be flexible enough to consider the speci�c country situations. This paper presents the overall Strategy. It provides a ground-breaking effort to improve agricultural statistics that has implica- tions for other sectors in the national statistical system. While it took many years for agriculture statistics to deteriorate to their current state, the implementation of the Strategy provides a fresh start. E C O N O M IC AND S E CT OR WORK I N T R O D U CT ION 1 Chapter 1: INTRODUCTION THE AVAILABILITY AND QUALITY OF face, and the interaction of these with issues concerning AGRICULTURAL STATISTICS HAS DECLINED, climate change. The impacts of these factors can only be ef- JUST AT THE WRONG TIME fectively measured and evaluated with appropriate statistics. Three out of four poor people in developing countries live in However, at present there is a serious paucity of statistical rural areas. Most rely directly or indirectly on agriculture for data on which to base marketing, investment, or policy deci- their livelihoods. Agricultural development is vital to achiev- sions, or with which to assess the ef�cacy of current com- ing the Millennium Development Goals, particularly those mitments or policies. related to poverty and food security and to environmental Many countries, especially in the developing world, lack the sustainability. Agriculture contributes to development as an capacity to produce and report even the minimum set of agri- economic activity, as a source of livelihoods, and as a pro- cultural data necessary to monitor national trends or inform the vider of environmental services—roles that were spelled international development debate. The Independent External out in substantial detail in the 2008 World Development Evaluation of the Food and Agriculture Organization (FAO2006) Report “Agriculture for Development� (World Bank 2008a) argued that “the time has come for a total re-examination of Recognition of its importance has led to renewed commit- the statistical needs for the 21st century and how they can ment to agriculture within the international development best be met.� The evaluation report concluded that “the quan- community. This commitment has assumed mounting ur- tity and quality of data coming from national of�cial sources gency in a global context of skyrocketing food prices and has been on a steady decline since the early 1980s, particu- lowered food reserves. Globally, food prices doubled be- larly in Africa.� It also found that “of�cial data submissions tween 2006 and mid-2008, a trend driven in part by droughts from countries in Africa are at their lowest level since before in grain-producing regions, increased oil prices, and sales of 1961, with only one in four African countries reporting basic corn to produce biofuels. Future food prices are expected to crop production data.� The evaluation also recognized the in- remain higher than they were in the 1990s and to be more creasing demands for new statistics and the need to integrate volatile. The role of agriculture as a source of greenhouse gas data on agriculture, �sheries, and forestry to understand their emissions and other environmental problems has also as- effects on the environment and climate change and on the sumed prominence, given the need to raise production, but use of biofuels to deal effectively with policy issues. with little latitude to expand production into new areas. The need to measure agricultural performance and the results of Low response rates to the United Nations Food and Agricultural agricultural investment has therefore become an increasingly Organization (FAO) questionnaires limit the availability of data. pressing priority.1 Figure 1 presents the response rates by data domain (produc- tion, land use, agricultural machinery, trade, fertilizer, and pes- Decisions about aid and investments that are intended to ticides) and by region. Response rates from the Paci�c, Africa foster agricultural growth need to be based on sound infor- (except for trade and pesticides data), and the Middle East mation about land use, factors of agricultural production, are the lowest, while Europe has the highest rates. Response the prevailing economic and social situations that producers rates from Latin America for basic data on production, land use, machinery, and pesticides are also very low. 1 See “The Comprehensive African Agriculture Development Program� www.nepad-caad.net; “Joint Donor Principles for Agriculture and Rural Development Programs� (The Global Factors contributing to the decline Donor Platform for Rural Development 2009; “World Develop- ment Report, Agriculture for Development,� (World Bank 2008) A number of likely reasons are attributable for the decline for a detailed discussion of these issues. in the quantity and the quality of statistics pertaining to E C O N O M IC AND S E CT OR WORK 2 INTR OD UC TION FIGURE 1: Country responses to the FAO, 2007 Countries Response Rate to FAO questionnaires for 2007 data 100% 80% Percentage 60% 40% 20% 0% Machinery Land Production Trade and Fertilizers Pesticides Use Equipment Europe 66% 78% 71% 64% 71% 43% Asia and Pacific 63% 35% 32% 33% 43% 10% Americas 38% 51% 17% 28% 40% 15% Africa 34% 45% 13% 13% 25% 22% Near East 37% 37% 37% 26% 16% Oceania 15% Regions agriculture and rural development. One obvious reason is the systems, coordination between the national statistical of- lack of country-level capacity at public statistical agencies. �ce and the ministry of agriculture is often lacking. Nor do An evaluation of FAO in 2008 argued that the most pressing many national strategies for the development of statistics need in national statistical systems is to improve the capacity adequately cover the agriculture sector. for agricultural statistics, which the evaluation described as a “re-emerging� need. The decline in the priority and in the A recent review by the Partnership in Statistics for resources that national agricultural systems assign to collect- Development in 21st Century (PARIS21) found that, of a total ing and reporting reliable agricultural statistics is paralleled of 78 International Development Association (IDA) countries, by a general lack in donor interest. The need to quantify such 43 (55 percent) have a national strategy for the development matters as the impacts of agricultural production on the envi- of statistics in which agriculture is or is supposed to be in- ronment and the impacts of biofuel production on food prices cluded. Among these 43 IDA countries, only 4 to 10 countries entails the development of new conceptual frameworks that (therefore only around 10 percent of all IDA countries in the go well beyond traditional domains of agricultural statistics. world) have included agriculture more or less appropriately It also makes this an exceptionally inopportune time for the in the National Strategy for the Development of Statistics collection of even those traditional domains to be assigned (NSDS) process (PARIS21 2009). lower priority than they have in the past. A number of problems are common to many developing Not unrelated to the lack of capacity is the lack of �nancial countries: resources to collect data. The dilemma is that agricultural sta- Limited staff and capacity of the units that are respon- tistics are often outside the national statistical system, with sible for collection, compilation, analysis, and dissemi- ministries of agriculture and other organizations responsible nation of agricultural statistics. for sectors such as land, water use, �sheries, and forestry Lack of adequate technical tools, statistical methodol- also failing to keep up with the increasing demand for data. ogy, and survey framework to support data-production In many countries, the lack of integration into the national efforts. statistical system is a major reason for the weakness of ag- Insuf�cient funding allocated for agricultural statistics ricultural statistics. In countries with decentralized statistical from development partners and national budgets. GLOB A L STR ATEGY TO IMPR OVE A GR ICULTURA L A ND R UR A L S TATIS TIC S I N T R O D U CT ION 3 Lack of institutional coordination, which results in the resources to carry it forward. The Global Strategy continues lack of harmonized and integrated data sources. with the following chapters. Lack of capacity to analyze data in a policy perspec- Chapter 2. A Conceptual Framework for the Collec- tive, which results in a signi�cant waste of resources tion of Agricultural Statistics. A conceptual frame- as large amounts of raw data are not properly used. work based on a thorough assessment of users’ data Dif�culty for data users in accessing existing data with needs was developed. It pointed to many emerging no metadata or indication of quality. requirements from issues closely linked to agriculture A systematic assessment is needed, using a standard in- such as poverty and hunger, the environment and cli- ternational framework, to provide a detailed diagnosis and mate change, the use of land and water, and the in- analysis of the current statistical capabilities by country. creasing use of food and feed commodities to produce The assessment should cover all principal data domains, biofuels. Based on these requirements, the concep- including data gaps, data quality, and related institutional tual framework broadens the scope and coverage of and methodological limitations with regards to priority data agricultural statistics to include aspects of �sheries, needs. The information available suggests that Africa, the forestry, and rural households and provides a menu of Middle East, the Paci�c, and Latin America have the largest indicators. The conceptual framework translates policy number of countries with weak agricultural statistics sys- issues into statistical language by identifying the need tems. Countries in these regions require a comprehensive for the survey framework to link the farm as an eco- capacity-development effort to enable them to provide the nomic unit, the household as a social unit, and the land minimum data requirements. they occupy in the natural environment. The framework suggests that the fundamentals of the Global Strategy be based on three pillars: identifying a minimum set of core data; the integration of agriculture into the na- FILLING THE VOID: THE GLOBAL STRATEGY tional statistical system; and the sustainability of the TO IMPROVE AGRICULTURAL AND RURAL agricultural statistical system through governance and STATISTICS statistical capacity building. The Global Strategy provides a blueprint for a coordinated Chapter 3. The First Pillar—Identifying a Minimum and long-term initiative to address the decline in agricultural Set of Core Data and Determining National statistics systems. Several efforts related to the goal to Priorities. Because the complete set of data require- improve agricultural statistics provided valuable input to the ments identi�ed in the conceptual framework exceeds development of the Global Strategy. These include Tracking the existing statistical capacity of many countries, a Results in Agriculture and Rural Development in Less than minimum set of core data is to be used as a starting Ideal Conditions: A Sourcebook of Indicators for Monitoring point upon which to develop the Global Strategy. This and Evaluation (World Bank 2008b), The World Programme core set of data will provide national and international for the Census of Agriculture 2010 (FAO 2005b), The Guide policy makers necessary information that goes across to Designing a National Strategy for the Development of national boundaries. The Global Strategy provides a Statistics (PARIS21 2007), and the Wye Group Handbook on framework for countries to add items of national inter- Rural Households Livelihood and Well-Being (United Nations est to the set of core data and to determine the fre- 2007). quency with which they will be provided. The set of The Global Strategy is also based on extensive consultations core data provides the beginning point for the improve- with national statistical of�ces, ministries of agriculture, and ment of agricultural and rural statistics. other national institutes as well as with all international statis- Chapter 4. The Second Pillar—The Integration tical organizations that have a stake in improving agricultural of Agriculture into National Statistical Systems. statistics. It takes into consideration the different stages of Overlapping data requirements and the need to improve statistical development across countries and the technical underlying statistics and methodology point directly to developments that can contribute to the improvement of sta- the need to integrate agriculture into the national sta- tistics. The Strategy should therefore be considered a long- tistical system. Incorporating agriculture into national range plan requiring an examination of governance at the na- statistical systems will facilitate the concentration of tional level, the establishment of statistical capacity building resources from multiple sources, and remove the dupli- across the national statistical system, and the restoration of cation of efforts in producing statistics that is so E C O N O M IC AND S E CT OR WORK 4 INTR OD UC TION common in developing countries. The Strategy provides Chapter 6. Summary of Recommendations and the the framework to achieve the integration based on the Way Forward. The Global Strategy concludes with a development of a master frame for agriculture, its use in summary of major recommendations and the conclu- an integrated survey system, and the implementation of sions reached by the UNSC in its acceptance and en- a data management system. dorsement of the Global Strategy and a summary of ac- Chapter 5. The Third Pillar: The Sustainability of Agri- tions to be taken to develop an implementation plan. cultural Statistics by Governance and Statistical Annex A provides a menu of indicators, data sources, Capacity Building The conceptual framework leading and technical notes. The core indicators provided in the to the integration of agriculture into national statistical Sourcebook (2008) and the emerging requirements de- systems points to requirements for governance that scribed in the FAO evaluations were used as starting bring together the efforts of the different stakehold- points to develop the menu. The menu of indicators ers, especially the national statistical institutes and also includes those needed to understand the issues ministries of agriculture. While the Strategy provides involving the environment, climate change, and the in- the framework for integration, it leaves the implemen- troduction of biofuels. Because countries have varied tation to each country to decide and suggests they do and limited capabilities, it will be necessary for each so by forming national statistics councils. Other issues country to establish priorities for the collection of the addressed are the steps needed to implement the basic data in addition to a core set that are universally Strategy, including the inclusion of the fundamentals of needed and are comparable across countries. the Strategy in the national strategies for the develop- Annex B provides an overview of sample frames used ment of statistics. for agricultural statistics. GLOB A L STR ATEGY TO IMPR OVE A GR ICULTURA L A ND R UR A L S TATIS TIC S A C O N C E P T UAL F RAME WORK F OR T HE CO LLECTI O N OF A GRICULTURA L STATISTICS 5 Chapter 2: A CONCEPTUAL FRAMEWORK FOR THE COLLECTION OF AGRICULTURAL STATISTICS Statistics on agricultural and rural development are used by the need for improved integration and for more accessible policy makers, donors, and private sector decision makers to and searchable databases. inform their decisions regarding a variety of important issues. It is these priority issues that drive the choice of indicators to The most critical issues are not independent of each other, be developed and the core data to be collected. and much of the data are needed for more than one indicator. The goal of the Strategy is to capture the interrelationships of A variety of attempts have been made to quantify the value these emerging issues and to ensure that appropriate indica- of information to public and private sector decision makers. tors are de�ned and underlying data provided. This points to Case studies examined in literature reviews suggest that the a major problem with current agricultural and rural statistics. bene�ts of having the information far outweigh the costs of Many of the issues have been considered in isolation, and providing the information. For example, Bruce Gardner re- this does not allow the cross-cutting analysis that is most viewed literature that quanti�es the value of agricultural mar- desperately needed. ket information to private and public decision makers regard- ing domestic U.S. policy reform, trade policy reforms, and DIMENSIONS OF DATA DEMAND investments in public research and development (Gardner 2004). In the same volume, George Norton and Jeffrey While agriculture is fundamentally an economic activity, in that Alwang consider case studies on the value of information its purpose is the production of food and other commodities, regarding deforestation in the Amazon and pesticide use in concern about its relationships to environmental and social the Philippines. Both reviews found high net bene�ts to pro- issues has been increasing. These relationships have to be viding decision makers the information (Norton and Alwang considered in a broader context, in which agriculture, the 2004). environment, and social factors are no longer treated as dis- crete disciplines. Institutions and enterprises affect all three Many of today’s critical issues are not new, but they have through policies, regulations, taxes, and infrastructure such increased in importance, have come to be framed differ- as transportation, education, markets, and processing facili- ently, or have been newly recognized. Many of the traditional ties. The signi�cance of the institutional framework applies at indicators in use, therefore, remain relevant, while others the local, national, and international levels. The international need to be refocused or newly developed. The indepen- level warrants consideration owing to the globalization of dent review of the United Nations Food and Agricultural markets and to the reality that some of the most important Organization (FAO) statistics program included an effort to issues, such as global warming and many facets of poverty seek input on emerging data needs from major users and reduction, transcend national or regional boundaries. Many of partners. One overall �nding of the report was that there the enterprises involved in this bigger picture are not directly was a great deal of overlap in the issues identi�ed among engaged in agricultural production, but they provide services stakeholders, including national statistical centers, nongov- that connect production to markets and consumers. ernmental and donor organizations, research institutions, and a variety of other users. Multiple users expressed the The economic dimension of agriculture consists of the land, need for new and improved indicators on prices, energy and labor, and capital that enter into the production process and biofuels, agricultural environments, climate change, trade, the outputs that result from it. The output of the production water, land, soils, household consumption, food security, process takes many forms. Some are consumed by the socioeconomic data, economic accounts, management of household, some are retained for seed or feed to be used natural disasters, and �sheries. Users also had high expecta- on the holding, and others enter supply chains that extend to tions for geospatial and remote sensing data and expressed markets. Some of the products require processing, such as E C O N O M IC AND S E CT OR WORK 6 A C O N C EPTUA L FR A MEWOR K FOR TH E C OLLECTION OF A GRICULTUR A L S TATIS TIC S crushing soybeans for oil, ginning cotton, or the slaughter of FIGURE 2: The conceptual framework for agricultural livestock by nonagricultural enterprises. An emerging output statistics is the use of agricultural commodities for the production of energy products. The outcome of the production process is Households, institutions, enterprises income to the agricultural and nonagricultural enterprises, and to households—both agricultural and nonagricultural. The impact of the production process affects food security, Agricultural Environmental services/sustainability Economy Social statistics poverty, and the performance of the economy. (scope and agricultural pollution, biodiversity, mitigation, investments, land labor etc. safety nets, food security, gender etc. coverage) adaptation etc. The environmental dimension of agriculture consists of the Inputs, outputs. sector’s role as a user of natural resources—principally land outcomes, impact and water—and as a provider of environmental services. In Agricultural production processes, agro-processing and markets crops, livestock, fisheries, etc. addition to its direct use of natural resources in production, its impacts also relate to the waste and emission byproducts generated by production. Agriculture can affect the condition of the resources it uses, and with important implications for households concerned. Environmental standards can have climate change and biodiversity. Recognition of the nega- serious economic consequences for household income. tive and potentially positive impacts that agriculture has on global, regional, and local environments points to the need for statistics that enable informed analysis of the interac- THE CONCEPTUAL FRAMEWORK tions between agriculture’s roles in the economy and in the The conceptual framework presented in �gure 2 brings to- environment. gether the economic, environmental, and social dimensions Data that relate to the social dimension of agriculture and of agriculture and the cause-and-effect relationships that rural development begin with households and household connect them. These relate to agricultural production and members—both farm and nonfarm. This represents a higher extend to processing and markets as well as income distri- level of detail than was captured by much conventional data, bution, accumulation, and consumption. The relationships which more often began with the farm enterprise as its are also a function of the prevailing institutional framework basic unit. Including nonfarm rural households also serves within which agriculture operates. Agricultural statistics are to develop a broader and more complete picture of rural needed at each respective stage: inputs, outputs, outcomes, communities and the multitude of interdependencies that and ultimate impacts. characterize them. Rural communities are far more than spa- tial clusterings of households located in sparsely populated areas. Understanding the interactions between rural house- THE ECONOMIC DIMENSION holds, businesses, and government agencies—and between The economic dimension covers agricultural production, mar- communities—generates the need for extensive data. It is kets, and farm and nonfarm income. especially important that the combination of agricultural and nonagricultural income sources among households, farms, Agricultural production. Data on agricultural productivity are and nonfarm businesses be represented with data, given the important to decision and policy makers. Productivity rises signi�cance of seasonality to food security among vulnerable when additional output is produced for the same level of households and individuals. The data are also very important inputs, or alternatively, the same amount of output is pro- in examining the relationships that exist between agriculture duced with fewer inputs. Data on the quantity and prices and other sectors in rural society. Finally, social data are of outputs and inputs are therefore the starting points for needed to examine households and individuals not only in measuring changes in agricultural productivity. Information their roles as producers and consumers, but also as users of is also required on capital stock that is used over multiple social services such as health and education programs. years of production in order to determine the equipment’s rate of depreciation. Together this information can be used The output of the agricultural process affects the income of to develop a balance sheet. both agricultural and nonagricultural households. Policy deci- sions that affect the choices made about different patterns While growth rates in agricultural productivity is a long- of production have implications for the well-being of the standing concern, the role of growth in reducing poverty and GLOB A L STR ATEGY TO IMPR OVE A GR ICULTURA L A ND R UR A L S TATIS TIC S A C O N C E P T UAL F RAME WORK F OR T HE CO LLECTI O N OF A GRICULTURA L STATISTICS 7 hunger has more recently become a focus on monitoring and value added is distributed among the factors of production: evaluation. The 2008 World Development Report Agriculture land, labor, and capital, as well as to entrepreneurial manage- for Development (World Bank 2008a) relates evidence that ment. While agricultural GDP is useful in measuring the over- growth in GDP that originates in agriculture is at least twice all performance of the sector over time, it is less informative as effective in reducing poverty as growth originating in other about the well-being of different categories of producers sectors. The effectiveness of public and private agricultural and households. Many households engage in nonfarm work investment has therefore become more closely monitored activities, and sometimes more than one household shares than it was in the past, whether that investment is in infra- the returns of a farm holding. These complex resource alloca- structure, new technology, or physical or human capital. tions within and among households, along with an uneven distribution of income, mean that average GDP income is Fish and other aquatic organisms are major sources of food not a useful indicator of well-being. Because the ultimate and household income throughout much of the developing goal of nearly all development projects is to reduce poverty, world. The subsector includes �sh captured on the open- more detailed indicators are required to monitor progress, sea commons, in coastal zones within the territory of indi- and in this capacity the use of household surveys has been vidual countries, in rivers and other freshwater sources, and prescribed. Using household surveys to collect data on ag- through aquaculture. Countries are responsible for providing riculture, however, is very resource-intensive, to the point statistics on all �sheries and aquaculture within their national that the surveys are often not practical in many develop- jurisdiction and on vessels that fly their flag. Regional �shery ing countries (although the Living Standards Measurement bodies have been formed to coordinate the collection of data Survey-Integrated Surveys on Agriculture (Carletto 2009) and management of �shery resources. These data gener- represents an important recent success in this respect). The ally contain more detailed information on operational and Sourcebook (World Bank 2008b) presents a number of al- biological aspects of capture �sheries, including the species ternatives to household surveys, including a service delivery composition of the catch. Small-scale and subsistence aqua- survey to determine whether services are actually reaching culture and capture �sheries often provide an opportunity of the poor and vulnerable. last resort for earnings and food security for people without access to land. Small households that engage in aquaculture tend to combine �sh production with other activities such as crop agriculture. The competition between aquaculture and THE SOCIAL DIMENSION agriculture for land and water use is intensifying and is likely The social dimension covers the need to reduce risk and to increase with the impacts of climate change. vulnerability, including food security, and issues related to gender. Markets. Effective marketing systems depend on informa- tion about supply and demand and market prices, which is Reducing risk and vulnerability. National leaders and private freely available to all participants in the system. The most decision makers in the marketplace will be better able to essential information is timely forecasts and estimates of manage risk and vulnerability with information that enables production. Timeliness is a critical factor. The lack of timely them to recognize or forecast potential hazards. While natu- production data was one of the major factors leading to food ral disasters such as droughts and storms are relatively con- shortages and spikes in consumer prices. The marketing stant sources of risk, market factors can seriously compound system should be considered in the broader sense to include that risk. For instance, a drought in a major producing region markets for inputs and those involved at every stage of the combined with large-scale transition from food to biofuel pro- modern supply chain from production to �nal delivery to the duction in another major producing region can, as we have consumer. (These indicators are also required to measure seen, seriously inflate global food prices. Effectively manag- agricultural productivity.) ing risks at this level requires excellent and timely data, and ensuring access to these data is a concern that has major Farm and nonfarm income and survey data. Net farm income rami�cations for international food security. and GDP from agriculture are basic indicators of a country’s agricultural performance. Tracking them for policy purposes Food security. Assessing food security at the national level can provide an understanding of the conditions facing pro- requires information on commodity production, using a ducers as a group—and whether or not they are likely to have number of the indicators used to measure productivity and adequate resources for the next production cycle. One of the market ef�ciency. In addition, food security includes consid- ways that national accounts are utilized is by examining how eration of food trade and nonfood use (fuel, drug industry, E C O N O M IC AND S E CT OR WORK 8 A C O N C EPTUA L FR A MEWOR K FOR TH E C OLLECTION OF A GRICULTUR A L S TATIS TIC S seed, feed, etc.). Information on consumption by agricultural Estimating the payoff of switching from less sustainable pro- and nonagricultural households is also required. The informa- duction systems to more sustainable ones requires disaggre- tion on food demand collected in household surveys involves gated data in a number of areas, including data on agricultural all households in the country, urban and rural, agricultural and inputs. For instance, the question of how much fertilizer is nonagricultural. Food security also requires information to as- used and how much of it returns to the watershed as runoff sess the food gap in terms of nutrients. is particularly important because more ef�cient fertilizer use can increase productivity even as it reduces the amount that Gender. In many developing countries and in rural societies becomes a source of water pollution. in particular, household roles, responsibilities, and rights are highly gendered. Income commanded by women has a dis- Food and feed products for biofuels. Biofuels may reduce proportionately positive effect on the health, nutrition, and carbon emissions from burning fossil fuels and raise income education of other members of their households. Women for producers. The supply effects of converting food and feed have also proven to be highly receptive adopters of technolo- crops to biofuel production can also raise food prices, quite gies that raise yields and improve environmental manage- possibly to a level that pushes consumers into poverty. For ment, such as agroforestry techniques—once their property this reason the use of nonfood crops such as switchgrass rights have been secured. The Third Millennium Development and jatropha to produce biofuels has received increased Goal, to “promote gender equality and empower women,� attention. Switchgrass can be cultivated on marginal, highly therefore, carries particular weight in the rural and agricul- erodible lands with little other agricultural potential, and tural development agenda, and the need to disaggregate converting its biomass to fuel requires less energy than the pertinent data by gender is generally acknowledged. conversion of food crops. Jatropha is a small seed-producing tree whose seeds can be crushed to produce oil that can be used to produce biodiesel. The crop is being grown in South THE ENVIRONMENTAL DIMENSION America, Africa, and Asia, and it is resistant to drought and The environmental dimension of agriculture generally applies pests. The production of these nonfood commodities could to the sector’s sustainability and to its provision of environ- still raise food prices if they replace traditional crops and the mental services. input and marketing infrastructure that was developed around them. Output quantities and input prices relevant to biofuel Agriculture and the environment. Policies and programs that issues are also relevant to the measurement of productiv- seek to mitigate its environmental impacts or to capitalize on ity. These, however, must be available in a disaggregated its potential as a source of environmental services require form in order to measure the relative costs and bene�ts of extensive information. Public of�cials and development prac- biofuel commodities and other agricultural commodities— titioners who champion such policies or who promote such particularly food crops. investments are often in a position of advocacy in which they must solicit the commitment of scarce resources. In this role Land cover and use, including forestry. Land is the foundation they require the credibility that only reliable information can of agriculture and forestry. How the land is used determines give them. Their ability to present informed estimations of its sustainability and productivity. The use of land can also the likely impacts of environmentally sustainable agricultural have environmental consequences that range from pollution initiatives is essential, particularly given the politically sensi- of waterways to global warming. Land cover is de�ned as tive contexts in which they often operate. The scale of agri- “the observed physical cover including the vegetation (natu- culture’s impacts on the environment remains indeterminate ral or planted) and human constructions that cover the earth’s without sound data. surface� (FAO 2005a). Agricultural expansion is the principal factor contributing to deforestation, which results in increas- While more detailed data on agriculture’s bene�cial and ing levels of carbon dioxide in the atmosphere. Forests and adverse impacts on the environment are urgently needed, woodlands absorb carbon dioxide (a major cause of global some broader facts point to the sector’s general signi�cance. warming) from the atmosphere, thus mitigating the effect of The opportunity costs of foregoing its potential as a source of carbon emissions from burning fossil fuels. It is necessary environmental services are likely very great. Environmentally to monitor land cover over time to reveal changes resulting sustainable agriculture can sequester large volumes of carbon from deforestation, urbanization, deserti�cation, and other from the atmosphere. It can also play a positive role in man- measures related to not only agricultural productivity but also aging watersheds and in preserving agricultural biodiversity. to the overall affect on the environment and global warming. GLOB A L STR ATEGY TO IMPR OVE A GR ICULTURA L A ND R UR A L S TATIS TIC S A C O N C E P T UAL F RAME WORK F OR T HE CO LLECTI O N OF A GRICULTURA L STATISTICS 9 The System of Integrated Environmental and Economic crop and livestock production. The FAO uses this classi�ca- Accounting (SEEA) uses two land classi�cation systems. tion to determine the scope of the agricultural census as de- The Land Cover Classi�cation System manual (FAO 2005a), scribed in The World Program for the Census of Agriculture jointly prepared by the FAO, the United Nations Environment 2010 (FAO 2005b). The Central Product Classi�cation (CPC) Programme (UNEP), and Cooperazione, Italiana, provides an provides an additional international standard. Its most recent international standard with which to categorize land cover and revision, CPC 2.0, contains a number of important amend- how humans activities affect it. This establishes a direct link ments and re�nements in the area of agriculture, forestry, between land cover and the actions of people in their environ- �sheries, and food. Items such as crops, livestock products, ment. For example, “grassland� is a land cover, while “range- machinery and equipment, and fertilizers and pesticides land� refers to its use to support livestock. The other clas- that are included in World Programme for the Census of si�cation is contributed by the FAO based on global statistical Agriculture 2010 (FAO 2005b) are also classi�ed in the CPC databases of agricultural and forestry land-use structures. 2.0. Both ISIC and CPC provide important instruments for integrating agricultural statistics into national statistical Water use. Like land, water is a critical integrating variable systems. that cross cuts with agriculture, forestry, and �sheries, which, in combination, affect the environment, climate change, and Agroforestry and aquaculture are considered to be agricul- food security. Water for irrigation is a major factor in improv- tural activities, although other activities related to forestry ing land productivity and crop yields. According to AQUASTAT, and �sheries are generally outside the scope of the agri- FAO’s global information system on water and agriculture, ag- cultural census—unless they are carried out in association riculture uses 70 percent of freshwater withdrawals globally with production on an agricultural holding. In many countries and 85 percent in developing countries. Demand for water is these compete with agriculture for land and water and are increasing for both agricultural and nonagricultural uses. In often the object of land-and-water-use policies that have both some countries, this is leading to unsustainable extraction of economic and environmental consequences. groundwater. There is a lack of data concerning water use for agriculture, the distribution of irrigated land, and water use The scope of agricultural statistics based on the broader con- practices, including aquaculture. ceptual framework includes aspects of forestry, �sheries, and land and water use. This expanded purview is required to address the merging and often closely related economic, AGRICULTURAL STATISTICS: SCOPE social, and environmental issues faced by policy makers. AND COVERAGE Because of the fundamental relationship between agriculture Scope. The starting point in determining the scope of required and land, the geospatial aspects of land should be seen as agricultural statistics is the system of national accounts an element of the scope of agricultural statistics. The geo- (SNA), which provides international standards for concepts, spatial scope for agricultural statistics should focus on the de�nitions, and classi�cations of economic activities. The use of land for agriculture and forestry and take place within conceptual framework also points to the need for a system a broader scope of national land-use statistics. of environmental accounts with which to monitor the effects of agriculture on the environment. The System of Integrated Forestry and agroforestry relate both to the production of Environmental Economic Accounting (SEEA)—which is a forest products and to the interface between forestry and satellite account of the SNA—should be the starting point agriculture as an area of environmental impact. Collecting for environmental statistics. While there is a framework for and reporting the data required for forestry and woodlands household decision making, there is no equivalent interna- outside of agriculture will be the responsibility of the con- tionally accepted standard for social statistics. The guiding ventional sources, which, from a governance standpoint, will principle will be to follow the socioeconomic variables cap- become part of the national statistical system for coordina- tured within the national accounts. tion purposes. The International Standard Industrial Classi�cation of Aquaculture and capture �sheries are important components Economic Activities (ISIC) divides agricultural production into of both food supply and security and household income. three categories or groups. Group 011 encompasses the cul- Aquaculture is de�ned by the FAO’s World Programme tivation of crops, market gardening, and horticulture. Group for the Census of Agriculture (FAO 2005b) as the farming 012 relates to “farming of animals,� and group 013 to mixed of aquatic organisms such as �sh, crustaceans, mollusks, E C O N O M IC AND S E CT OR WORK 10 A C O N C EPTUA L FR A MEWOR K FOR TH E C OLLECTION OF A GRICULTUR A L S TATIS TIC S aquatic plants, and other aquatic organisms. This implies In some cases, there will not be a one-to-one correspondence feeding, regular stocking, protecting, and raising organisms between the agricultural holding and the household. However, through one or more life cycles. All aquaculture and capture a goal should be to statistically establish the links between the production, employment, and food security information economic, environmental, and social dimensions as described will be within the scope of agricultural statistics. This does in the conceptual framework. Chapter 4 will discuss a strat- not mean the national statistical of�ce undertakes the data egy for establishing these links via an integrated statistical collection if it is the responsibility of another governmental system. body. However, the responsibility for oversight should reside with the national statistical system, which will use common Rural households fall within the scope of agricultural sta- standards, de�nitions, and coordination of the data that are tistics. Agricultural development provides a pathway out of published. poverty and hunger for the rural poor. These pathways can include improving the income of small agricultural holders The scope of agricultural statistics will include use of water through wage employment in agriculture or the rural nonfarm for agricultural purposes, including irrigation and other uses, economy, or through migration. The need for statistics for the source of irrigation water, the land under irrigation, the rural development led to the production of The Wye Group irrigation method, and the resulting production. This will be Handbook on Rural Households Livelihood and Well-Being done in collaboration with the FAO-AQUASTAT Programme, (United Nations 2007). The necessary data underlying many the global information system on water and agriculture. of the indicators needed to monitor rural development and economic growth leading to poverty and hunger reduction The intersection of the connections between the dimensions are based on the rural household as a statistical unit. of the conceptual framework points to the need for data described in systems of accounts such as supply utilization Other statistical units required by the Global Strategy are accounts, food balances, and income accounts for the house- enterprises that service agriculture such as input suppliers, hold and agricultural enterprises. These accounts require data processors, and transporters of agricultural goods. While from many sources, including the government, households, these economic units are outside the conceptual framework agricultural holdings, and agricultural businesses. The follow- for agriculture, they do provide information on prices and ing paragraphs de�ne the coverage and statistical units to be quantities that are important for economic and environmen- included in the scope of agricultural statistics. tal accounts. Local communities are important sources of information on social services provided to agriculture holders Coverage. The FAO’s World Program for the Census of and rural households. Agriculture (FAO 2005b) recommends that the 2010 round of the census consider the agricultural holding as the basic The coverage of agricultural statistics should be as exhaustive unit for economic statistics. However, the same report pro- and as comprehensive as possible, and any omission of units vides guidelines about the use of a population census and based on their size, importance, location, or other criteria should the collection of agricultural data for households that are be avoided. Many countries apply such criteria to reduce the not agricultural producers. The use of the population cen- costs of collecting data. Some stipulate a minimum size that a sus to obtain basic information about agricultural and rural holding must be in order to be included in a census or survey. households provides a means of broadening the scope of Some concentrate data collection in major producing areas. the coverage required to meet the emerging data require- This selective focus leaves smaller plots and remote parts of a ments of the conceptual framework. The World Program for country unrepresented in agricultural statistics, although these the Census of Agriculture (FAO 2005b) and the Sourcebook areas may account for a majority of the country’s food insecu- (World Bank 2008b) also consider the rural community as a rity and poverty. The omission of small-holder and household unit for social statistics. The basic unit for social statistics plots also deprives decision makers of information about local is the household; for environmental statistics it is the land subsistence strategies or the amount of income households parcel. The challenge will be to link these statistical units. In receive from selling produce from gardens and small plots. many cases, there will be one-to-one relationship between Because many small holdings are often the responsibility the agricultural holding, the household, and the land parcel. of women, the omission of this information overlooks a key In these cases, it will be feasible to collect economic, social, source of gender-disaggregated data on well-being. and environmental information from one unit. If these units are georeferenced, then the three dimensions can also be For the purposes of the Strategy, all agricultural units regard- associated with the overall land use. less of size and location should be included in the scope of GLOB A L STR ATEGY TO IMPR OVE A GR ICULTURA L A ND R UR A L S TATIS TIC S A C O N C E P T UAL F RAME WORK F OR T HE CO LLECTI O N OF A GRICULTURA L STATISTICS 11 agricultural statistics. This would be made possible by the its production need to be set aside for reproduction even inclusion of some basic questions about agriculture in the in times of drought or famine. A complicating factor is that population census. The inclusion of the small and geographi- many items are perishable. These issues create fundamen- cally isolated household holdings in the annual statistical pro- tal requirements for current, timely, and accurate measures gram will be considered in the following chapters on method- of production. The need for timeliness is a major factor un- ology. While size criteria and geographic coverage should be derlying agricultural statistics—information about a bumper inclusive for the population and agricultural censuses, these crop is worth little if it becomes available after the crop has can be different for the annual survey program, which more rotted in storage because of lack of information about its often covers commercial farms. availability. In order to provide information that generates an accurate picture of the activities of the agriculture sector, The agricultural statistician faces special problems. statisticians, therefore, require data that take into account Agricultural production is seasonal and related to the bio- seasonal variation and the heterogeneity of production logical life cycle of plants and animals. More than any other patterns. sector, it is dependent on the physical landscape (fertility and type of soil), weather, and climate. Agriculture must be The data requirements and the conceptual framework for self-sustaining; grain harvested provides the seed for next agriculture have been described. The next chapter builds on year’s crop, and animal births represent the next genera- these topics to establish a minimum set of core data that can tion. While agriculture feeds the world, certain amounts of be used to derive many of the required indicators. E C O N O M IC AND S E CT OR WORK T H E F I R ST P IL L AR —IDE NT IF YING A MINIM U M SET OF CORE D ATA A ND D ETER MINING NATIONA L PR IOR IT IES 13 Chapter 3: THE FIRST PILLAR—IDENTIFYING A MINIMUM SET OF CORE DATA AND DETERMINING NATIONAL PRIORITIES This chapter de�nes a minimum set of internationally com- for all of those several hundred items are needed for indica- parable core data that countries should provide. The data tors such as GDP growth from agriculture value added and a requirements are shown as a menu of indicators in annex A. number of others. The United Nations Food and Agricultural This menu of indicators includes those that are sector wide Organization (FAO) sends annual questionnaires to countries for agricultural and rural development as well as others for requesting data on production, trade, land use, agricul- subsectors such as crop and livestock, indicators for climate tural machinery and equipment, fertilizer, and pesticides. change, land and the environment, and the rural economy. Producer price data are also requested. The problem is that The menu shows data requirements to provide the indicators, these annual requests cover the population shown above. data sources, and technical notes. Box 1 provides examples These data requirements exceed what any country can pro- of indicators, data items, and variables. The selection of these duce on an annual basis. Therefore, the �rst step is to select core data is necessary because the total amount of data that a minimum subset that countries will provide using common would be required to meet all requirements exceeds what de�nitions and methodologies to ensure that measurements most developing countries can currently provide—until the are internationally comparable. The goal is to determine the capacity of their statistical systems has been substantially minimum subset of items for which data will be provided increased. A framework is presented for countries to identify annually and the frequency with which the remaining data other items in addition to this core set of data, to determine will be furnished. the frequency that data should be provided, and to establish the extent of the national coverage that is required. The core The following paragraphs describe the process to establish and national items and associated data will be used as start- an internationally agreed upon set of core data items that ing points to implement the Global Strategy as it is de�ned each country will provide. Because countries have varied and in chapter 4. limited capabilities, it will be necessary for each of them to establish priorities on what to include in their national statisti- The indicators in annex A reveal that basic statistics on crop cal system in addition to the core set. production, livestock, aquaculture, �sh captures, and timber removal from forests are major sources of information. The Core data items are selected on the basis of their importance World Program for the Census of Agriculture (FAO 2005b) to agricultural production globally. For example, only about lists 149 crops, 28 livestock species, and about 1,400 �sh- 10 crops and 4 livestock species account for over 95 per- ery and aquaculture species. Not all are produced in every cent of the world’s production of cereals, meat, and �ber. A country, and they are not of equal importance everywhere core item is one whose data enter into a multitude of indica- they are produced. Data on inputs, production, and prices tors needed to monitor and evaluate development policies, food security, and progress toward meeting the Millennium Development Goals (MDG). Core data should provide inputs to the national accounts and global balances of supply and BOX 1: Indicators, variables, and data items demand for food and other agricultural products. Core data items that are crops should account for a major proportion A food production index is an indicator. of land use, contribute signi�cantly to farm and rural house- Maize is a data item that enters into the index. hold well-being, and have an effect on the environment and Variables about maize include area harvested, yield, pro- climate. A core item should be the �rst to be included in the duction, utilization, prices, etc. statistical system and the last to be removed as a result of budget shortfalls. E C O N O M IC AND S E CT OR WORK 14 T HE F IRST P IL L AR—IDE NTI FYI N G A MINIMUM S ET OF CORE D ATA A ND D ETER MINING NATIONA L PR IOR ITIES Core items and their related data are required by the global products leads directly to increased usage of feed grains, and statistical system to monitor issues that go beyond national can lead to situations in which feed production competes boundaries. The globalization of the world’s economies with food production, even though the feed is ultimately means that an action in one part of the world affects the food an input to food production. Livestock are also sources of supplies, the environment, and the climate in other areas. methane emissions, water pollutants, and disease risk. All of these factors can be affected by policy decisions. Data The list of core items and associated data should establish the required for these livestock items include: framework for the agricultural and rural components of the National Strategies for the Develop of Statistics (NSDS) when a. Inventory and annual births. they are being implemented. The set of core data items will b. Production of products such as meat, milk, eggs, and be the building block to establish methodology and to inte- wool, and net trade or imports and exports. grate agriculture and rural statistics into the national system. c. Producer and consumer prices. The designation of core data starts with basic production Core aquaculture and �sheries products. These contribute statistics for the major crop items, livestock, aquaculture, signi�cantly to food supplies, and in the case of aquacul- �shery products, and forestry; and continues with agricultural ture, production entails the use of land as well as of water inputs, socio-economic data, land cover, and public expen- resources. Fisheries provide livelihoods for small-scale and diture. These are presented in the following section, which inland holdings. Data required include: is itself followed by a framework for countries to add their additional national requirements to the core list and to also a. Area cultured, production, prices, and net trade or determine the frequency for which both core and national imports and exports for aquaculture. data will be provided. b. Quantity landed and discarded, number of days �shed, amounts processed for food and nonfood uses, prices, and imports and exports. SET OF CORE ITEMS AND ASSOCIATED DATA Core forestry production. Forestry is a major land use, pro- Core crop items. Wheat, maize, barley, sorghum, rice, sugar vides income, and has a signi�cant role in understanding the cane, soybeans, and cotton are core crop items. These forces affecting climate change. Data required include: account for a major proportion of agricultural land use, of overall food supply, and of value added from agriculture. a. Area in woodlands and forests, quantities removed, Their production can vary considerably from year to year. and their prices for land associated with agricultural Because their products can be used for a variety of purposes, holdings. including bioenergy, decisions about which commodities to b. Area in woodlands and forests, quantities removed, produce can have important implications for food supply. and their prices for products from nonagricultural Data required for these core items include: holdings and respective utilizations. a. Area planted and harvested, yield, and production. Core agricultural inputs. Core inputs to agricultural production b. Amounts in storage at the beginning of harvest. include labor, chemicals, water, energy, and capital stocks. c. Area of cropland that is irrigated. Inputs are considered core because, in combination with d. Producer and consumer prices. data about outputs, they provide measures of agricultural productivity important to monitoring and evaluating steps to e. Amounts utilized for own consumption, food, feed, reduce poverty and hunger. Data required include: seed, �ber, oil for food, bioenergy, and net trade or imports and exports. a. Quantities of fertilizer and pesticides utilized. f. Early warning indications such as precipitation, wind- b. Water and energy consumed. shield surveys of crop conditions, and vegetative c. Capital stocks such as machinery by purpose (i.e., indices provided by satellite observations. tillage or harvesting). Core livestock items. These include cattle, sheep, pigs, d. Number of people of working age by sex. goats, and poultry. These are major sources of food supply e. Number of workers hired by agricultural holders. and agricultural income. Consumption increases as countries f. Employment of household members on the agricul- develop and incomes grow. Increased demand for livestock tural holding. GLOB A L STR ATEGY TO IMPR OVE A GR ICULTURA L A ND R UR A L S TATIS TIC S T H E F I R ST P IL L AR —IDE NT IF YING A MINIM U M SET OF CORE D ATA A ND D ETER MINING NATIONA L PR IOR IT IES 15 Core socioeconomic data. The socioeconomic characteris- Each country, therefore, needs to select which core items tics of agricultural and rural households include household to include in its national system. It must add other items income by source as a key measure of the economic well- relevant to its economy and determine how frequently data being of rural households to guide policy decisions about will be provided and the scope of the national coverage developmental efforts to reduce poverty. Periodic data about required. For example, the core data do not include fruits the number of households, employment, population, age, and vegetables or other livestock items that contribute to a gender, and education levels are also required. country’s food supplies and household income. Each country should consider how these should be included in its national Land cover. A fundamental way to evaluate agriculture’s af- system. fect on the environment is to monitor changes in land cover and use. Land cover does not change rapidly and data are Annual data are generally required for those items that, not, therefore, required on an annual basis. However, map- combined, account for more than three-fourths of a coun- ping products or digitized data from remote sensing should try’s value of production. Items with production that can vary provide complete coverage for the entire land mass of a signi�cantly from year to year should be included, particularly country with the following classi�cations: if the production fluctuations are a major source of risk for vulnerable households and food supplies. Items that account a. Cropland for a signi�cant proportion of land used and that have short- b. Forest land term effects on land use and the environment should be rep- c. Grassland resented as well. Including items that are produced by only a d. Wetlands small number of households or holdings or that account for e. Settlements only a small share of the country’s land has sample design f. Other land and resource implications. For example, sampling theory g. Water shows that the relative variance of the estimated mean is approximated by the relative variance of the positive sample Public expenditures on subsidies, infrastructure, and health units plus the relative variance of the estimated proportion of and education in rural areas are core items. This should positive population units. include the availability of roads, transport services, commu- nications, and extension services. CV2 (Y) = CV2 (Yp) + CV2 (P) where Yp is the mean of the positive responses and P is the proportion of the popula- Table 1 below shows the core data items grouped by key tion that has the item variables in the economic, social, and environmental di- mensions. Note that the basic production data items are Assuming that only a third of the households or holdings required annually. The strategy to establish the frequency of have a particular item, the sample size will have to be four the remaining core data items is described in the next sec- times larger than if three-fourths have the item in order to tion that elaborates on the steps to determine the national achieve the same level of precision. If only 10 percent of priorities. The frequency requirement is also considered in the households or holdings have the item, then sample sizes the design of the integrated survey framework presented in triple over what is needed if a third are positive and would be chapter 4. 12 times greater than if (P) > 0.75 for the same level of preci- sion. The general conclusion of this exercise is that minor and relatively rare commodities should be con�ned to the 5 to 10 year agricultural census and omitted from more fre- DETERMINING NATIONAL PRIORITIES: quent surveys. The exception would be if the sample frame CONTENT, SCOPE, AND FREQUENCY contains suf�cient data that can be used in the survey design Data for some core items will not be required every year either to target the rare items. because they do not change much from year to year or because they are dif�cult and expensive to obtain annually. Countries The next step is to review the rural development indicators will also have additional items to add to the list of core items for monitoring and development in the Sourcebook (2008) to meet national data needs. Teff, for example, is a major crop and include those relevant to the national situation. Then and food source in Eritrea and Ethiopia, but not in other coun- each country should determine the level of geographic tries. Items such as rice, on the other hand, are major global coverage and detail to be provided for the core plus addi- food sources, but are not produced in every country. tional items added. The same issue raised above about the E C O N O M IC AND S E CT OR WORK 16 T HE F IRST P IL L AR—IDE NTI FYI N G A MINIMUM S ET OF CORE D ATA A ND D ETER MINING NATIONA L PR IOR ITIES TABLE 1: Minimum set of core data GROUP OF KEY VARIABLES CORE DATA ITEMS FREQUENCYa VARIABLES ECONOMIC Output Production Core crops (e.g., wheat, rice, etc.) Annual Core livestock (e.g., cattle, sheep, pigs, etc.) Core forestry products Core �shery and aquaculture products Area harvested and planted Core crops (e.g., wheat, rice, etc.) Annual Yield/births/productivity Core crops, core livestock, core forestry, core �shery Annual Trade Exports in quantity and value Core crops, core livestock, core forestry, core �shery Annual Imports in quantity and value Core crops, core livestock, core forestry, core �shery Annual Stocks Quantities in storage at beginning of Core crops Annual harvest Stock of resources Land cover and use Land area Economically active population Number of people in working age by sex Livestock Number of live animals Machinery Number of tractors, harvesters, seeders, etc. Inputs Water Quantity of water withdrawn for agricultural irrigation Fertilizers in quantity and value Core fertilizers by core crops Pesticides in quantity and value Core pesticides (e.g., fungicides herbicides, insecticides, disinfectants) by core crops Seeds in quantity and value By core crops Feed in quantity and value By core crops Agro processing Volume of core crops/livestock/�shery used By industry in processing food Value of output of processed food By industry Other uses (e.g., biofuels) Prices Producer prices Core crops, core livestock, core forestry, core �shery Consumer prices Core crops, core livestock, core forestry, core �shery Final expenditure Government expenditure on agriculture and Public investments, subsidies, etc. rural development Private investments Investment in machinery, in research and development, in infrastructure Household consumption Consumption of core crops/livestock/etc. in quantity and value Rural infrastructure (capital stock) Irrigation/roads/railways/communications Area equipped for irrigation/roads in km/railways in km/ communications International transfer ODAb for agriculture and rural development SOCIAL Demographics of urban and rural Sex population Age in completed years By sex a The frequency for the items not speci�ed will be established by the framework provided in the Global Strategy to determine the national priorities for content, scope, and frequency. The frequency requirement will also be considered in the establishment of the integrated survey framework where the data sources will be de�ned. b ODA = Of�cial Development Assistance GLOB A L STR ATEGY TO IMPR OVE A GR ICULTURA L A ND R UR A L S TATIS TIC S T H E F I R ST P IL L AR —IDE NT IF YING A MINIM U M SET OF CORE D ATA A ND D ETER MINING NATIONA L PR IOR IT IES 17 TABLE 1: Minimum set of core data GROUP OF KEY VARIABLES CORE DATA ITEMS FREQUENCY VARIABLES Country of birth By sex Highest level of education completed One digit ISCED by sex Labor status Employed, unemployed, inactive by sex Status in employment Self employment and employee by sex Economic sector in employment International standard industrial classi�cation by sex Occupation in employment International standard classi�cation of occupations by sex Total income of the household Household composition By sex Number of family/hired workers on the By sex holding Housing conditions Type of building, building character, main material, etc. ENVIRONMENTAL Land Soil degradation Variables will be based on above core items on land cover and use, water use, and other inputs to Water Pollution due to agriculture production. Air Emissions due to agriculture GEOGRAPHIC LOCATION GIS coordinates Location of the statistical unit Parcel, province, region, country Degree of urbanization Urban/Rural area TABLE 2: Frequency of coverage by geographic and structural detail Level of geographic and structural detail DATA ITEM MAJOR PRODUCTION NATIONAL COVERAGE WITHIN COUNTRY INCLUSIVE OF AREAS ONLY— OF PRODUCTION BY ADMINISTRATIVE HOUSEHOLDS AND HH PRODUCTION BY HOLDINGS AREAS—PRODUCTION PLOTS HOLDINGS BY HOLDINGS Crop A Annual Annual Decennial census Decennial census Crop B Biannual Biannual Decennial census Decennial census Crop C Decennial Crop Z Time and available resources result in a necessary compromise between frequency, level of geographic detail, and other breakdowns. Livestock A These categories need to be considered for each data item. Livestock B Livestock Y Aquaculture and �shery Forestry Inputs Household income Change in land cover E C O N O M IC AND S E CT OR WORK 18 T HE F IRST P IL L AR—IDE NTI FYI N G A MINIMUM S ET OF CORE D ATA A ND D ETER MINING NATIONA L PR IOR ITIES proportion of households or holdings that have the item will It is generally true that policy makers will want data for within also determine the level of geographic detail or other break- country administrative areas such as provinces; if so, this down that can be provided from the sample surveys. These should be included in the national framework. have implications about the methodology to be used and the At this stage, each country should have an overall picture of resources required. The annual collections of data will rely the content of its national statistical system for agriculture, upon sample surveys that will limit the geographic detail that including the rural, forestry, and �shery components and the can be provided. Therefore, it may be only through an agri- coverage and frequency of the data provided. Input from cultural census that detailed geographic or size distribution policy makers and other data users should shape this �nal data can be provided. picture. The question of what level of detail is required and how The data-user requirements, conceptual framework, and often data are required may be dif�cult to answer. Table 2 steps to determine the content of the national statistical presents a decision matrix that is useful in many contexts. programs have been de�ned. The following chapter pro- For example, it should be determined for each item whether vides the strategy and methodology to integrate agriculture the data will be provided for the entire country or only for the into the national statistical system and improve agricultural major producing areas. statistics. GLOB A L STR ATEGY TO IMPR OVE A GR ICULTURA L A ND R UR A L S TATIS TIC S T H E S E C OND P IL L AR—INT E GRAT ING AGRI CU LTU RE INTO NATIONA L STATISTICA L SY S TEMS 19 Chapter 4: THE SECOND PILLAR—INTEGRATING AGRICULTURE INTO NATIONAL STATISTICAL SYSTEMS This chapter provides an overview of the statistical meth- samples. The separate data provide no basis for analyzing odology to improve agricultural statistics that will meet the the characteristics of farms that produce both crops and live- requirements of policy makers and other data users. The stock, or for comparing them to farms that specialize in one conceptual framework is used as a base. The statistical or the other. Household surveys are often conducted in isola- framework will provide the blueprint for the methodological tion from production surveys with no coordination or with requirements for agriculture in the national strategies for the sample sizes too small to disaggregate the data into the rural development of statistics. The integration and underlying and farm sectors. The results generated from these surveys methodology described below considers the quality dimen- are also not integrated into a common database for access sions, which include relevance and completeness, accuracy, by data users. timeliness, accessibility, coherence, and comparability. More than one governmental organization is often involved The process of improving agricultural statistics will begin in the collection and analysis of agricultural, �shery, and for- with the integration of agriculture into the national statisti- estry data without coordination. While the National Statistical cal system. This integration will be accomplished by the Of�ce may produce the agricultural census, the annual pro- development of a master sample frame for agriculture to duction data could come from the ministry of agriculture, ensure relevance and completeness; its use in implement- and the contribution of the �shery and aquaculture sectors ing a coordinated data collection program to produce timely may come from another authority and may be ignored or and accurate data that are coherent and comparable; and a neglected by the National Statistical Of�ce. In some cases, strategy for data dissemination to ensure accessibility. This different organizations produce statistics for the same items, integration of agriculture into the national statistical system with different results, which confuse the data users and is needed for several reasons. make it dif�cult to aggregate results across countries. This means that results then differ also at the international level One of the shortcomings of current statistical systems in if those organizations use different sources to populate their both industrialized and developing countries is that data are data bases. collected by sector, using different sampling frames and sur- veys. The division of data by sector leaves no opportunity to Integrated statistical systems can resolve many of these measure the impact of an action in one sector on another. problems by avoiding duplications of effort, preventing the Surveys are often conducted on an ad-hoc basis with no links release of conflicting statistics, and ensuring the best use of to a master sampling frame or the use of georeferenced units resources. Concepts, de�nitions, and classi�cations become for data collection. It is therefore dif�cult to integrate data standardized, allowing more systematic data collection across from various surveys for in-depth analysis with cross tabula- sources. These practical advantages of integrated data sys- tion of variables. Data on crop and livestock production are tems together with the increasing need for reliable and com- drawn from separate surveys, which are based on separate parable data in a context of globalization and international concern about environmental issues point to the need for integrated national statistical systems. The World Program BOX 2: Reminder: for the Census of Agriculture (FAO 2005b) argues forcefully for the development of such integrated systems. The use of the word “agriculture� in the strategy is inclu- sive of the broader scope to include forestry, �sheries, In some countries, centralized organizational structures are and aquaculture as described in chapter 2. already in place, and national statistical of�ces maintain the principal responsibility for agricultural statistics. However, E C O N O M IC AND S E CT OR WORK 20 T HE SE CO N D PI LLA R —INTEGR ATING A GR ICULTURE INTO NATIONA L STATIS TIC A L S Y S TEM S this centralized role may not always meet the needs of the for shortening the period between data collection and dis- line ministries such as the ministry of agriculture. For that semination with improved data quality. reason, the statistical responsibilities in many countries are decentralized with the agricultural statistics produced by the ministries of agriculture. Both systems have advantages and FRAMEWORK TO DEVELOP A MASTER SAMPLE disadvantages. National statistical of�ces have experience FRAME FOR AGRICULTURE with statistical methodology and sample frames that other The development of the master sample frame for agriculture ministries do not have. However, the other ministries have starts by de�ning the population parameters, which are the more knowledge about agriculture, forestry, �sheries, and physical land mass and natural environment of the country, land use. The purpose of the Global Strategy is to propose the economic output of agriculture, and the well-being of the a framework for integration that builds off the strengths of farm and rural populations. For data-collection purposes, both systems. the population needs to be de�ned in terms of the unit of The integration of agriculture into the national statistical measure or the statistical units. The statistical units de�ned system will be based on statistical methodology using tools in the conceptual framework include the farm or agricultural that establish a closer link between results from different holding, the household, and the land parcels. The conceptual statistical processes and different statistical units. This framework requires a link between the economic, environ- can be achieved by the development of a master sampling mental, and social dimensions and their statistical units. This frame, the adoption of sample designs such as overlapping entails the need for georeferencing the farms and house- samples, and the synchronization of questionnaire designs holds. All of these issues are considered in the development and surveys. of the master sample frame. The master sample, sample designs, and the survey frame- Annex B provides an overview of the different approaches work need to be considered together because there are currently used by countries to establish sampling frames choices to be made, such as whether to monitor the same for agricultural statistics. Developing countries often use farms and households, or to use different samples, and col- the enumeration or administrative areas established for the lect some of the same variables across surveys. It is also population census as the sample frame. Samples of farms necessary that countries have some flexibility in how the are obtained by �rst selecting enumeration areas, screening master sample frame and resulting survey designs are imple- them for farms or households, and then selecting a sub- mented to consider their national requirements as well as sample for the surveys. Other countries prepare registers of statistical capabilities. farms for sampling purposes and must devote considerable resources to keep them up-to-date. A less used approach The statistical methodology to be used also needs to con- is an area sample frame, which is essentially the country’s sider some basic data quality dimensions—timeliness, com- land mass divided into sampling units (Gallego 1995). Many pleteness, comparability, and accuracy. Measures for each of the requirements posed by the Global Strategy point to an quality dimension will be considered in the development of increased use of area frame methodologies. A �nal approach the strategy. The following sections provide the strategy to is to use multiple frames (FAO 1998) to create a master frame create a master sample frame followed by the sample and that builds on the advantages of area frames and registers. survey frameworks to achieve the integration. The master sample frame must provide the basis for the The following strategy also builds on recent developments selection of probability-based samples of farms and house- in agriculture statistics, including the use of satellite imagery holds with the capability to link the farm characteristics with for monitoring land use, estimating crop areas, and providing the household and then connect both to the land cover and early warnings of changing growing conditions, to name a use dimensions. The area sample frame meets this require- few examples. In addition, the development of global po- ment. The methodology using the population census recom- sitioning systems (GPS) makes it possible to georeference mended for the World Program for the Census of Agriculture observations and data collection to the land cover provided (FAO 2005) will also meet this requirement—if households by the satellite imagery. The emergence of the Internet and from the population census are georeferenced and used as other technology, such as the use of personal digital assis- the frame for the agricultural census and linked to satellite tants (PDAs) equipped with GPS systems for data collection images of land use. At this stage, only a limited number and their connection to databases, has tremendous potential of countries have included agriculture in their population GLOB A L STR ATEGY TO IMPR OVE A GR ICULTURA L A ND R UR A L S TATIS TIC S T H E S E C OND P IL L AR—INT E GRAT ING AGRI CU LTU RE INTO NATIONA L STATISTICA L SY S TEMS 21 census. According to information currently available to the BOX 3: The Brazilian Institute of Geography and FAO, only 71 countries out of a total of 189 member coun- Statistics integration of the agricultural tries have plans to undertake an agricultural census during census with the population counting 2006–15. Given these constraints, it becomes important to provide alternative methods to develop the master sample Integration facilitated by the use of PDAs frame for agriculture. equipped with GPS for data collection. The list of 5.2 million agricultural holdings is refer- The strategy to follow starts with a long-term vision for how enced to the households listed in the population the master sample frame for agriculture should be devel- counting. oped. The strategy is mindful of differing levels of capacity Each agricultural holding can be visualized by between countries; therefore, alternative methods to de- means of Google Earth images combined with velop the master sample frame are also provided. the grid of the agricultural census enumeration The development of the master sample frame for agriculture areas. begins with the need to link the economic and social dimen- The list frame of agricultural holdings with their sions of agriculture with those relating to land cover and oth- respective coordinates and the set of enumera- er environmental issues. Because the master sample frame tion areas surveyed by the agricultural census should be linked to land use, obtaining satellite imagery of forms the area frame and becomes the master the country’s area is a useful starting point. The land cover sample frame. as recorded by the satellite imagery should be classi�ed into major categories such as cultivated land, woodlands, grass- lands, idle land, and urban areas. Unless land use is changing with a farm, and if so, what are the indicators of size, type, rapidly, this imagery only needs to be updated periodically. and the location of the land (census enumeration area or This �rst step in creating the digitized land cover database administrative unit)? This information can be used to create should play a prominent role in efforts to build statistical a register of households and farms with their land linked to capacity. georeferenced census enumeration areas or administrative Once the land-use mapping is complete, the next step is to units. In census enumeration areas in which agricultural data georeference (or digitize) the population and agricultural cen- are collected, nonfarm households should be included in the sus enumeration areas to the satellite imagery. Countries, register. This will provide a link between the agricultural data districts, townships, and villages should be georeferenced and all characteristics contained in the population register. so that they are associated with the land-cover imagery. This While linking data from farm censuses and from population enables monitoring of land use over time, and can be used censuses provides a powerful tool for data analysis, several to relate land use to local administrative structures. This issues will need to be resolved. First, con�dentiality rules may information becomes an important component of the master limit how the census data can be used to construct a master sample frame for agriculture. frame for agriculture. In addition, the register will need to be supplemented by a register of commercial farms not associ- A number of strategies can be employed to create a master ated with households in order to provide a complete register sample frame. The �rst method discussed below is used for agricultural surveys. A more ideal approach would be to establish a link between the agricultural master sample to use the household or farm register as an input into the frame and the population census. Given the fact that the agricultural census. Then the master frame for agriculture link cannot be made for many years because of the infre- would be the same as described below when a census of quent nature of population censuses, additional strategies agriculture is the base. are also offered for: countries with recent agricultural cen- suses; countries that use administrative data to construct a Master sample frame with an agricultural census. The devel- sample frame; and those that do not have recent agricultural opment of the master sample frame using the agricultural censuses. census includes the need to associate farms with households and both with land use. Historically, the reporting unit for the Coordinated population and agricultural census data collec- agricultural census is the farm. The �rst step is for data collec- tion. The basic information that should be obtained in the tion to not only de�ne the farm along with obtaining produc- population census is whether the household is associated tion and economic information, but also to obtain information E C O N O M IC AND S E CT OR WORK 22 T HE SE CO N D PI LLA R —INTEGR ATING A GR ICULTURE INTO NATIONA L STATIS TIC A L S Y S TEM S about the household(s) associated with the farm and the association during data collection to assign both farms and household characteristics. The coverage of the census should related households to the segment or point that will be al- be inclusive of both commercial and small-scale farms plus ready georeferenced to land use. The data quality dimension subsistence farming households. The goal should be that the of completeness is satis�ed because the entire country is farms counted in the census be used to develop a register, mapped and every farm, household, and parcel of land has a and each farm should be associated with a household unless known probability of selection. It is also comparable because it is a corporate or institutional farm. A problem is that the the same segments or points can be used for multiple surveys point of data collection is the farm headquarters or household and over time. Once the country has established the area whose distance from the land holding poses dif�culties for frame, it may begin creating a list register of large or special- georeferencing each land holding to land use. Therefore, land ized farms to use in a multiple frame context. Nevertheless, associated with each farm and associated household needs to the area frame described above becomes the master sample be linked to the appropriate georeferenced census enumera- frame for agriculture with the capability to directly link or geo- tion areas or administrative units, or both. In this example, reference the farm or household to its associated land hold- the master sample frame for agriculture will be a register of ing. This is an important advantage, as the households can be farms or households and commercial farm enterprises with located in villages some distance from the land holding. The their land georeferenced to enumeration areas or administra- sample segments or points should also be associated with tive units. Where the census is repeated at 10-year intervals, the census enumeration areas or administrative units. The link it will be necessary to update the register in the interim period of the sample units with census enumeration areas also puts using administrative information. An alternative procedure is the master frame into the population census framework. to use two-stage sampling in which the �rst stage is census enumeration areas or administrative units. The �rst stage In summary, the master sample frame for agriculture can units could be screened annually for updating purposes. be established several ways. The common element for the three methods provided above is the georeferencing of cen- The georeferencing of farms or households to the census sus enumeration areas and administrative units to digitized enumeration or the administrative units that are part of the satellite imagery classi�ed by major land cover. The area data layer in the satellite imagery in effect establishes an frame sampling units can be directly associated with the area sample frame—and becomes the master sample frame land-cover classi�cation. The land associated with the farms, for agriculture. households, and enterprises in the census or administra- tive registers is indirectly associated with land cover via the Countries using administrative data to construct registers of mapping to the census enumeration area or an administra- farms. The procedures described above to develop a master tive unit. A longer-term goal would be to georeference each sample using the census of agriculture should also be followed parcel associated with an agricultural holding directly to the where administrative information such as tax records, licens- satellite imagery. ing, or regulatory registers is available. However, additional steps may be required if the administrative data do not in- Once the master sample frame of farms and households has clude small or subsistence farms. This could include selecting been established, the next and longer-term step is to create a samples of administrative units or census enumeration areas, register of agricultural enterprises that furnish inputs, provide which would be screened for small and subsistence farms. Again, the georeferencing of the farms or households in the BOX 4: Master sample frame: business register to either census enumeration or adminis- trative areas in effect establishes an area sampling frame— The underlying principle is that the master sample frame which becomes the master sample frame for agriculture. be the source for all samples for surveys of agricultural holdings, farm households, and rural nonfarm house- Master sample frame when there is not a recent agricultural holds. This means the samples can be designed so that census. The starting point should be the development of an data can be analyzed across surveys. Once the master area sample frame. The georeferenced satellite imagery by sample frame has been developed, it should be possible land-use category can also be used as the basis for an area for different institutions in the national statistical system sample frame as described by Gallego (1995). The land-use to access the master sample for survey purposes with characteristics of the country should be used to select the another guiding principle that the resulting data be avail- sampling unit—segments with identi�able boundaries or a able for analysis across other data collections. sample of points. Either method can be used with rules of GLOB A L STR ATEGY TO IMPR OVE A GR ICULTURA L A ND R UR A L S TATIS TIC S T H E S E C OND P IL L AR—INT E GRAT ING AGRI CU LTU RE INTO NATIONA L STATISTICA L SY S TEMS 23 transportation, and are the �rst-stage processors of crop and The preparation of the integrated survey framework begins animal products. by �rst considering the set of core data requirements fol- lowed by the additional information needed by each country, The master sample frame enables the use of a rich assort- as summarized in table 2 in chapter 3 showing by item the ment of sample designs including single- and multiple-stage frequency of coverage, geographic detail, and inclusion of sampling. If enumeration or administrative areas are the �rst commercial agriculture versus small and subsistence farms. stage of sampling, they can be selected with probabilities proportional to measures of size reported in the population or The minimum set of core data includes statistics about the agricultural censuses. The use of enumeration or administra- production of major crop, livestock, aquaculture and �sher- tive areas provides a means of selecting farms, households, ies, and forestry products. The second requirement is for or a combination of them as the statistical unit. economic data on the agricultural holding, including inputs and outputs. The third requirement is to collect data on the Landing sites are the appropriate unit for surveying capture use of fertilizers, chemicals, tillage methods, and other land- �shery production, while the master sample frame can be use activities to monitor how agricultural production affects used to monitor other �shery-related units such as house- the environment. The fourth requirement is to measure the holds, holdings, and enterprises. When utilizing landing sites social well-being of the farm and rural households. The tra- as the sampling unit for data collection of capture production, ditional methodology is to select independent samples and the survey on the other aspects of the �shery sector will conduct separate surveys for each of the categories. While need to include questions about the landing sites used by the optimum sample design often leads to the selection each household, holding, and enterprise to allow integration of samples speci�c to crops, livestock, and the respective of two different sampling schemes. economic, environmental, and social surveys, it limits data analysis across the respective categories. VISION FOR THE INTEGRATED Single-purpose surveys generally make it easier to target SURVEY FRAMEWORK the selected sample such as crops or livestock, especially This section presents a vision for the integrated survey where both are not present on most farms, or, when pres- framework. The complete survey framework includes the ent, differ considerably in size. It is dif�cult to use strati�ed sample design, questionnaires, data-collection methods, designs using many different measures of size. Recent de- analysis, and estimation. It also takes into consideration the velopments in sampling theory provide an alternative using data sources in addition to sample surveys that provide input selection probabilities based on the measures of size for a into the survey framework. The overall strategy is presented. number of different variables. This design is termed “Multiple The technical and methodological elements will be part of Probability Proportional to Size� (MPPS) because the relative the implementation plan. size of each farm (or enumeration area) is determined for more than one item of interest. The use of this method in The timing and frequency of data collection are major issues China is described by Steiner (2007). It takes advantage of for agricultural statistics. Crops have different production ef�ciencies of the Probability-Proportionate-to-Size sampling cycles that are seasonal, while livestock production is deter- while adding the use of multiple measures of size. The use mined not only by the respective reproductive cycles, but of MPPS is appropriate for multiple-purpose surveys in which also the continuous production of commodities such as milk the population sample units each only have a subset of the and eggs. Aquaculture has characteristics similar to livestock items of interest. production. The rural labor force is also affected by the sea- sonal nature of agriculture, which affects opportunities for For the purposes of data analysis, it is desirable to select work and earnings. The timing of data collection affects the one large sample to provide all of the data for production, quality of the data, especially if a lengthy recall is required. As a result data collection should coincide with harvest periods. BOX 5: China’s integrated statistical system For example, if crop yields are determined by crop-cutting surveys, then these have to be measured shortly before har- MPPS sampling using multiple variables from the cen- vest. Fish capture requires frequent sampling and surveys— sus of agriculture is used to support an expanded survey for instance, twice a week or once every �ve days—in order program and to integrate the statistical needs for differ- for the data to reflect developments such as frequent and ent levels of government. unpredictable changes in species composition. E C O N O M IC AND S E CT OR WORK 24 T HE SE CO N D PI LLA R —INTEGR ATING A GR ICULTURE INTO NATIONA L STATIS TIC A L S Y S TEM S the economic situation of the holding, its environmental im- (minor crop or livestock items, for example). Data for pact, and the social well-being of the household. It would these items will come from rotating panel surveys also be desirable for the same sample to be used over time based on a subsample of the core survey. for longitudinal data analysis. While the MPPS sample design b. Select a replicated sample for the annual core items provides the basis to use a single sample, at the same time using MPPS. In other words, instead of select- it requires lengthy and complex questionnaires to include all ing one large sample, select several replicates. As items of interest. For this reason, a strategy to collect data shown in table 3, this allows a process to include for some core items annually coupled with periodic data col- some of the sample units in the survey across lection for other items is required to allow analysis across time for longitudinal analysis. Table 3 shows 12 subjects. replicates; 1 through 5 for year 1, 2 through 6 for year 2, etc. This provides longitudinal data, but also STEPS TO IMPLEMENT AN INTEGRATED limits the number of times for respondent burden SURVEY FRAMEWORK considerations. The integrated survey framework should be based on the c. Design a survey questionnaire to obtain the annual minimum set of core and national data and the determination core data items. Each year the core questionnaire of how frequently they are required. should contain supplemental questions regarding one of the subject matters described above. For example, a. Determine the set of core items for which at least in year 1 replicates 1 through 5 will be surveyed annual data are required. For those core items not using the core questionnaire, which will also con- needed annually, group them by category, including tain key questions about economic variables. The economic variables such as farm structure expen- core questionnaire can either obtain all information ditures and income; environmental measures such required, or a subsample could be selected for the as the use of fertilizers and chemicals and land and collection of the detailed data. In year 2, replicates 2 water use; social variables such as household income through 6 will be surveyed using the core ques- and well-being; and other items of national interest tionnaire, which will also contain questions about TABLE 3: Example of a replicated survey design with the use of an annual core questionnaire and rotating sets of supplemental questionnaires REPLICATE REP REP REP REP REP REP REP REP REP REP REP REP YEAR 1 2 3 4 5 6 7 8 9 10 11 12 1 A A A A A 2 B B B B B 3 C C C C C 4 Detailed Questionnaires for D D D D D 5 Rotating panel surveys A A A A A Every replicate receives the same core questionnaire 6 every year for annual core data items plus obtains B B B B B data for one following rotating panels: 7 C C C C C 8 A. Economic items including Farm structure, expenditures, D D D D D income 9 A A A A B. Environmental items including inputs, chemicals, 10 tillage, water use, land use B B B 11 C. HH income, consumption, employment C C 12 D. Items of national interest D GLOB A L STR ATEGY TO IMPR OVE A GR ICULTURA L A ND R UR A L S TATIS TIC S T H E S E C OND P IL L AR—INT E GRAT ING AGRI CU LTU RE INTO NATIONA L STATISTICA L SY S TEMS 25 environmental issues. By year 4 all of the subject estimators. While the direct unbiased estimators based on matters will have been included. the sample design form the foundation, they can be supple- d. Each year, one of the sets of panel data will be linked mented using ratio and regression estimators, or model- to the annual core items. Also note that starting with based estimators using census results. The use of multiple year 4, at least one of the replicates will have been estimators can improve data accuracy and reliability. surveyed by all of the rotating panel questionnaires in The integrated survey framework shown in �gure 3 below addition to the core questions. provides an overview of how the annual and periodic sur- Table 3 provides an overview of a survey framework based on veys are connected in the data system. Note that within-year replicated samples that are surveyed each year for the annual surveys can also be conducted using subsamples from the core data items. In addition, each year the core questionnaire annual survey. contains a set of supplemental questions for one of the sub- The survey framework also takes into account the additional ject matters that round out the minimum set of core data. data sources that need to be included in the overall frame- The above survey design provides a framework to collect work. These include: data for core items—some annually, others on a 4-year rotat- a. Administrative data. Governmental interventions ing cycle. Each country will need to make its own decisions such as subsidies, regulation, and legislation often about the content of each of the components. Once the above require agricultural holders to report production design is in place, the next consideration is whether some information. Land ownership and cadastral surveys of the data collections for the annual core items should take provide useful information for constructing regis- place more frequently during the year. One example would be ters. Food inspections, animal health inspections, to conduct a midyear survey to determine crop yields before and trade data provide input to the utilization harvest, another to obtain the �nal production and stocks. accounts. The integrated survey framework also offers the opportu- b. Remotely sensed data. These include vegetative nity to compare sample unit data across time, providing a indices that show overall crop conditions and infor- major validation tool to improve data quality. The integrated mation about changes in land cover and use. The framework also provides the opportunity to use alternative survey framework should include the need to provide FIGURE 3: The integrated survey framework Integrated Survey Framework Master Sample Frame Geo referenced to land use Annual Survey(s) Periodic surveys, Household holdings & enterprises (rotating panels), See Table 3 Within year Surveys-optional Data Supply and Utilization, income, & Management Environmental accounts, System Food Balances, other indicators Administrative Remote Agri Expert Community Data Sensing Businesses Judgment Surveys E C O N O M IC AND S E CT OR WORK 26 T HE SE CO N D PI LLA R —INTEGR ATING A GR ICULTURE INTO NATIONA L STATIS TIC A L S Y S TEM S ground truth data if remote sensing information is to along with the supporting administrative and other be used to estimate cropland areas. data sources. Not all survey results are published; c. Agribusinesses are the source of utilization data and however, they should be available for research and prices. analysis purposes. As described above, the sample d. Expert judgment and windshield surveys can be used and survey design enables the use of ratio and re- to collect data from experts whose judgments inform gression estimators requiring links to previous data. evaluations of agricultural conditions. For instance, iii. Build on the capabilities provided by the master the Sourcebook (World Bank 2008b) refers to a sample frame’s link to land use. The data manage- procedure in which experts travel a speci�ed route ment system should provide for the storage and on a periodic basis and record the condition of crops, maintenance of the farm and household survey data which provide an input into crop yield forecasts. and for the link between the different sets of data e. Community surveys. The World Programme for that are georeferenced to a common land use. For the Census of Agriculture (FAO 2005b) provides an example, there will be �ve consecutive years of core overview of data that can be collected at the village production data for the same sample units plus data level. These data include information about the infra- from the rotating panel surveys. The strength of the structure and services available to households and integrated survey system will come from the data agricultural holdings, occurrences of food shortages, analysis capabilities provided by this data set. frequency of natural disasters, etc. The data management system must also encompass the other data sources depicted in �gure 3. These are neces- The integrated survey framework will provide annual data for a sary for the compilation of supply and utilization accounts, core set of items on agricultural production and other variables food balance sheets, and other economic and environmental determined by the national statistical system. The survey accounts. The use of these accounts provides a means to framework enables longitudinal analysis of the core data, and ensure the consistency of data from different sources. At the it provides links to the data that are collected regarding eco- same time it helps to integrate agricultural statistics into the nomic, environmental, and social issues. The use of the mas- national statistical system by compiling them in parallel with ter sample frame ensures that data collection is connected indicators from other sectors that follow the same concepts, to land use as well. The remaining pillar of integration is the de�nitions, classi�cations, and accounting methodology. management of the data to maximize their use for analysis. The value of the integrated database will increase over time as the database itself grows. It will enable more analyses across THE DATA MANAGEMENT SYSTEM time, and it can be used to improve data quality by comparing The data management system ful�lls three functions—access survey information with census data or between surveys over to of�cial statistics for dissemination purposes; storage and time. The output of the aggregated values will be the input to retrieval of survey results; and access to farm, household, Country STAT,2 following its methods and principles. and georeferenced data for research. The data management The integration of agriculture into the national statistical sys- system should: tem through the implementation of a master sample frame, i. Support the dissemination of data to ensure the of�- an integrated survey framework, and an integrated database cial statistics are readily available, clearly identi�ed by will require countries to review their current governance source and time, and are comparable for aggregation structures. Some countries will have to make changes in purposes, both within and across countries. If more order to meet the challenges of coordination and to ensure than one institution is involved in the national statisti- that the statistical system is sustainable. cal system, there should either be a single database, or the databases should be coordinated to avoid duplication of of�cial statistics. Such duplication can lead to different numbers, causing confusion among 2. Country STAT is a Web-based information technology system those using the data. These data should become part for food and agriculture statistics at the national and subnational levels that provides decision-makers access to statistics across of FAOSTAT, the FAO statistical data base, which thematic areas such as production, prices, trade, and consump- becomes a public good for data access. tion. The FAO forms partnerships with statistical of�ces and ministries of agriculture, �sheries, and forestry to introduce the ii. Provide the framework for the storage of the aggre- system and build the national capacity to use it. www.fao.org/ gated survey results and georeferenced land use data economic/ess/countrystat/en/. GLOB A L STR ATEGY TO IMPR OVE A GR ICULTURA L A ND R UR A L S TATIS TIC S T H E T H I R D P IL L AR 27 Chapter 5: THE THIRD PILLAR—THE SUSTAINABILITY OF AGRICULTURAL STATISTICS THROUGH GOVERNANCE AND STATISTICAL CAPACITY BUILDING The third pillar of the Global Strategy is to establish the gov- units are already in place within the national statistical sys- ernance and capacity that are the foundations of sustainable tems and coordinate data collection and dissemination as statistical systems. The sustainability of a statistical system part of their larger responsibility for agricultural and other depends on stable and predictable funding that ensures statistics, coordination mechanisms may be required to ongoing support for data collection at appropriate intervals. ensure that the statistical system is fully meeting the Policy makers and others who use the data are more likely needs of line ministries. In most countries, however, sta- to support the system that provides it and to sustain their tistical responsibilities are decentralized, and ministries of demand for the data when it proves to be authoritative and agriculture produce the agricultural statistics. Both cen- relevant to their needs. In this way, the sustainability of a tralized and decentralized systems have advantages and statistical system is largely a function of the demand for the disadvantages. National statistical of�ces have experience data it produces and of the �nancial support that is required applying statistical methods and using sample frames— to satisfy that demand. experience that other ministries often lack. However, other ministries are likely to have greater technical knowledge The current situation of poor-quality data leads to their lim- about agriculture, forestry, �sheries, and land use. The ited use within countries and by the international community; purpose of the Global Strategy is to provide a framework this is an important underlying factor explaining the lack of for integration that builds off the strength of both areas of �nancial support for agricultural statistics. Understanding the expertise. demand for statistical information at the national level and what is required to supply that information is, therefore, a Governance at the national level involves the organization of key element of the sustainability of an agricultural statistics a national statistical system that includes sector ministries system. Demand can be supported and strengthened if the and other agencies that provide data. In the case of agricul- statistical system is responsive to users and provides statis- tural statistics, this will include the ministries responsible for tics that are relevant, accessible, timely, and with a level of agriculture, forestry, �sheries, and any other institutions that accuracy that meets their needs. collect agriculture-related data. While donor funding and support will continue to be essential A coordination mechanism is employed to ensure that to improve national statistical systems, the collection of core the different data producers adhere to a common set of data should, over time, become sustainable using national re- standards. Their compliance with these standards prevents sources. The integration of agriculture into national statistical duplications of efforts and resources as well as the publica- system will require many countries to develop an adequate tion of conflicting data from different reporting agencies. It governance structure and to build statistical capacity across also ensures statistical integrity by making the data avail- the different institutions concerned. able and accessible. The coordination mechanism should provide a common voice for seeking resources for the agricultural statistics system within the framework of the GOVERNANCE national statistical system. The governance it provides Because multiple governmental organizations are usually should enable the ministries and agencies involved in the involved in the collection of data on agriculture, forestry, collection of agricultural data to integrate agriculture into and �sheries, most countries will require a statistical the preparation of the national strategies for the develop- coordinating authority. Even in countries where centralized ment of statistics. E C O N O M IC AND S E CT OR WORK 28 TH E TH IR D PILLA R A governance body such as a national statistics council should particularly among policy makers and other data be established to organize the efforts of statistics stakehold- users. The goal is to increase country ownership of ers. Such a council would include the ministry of agriculture, the planning process to produce statistics and the the national statistical of�ce, and other organizations provid- outcome. ing statistics or administrative data to jointly organize and d. Provide common standards, salary scales, and coordinate the development and use of the master sample professional requirements across the organizations frame, the integrated survey framework, and the database. It in the national statistical system for agricultural may be determined that certain ministries are best suited for statistics. activities such as those involving the master sample frame e. Determine who does what in developing and main- or the collection of speci�c types of data. All data collected, taining the master sample frame, in determining the whichever ministry or agency collects them, will be based framework of the integrated surveys to be con- on the master sample frame in an integrated survey system ducted, and in assigning responsibility for the data with the outcomes stored in an integrated database. The role management system. of each institution should be clearly de�ned and build on its f. Reach agreement on the content, scope, and cover- knowledge and technical expertise (crops, livestock, aquacul- age of data, and how frequently it will be provided ture and �shery, forestry, land, and water). by the national statistical system in addition to the The integration of agricultural statistics into a country’s core data based on policy maker and other data user national statistical system does not mean that all responsi- requirements. bilities fall on the national statistical of�ce, the ministry of g. Establish a framework to ensure that the provision agriculture, or any other agency in particular. It does, how- of data is user driven and responsive to user require- ever, mean that the organizations with overlapping data ments for timeliness and quality. needs accept the master sample frame, integrated survey h. Work with the FAO, other international organizations, framework, and database principles. and donors to prepare a detailed assessment of the current national capabilities, and prepare a framework For international organizations, the integration of agriculture for statistical capacity building. into the national statistical system has several implications. i. With input from international and regional organiza- They will have to present their data requests to the national tions, determine funding requirements for capacity statistical system rather than to the individual sector institution building, development of the master sample frame, concerned. The consolidation of data requirements between and the costs necessary to sustain a survey system. multiple international organizations would reduce the amount of overlap and minimize the data reporting requirements. Integrating agriculture into the national system will change the focus of statistical capacity building. This capacity build- The Strategy has implications for donor organizations, includ- ing currently focuses mainly on national statistical of�ces. ing those that support statistical capacity building. Again their Including agricultural statistics into the national statistical efforts will need to focus on the governance structure each system makes those statistics a primary element of the country has organized rather than going directly to individual national strategies for the development of statistics. sectors. A country’s national statistics council will need to deal with STATISTICAL CAPACITY BUILDING the following cross-cutting and coordination functions: The statistical capacity building component of the implemen- a. Prepare or revise national strategies for the develop- tation plan of the Global Strategy should take into account ment of statistics, identifying the respective roles of the quality of agricultural statistics as a function of their ac- each organization in the national statistics council. curacy, relevance, timeliness, comparability, availability, and b. If necessary, implement or revise legislation accessibility. These different dimensions of quality all need regarding the authorities and responsibilities for to be incorporated into the design of systems for agricultural statistics—including legislation and regulations statistics; they should all also be addressed in efforts to build regarding con�dentiality of data. capacity for data collection and analysis. The seasonal na- c. Develop a strategy to foster public support for the ture of agricultural production presents a special challenge funding that a sustainable statistical system requires, for agricultural statistics, especially for the accuracy and GLOB A L STR ATEGY TO IMPR OVE A GR ICULTURA L A ND R UR A L S TATIS TIC S T H E T H I R D P IL L AR 29 timeliness quality dimensions. The often conflicting needs at country, regional, and international levels to identify pri- for accuracy and timeliness can be dif�cult to reconcile. The ority areas, resources required, and timeframes. For many need for some types of data such as timely crop production developing countries, assistance from donor agencies estimates minimizes the time available for data collection and technical cooperation agencies will be needed to sup- and analysis, which has implications for accuracy. port capacity building. This capacity building begins with support to: Carrying out the Strategy will require levels of expertise that can be dif�cult to �nd (or to maintain) in many developing Develop national strategies for the development of countries. The use of remote sensing technologies, the statistics; where such strategies are in place, review design of an integrated survey framework, and the use of them to determine where revisions are needed. a data management system require experienced technical Build a network of statisticians and supporting staff personnel. While building and maintaining technical capac- including data collectors. ity in countries will be problematic, there are possibilities. Educate staff on statistical methodology for sampling, One of these is to establish regional centers of excellence survey design, data compilation, and data analysis. that can provide remote sensing capabilities, develop statisti- Develop and maintain the master sample frame, cal methods, and guide the implementation of information implement the new survey framework, and develop technologies in providing support to national institutions. The the data management system. establishment of these centers could be a focal point for sup- Provide computers, software, and other technical port from donors and international organizations. equipment. The success of the Global Strategy will require a national Provide the satellite imagery georeferenced by land and international effort and commitment to implement the use. statistical capacity building required to rebuild the statistical Disseminate the results and respond to requests. systems in some countries and to make improvements in others. The implementation of the Strategy should build on The Global Strategy is a long-term plan that will face many chal- a detailed country assessment that de�nes speci�c actions lenges and require a concentrated effort from all stakeholders. E C O N O M IC AND S E CT OR WORK S U M MA RY OF R E C OMME NDAT IONS AND T H E WAY FORWA R D 31 Chapter 6: SUMMARY OF RECOMMENDATIONS AND THE WAY FORWARD SUMMARY that the implementation plan should include a well-targeted The Global Strategy to Improve Agricultural and Rural research agenda to support the implementation of the sta- Statistics addresses the declining state of both the quantity tistical methodology required by the strategy and also the and quality of data in developing countries. In addition, it development of methodological guidelines for speci�c condi- responds to increasing demand from policy makers and the tions such as small-scale agriculture, agriculture under dif- donor community for information to deal with sky-rocketing �cult conditions, and nomadic populations. The Commission food prices and emerging issues relating to the use of bio- recognized that the implementation of the Global Strategy fuels, the environment, climate change, and monitoring the will require the mobilization of resources and technical sup- MDGs. port from countries, the donor community, and international organizations. The Global Strategy broadens the scope of agricultural sta- tistics to include aspects of rural households, �sheries, and The Commission, in its endorsement of the integration of forestry. agriculture into the national statistical system, extended the scope of the integration to include other statistical domains, This broadened scope of agricultural statistics led to the basic including macroeconomic statistics and national accounts. fundamentals of the Global Strategy that are based on three pillars: (i) the identi�cation of a minimum set of core data; (ii) the integration of agriculture into the national statistical sys- THE WAY FORWARD tem by the development of a master frame for agriculture, The endorsement of the Global Strategy by the international its use in an integrated survey system, and the implementa- statistical community led the UN Statistical Commission to tion of a data management system; and (iii) the sustainability direct its Friends of Chair working group and the FAO to of agricultural statistics through governance and statistical develop the implementation plan which should reflect the capacity. varying capabilities of the countries. Because of the current poor state of agricultural statistics, This effort should begin with a complete assessment of the the Global Strategy essentially provides the basis for a statistical capabilities of each country, the data they currently re-engineering of the statistics system. This will require a provide, and their readiness to begin to implement the com- renewed commitment from the countries and international ponents of the Global Strategy. organizations to build statistical capacity. The National Strategies for the Development of Statistics The United Nations Statistical Commission at its 41st session (NSDS) should be reviewed and, where necessary, be re- in February 2010 welcomed the Strategy and endorsed its vised to reflect the integration of agriculture into the national technical content and strategic directions. The Commission statistical system and to also reflect the implementation of urged the expedited development of an implementation plan, the master sample frame, the integrated survey framework, including taking the steps necessary to develop the master and the data management system. sample frame, the integrated survey framework, and the data management system. Based on the assessment of the statistical capabilities and the revised NSDS, a comprehensive training and techni- With regard to the implementation plan, the Commission cal assistance program plus a research agenda should be further recommended that a comprehensive technical as- outlined and identify the responsibilities of the different sistance and training program be established. It also stated stakeholders. E C O N O M IC AND S E CT OR WORK 32 S UMMA RY OF REC OMMEND ATIONS A ND TH E WAY FORWA R D The implementation plan should provide the framework for meeting of the International Statistical Institute Conference countries to prepare detailed action plans within a period of 6 on Agricultural Statistics (ICAS-V). to 12 months after the Strategy is launched. The Global Strategy provides a ground-breaking effort to im- Because of the importance of the Global Strategy, the U.N. prove agricultural statistics that has important implications Statistical Commission will continue to be engaged and for other sectors in the national statistical system. While it requests a progress report be made to its 42nd meeting in took many years for agricultural statistics to deteriorate to February 2011. the current situation, the implementation provides a fresh The Global Strategy and implementation planning efforts start. The Global Strategy should be considered a living docu- should provide the main focus for the October 2010 ment that will be revised on a need basis. GLOB A L STR ATEGY TO IMPR OVE A GR ICULTURA L A ND R UR A L S TATIS TIC S ME N U O F INDIC AT ORS F OR AGRICULT URA L STATI STICS 33 Annex A. MENU OF INDICATORS FOR AGRICULTURAL STATISTICS TABLE 1: Menu of indicators, data requirements, data sources and technical notes3 INDICATOR DATA REQUIREMENTS DATA SOURCES TECHNICAL NOTES SECTOR WIDE INDICATORS FOR AGRICULTURE AND RURAL DEVELOPMENT 1 Gross domestic product Censuses and surveys of �rms, farms, Value added should include unreported activities as (GDP). and households for small holders. well as the value of informal or small-scale opera- tions. Annual estimates between census or surveys based on extrapolations based on other indicators. 2 GDP growth from agricul- Estimates of total production and Censuses and surveys, agricul- SNA concepts followed. Problems include estima- ture value added value for all commodities produced in tural enterprises, farm and rural tion of output consumed by the household and the country, including that from small households, and administrative and the annual coverage of all commodities for which holders and household plots minus processor data. only periodic census data are available. Annual estimates of the cost of inputs such estimates made using previous census and other as seed, feed, energy, fertilizer, labor, administrative data if available. etc. Agriculture includes forestry and �sheries. 3 Amount of public Government budget allocations and Ministry of �nance, national The de�nition for public spending on agriculture spending on agricul- spending related to agriculture. accounts, planning commissions, should follow the U.N. Classi�cation of Functions of ture, subsidies, and Agriculture includes forestry and donor reports. Government (COFOG) for agriculture. infrastructure �sheries. 4 Amount of public spend- Government budget allocations and Ministry of �nance, national Rural de�ned using national description. ing on rural infrastructure, spending related to rural areas. accounts, planning commissions, and health, and education donor reports. 5 Change in investment in Inventories of machinery and equip- Agricultural resource surveys of hold- Machinery and equipment inventories should be by capital stock ment owned by agricultural holdings, ings and agricultural enterprises. purpose (tillage, harvesting, etc.) and size. buildings such as milking purposes, animal breeding stock, area of semi- permanent crops such as trees and vineyards, number of trees and vines. 6 Demographics of agricul- Rural population and number of rural Census of Population, Census of Rural de�ned using national description tural and rural population households, number of agricultural Agriculture, Household surveys, households and population living administrative records in them, age and education levels. Agriculture includes forestry and �sheries 7 Rural poor as a percent of Household income and consump- Household surveys. International Countries should use poverty estimates based on total poor population tion estimates for national and rural Comparison Program for comparisons PPPs and extrapolate between ICP benchmarks. poverty lines. Purchasing Power across countries. Parities (PPPs) for comparisons across countries. 8 Rural hungry as a percent Household income and food Household surveys. International Countries should use hunger estimates for monitor- of total poor population consumption estimates for national Comparison Program for comparisons ing food deprivation levels. minimum energy requirements. across countries. 9 Food production index Area, production, and yield for food Agricultural census, surveys of agri- Follow FAO guidelines for inclusions and exclusions. crops, livestock numbers, and produc- cultural enterprises, processors, �sh tion of meat, milk, eggs, �sh captured landings, administrative data such as and cultured, and other food prod- imports and exports. Food balances ucts, nonfood use of food products, and household consumption surveys. and food imports and exports. 3 Indicators should be disaggregated by gender. continued E C O N O M IC AND S E CT OR WORK 34 MENU OF IND IC ATOR S FOR A GR ICULTUR A L S TATIS TIC S TABLE 1: Menu of indicators, data requirements, data sources and technical notes1 (Continued) INDICATOR DATA REQUIREMENTS DATA SOURCES TECHNICAL NOTES 10 Change in value of Imports and exports—quantities and Customs inspections—in some National statistical of�ces should collaborate with trade—imports and values of agricultural products includ- countries the customs of�ces collect customs of�cials to ensure coding and classi�ca- exports ing �shery and forest products. the data, which then are turned over tions follow international guidelines. to the national statistical of�ce for compilation. INDICATORS FOR AGRICULTURAL SUBSECTORS AND RURAL AREAS 11 Productivity of crop Quantity harvested per unit of area Census of agriculture, crop-cutting Dif�cult to measure with multi-cropping or with production as measured such as hectare and area har- surveys. Production sample surveys, crops that can be harvested more than once a year. by crop yields vested. Area harvested distinguished processor surveys, such as oil seed Crop cutting can over estimate yields. between irrigated harvested crops crushers and cotton ginners. and rainfed harvested crops. 12 Change in components of Area harvested, quantity harvested, Surveys of agricultural enterprises, Crop balances should reflect the growing cycle and crop balances quantities imported or exported, administrative data on trade, proces- marketing year, which could be different from the change in stocks, quantities by sors by utilization, and household calendar year. utilization such as food, biofuels, own surveys for own consumption. consumption for every crop including those produced for �ber and oil. 13 Livestock value added Estimates of quantity and value of Surveys of agricultural holdings, Own consumption should be included, dif�cult to production of meat, poultry, milk, enterprises such as slaughter plants, measure. eggs, by-products such as hides and dairies, and processors. Household skins and wool mohair minus costs surveys for own consumption of inputs such as feed and replace- ment stock. 14 Change in components Number of animals born, acquired, Surveys of agricultural holdings at Data collection intervals should reflect the repro- of livestock and poultry slaughtered, and deaths from dis- least annually but more often for spe- ductive cycles. This suggests annual for cattle, balances by species ease. Number of animals by purpose cies with more frequent births during semiannual for pork, and quarterly or shorter for such as breeding, meat, milk, wool, a reference period. This ranges from poultry and milk. and by age breakdowns relevant to annually for cattle to monthly for egg specie (see FAO 2010 Census). production. 15 Change in productivity of Quantity of �sh taken by unit of �sh- National �shery surveys, surveys at capture �sh production ing effort; scienti�c estimates of �sh landing sites, onboard observers, stock and exploitation rates. national, regional, and global assess- ment results. 16 Change in productivity of Estimates of quantity and value of Surveys of aquaculture enterprise, aquaculture production of �sh by species minus and holdings, aquaculture census, costs and quantity of inputs such as market certi�cations, seed, feed, and fertilizers. 17 Change in components of Quantities and value of captures from National �shery surveys, �shery See CWP Handbook and FAO coding and �sh balances coastal and offshore waters, rivers, census, aquaculture census, classi�cation. and lakes including nonlanded catch; surveys of �shery and aquaculture quantities and value of products from enterprises, processors, market aquaculture; utilizations including information, and administrative and own consumption and discards, inspection sources. imports and exports. 18 Change in components of Quantity and value of removals of Appropriate ministries, satellite forestry balances products from forested areas and imagery, price surveys, or processor respective utilizations. data. 19 Commodity price indexes Market reports of prices being Market observers, surveys of enter- Care needed to ensure units of measure for pricing offered by commodity and location. prises, agro-enterprises purchasing are comparable. Prices received by the enterprise at commodities from agricultural the �rst point of sale. enterprises. 20 Consumer price indexes Monthly or seasonal prices paid by Consumer price index. Care is needed to ensure highly seasonal products the consumer. do not distort the price series. 21 Early warning of change Monthly or seasonal prices paid by Windshield surveys of crop condi- These do not have to be statistically rigorous, in food security. the consumer. tions, amount of precipitation, satel- mainly to provide an early warning that other lite imagery of vegetative indexes, interventions are needed. changes in trade data, and animal disease outbreak. GLOBA L STR ATEGY TO IMPR OVE A GR ICULTURA L A ND R UR A L S TATIS TIC S ME N U O F INDIC AT ORS F OR AGRICULT URA L STATI STICS 35 TABLE 1: Menu of indicators, data requirements, data sources and technical notes1 (Continued) INDICATOR DATA REQUIREMENTS DATA SOURCES TECHNICAL NOTES CLIMATE CHANGE, LAND, AND THE ENVIRONMENT 22 Change in land cover Land Cover Classi�cation System Land use surveys, satellite imagery. Ground truth data required to provide more detailed and use (LCCS), area and georeferenced for Georeferenced data on economic breakdowns of cultivated land, especially for crops cultivated land, grass or pasture, situation of agricultural holdings in small plots. Dif�cult to apply in detail where inland water, marine water, wetlands, needed to understand effect of policy multicropping is used. shrubland, woodland, fallow or idle decisions on land use. cultivated land, barren land, urban or developed areas, areas equipped for irrigation. 23 Change in proportion of Area georeferenced to map materials. Ministry responsible for forestry, Follow LCCS classi�cation. land area covered by for- satellite imagery. ests, rate of deforestation 24 Percent of land and water Land and water area and georefer- Responsible ministry—satellite Follow LCCS coding with expansion covering inland area formally established enced to mapping material. imagery. and marine water bodies. as protected areas 25 Irrigated land as percent Total cropland and area irrigated by Agricultural census, other crop- Irrigation refers to the arti�cial application of water of total cropland source of water for irrigation (surface related surveys or water-user survey. to assist in the growing of crops (and pastures). water, groundwater, treated waste- Can be done by letting water flow over the land water, etc.) and by method (surface, (“surface irrigation�), by spraying water under pres- sprinkler, localized irrigation). sure over the land concerned (“sprinkler irrigation�), or by bringing it directly to the plant (“localized Productivity of irrigation Crop yields from irrigated land irrigation�). compared to yields from nonirrigated areas. 26 Withdrawal of water for Area under irrigation, number of Appropriate ministries, special stud- Should include both surface and ground water. agriculture as a percent of irrigations, irrigation intensity ies or surveys to estimate water use Coding and classi�cations should be de�ned. total water withdrawal and requirements by crop, water in agriculture and aquaculture, and withdrawal and turnover rate for surveys of aquaculture enterprises aquaculture consumption, and per and holdings. capita consumption by people and animals. 27 Change in soil loss from Reduction in crop yields, reduction in Appropriate ministries, georefer- watersheds area of cultivated land. enced data with satellite imagery. 28 Change in affect of inputs Fertilizer, pesticide, and other Agricultural census and or follow-up Data should be georeferenced to land cover and on the environment chemicals applied to the soil, water surveys to measure fertilize and use. bodies, and plants by type of crop and chemical use, tillage methods. watershed area, stocking. THE AGRICULTURAL AND RURAL ECONOMY 29 Number of family and Include unpaid labor of the operator Labor force surveys of holdings. Need to establish standards for minimum ages of hired workers on the of the holding and family members workers and the number of hours worked per week holding plus number of hired workers. to be considered a worker. Need to de�ne reference period. Need to ensure female workers are counted. 30 Number of household The employment status for work Labor force surveys—household Need to distinguish de�ned employment from members employed by off the agricultural holding for each surveys. unpaid household service work such as domestic farm and nonfarm household member. chores. 31 Change in farm and Income to the household by sector, Rural household survey. Rural to be de�ned using national de�nitions. rural nonfarm household crop, livestock, etc. Income from income from all sources investments or employment outside the agricultural holding. 32 Percent of rural popula- Total number of rural households, Central bank or commercial banks, tion using services of for- number using credit or savings special surveys, agricultural census. mal banking institutions services. 33 Change in sales of agro- Sales, net pro�ts of enterprises Special surveys. Use standard accounting principles. enterprises providing services to agriculture. E C O N O M IC AND S E CT OR WORK E X A MP L ES OF S AMP L E F RAME S USE D F OR A G R I CULTURA L STATISTICS 37 Annex B. EXAMPLES OF SAMPLE FRAMES USED FOR AGRICULTURAL STATISTICS Population census enumeration areas. The population the agricultural census to develop registers of farms. This census is usually conducted using an administrative struc- provides a powerful sampling tool because it allows a choice ture in which cartographic or other mapping materials are of many alternative sampling designs. A major weakness is used to divide the country into enumeration areas, which is that the registers rapidly become out-of-date. Out-of-date the �rst level of data aggregation. Depending on the coun- population and farm registers erode all of the data quality try’s capabilities, the only results from the population cen- dimensions because the completeness of coverage changes sus in some countries are the enumeration area totals for over time, thus affecting the comparability and accuracy of numbers of people, households, and so on. Therefore, the the resulting estimates. sampling frame is basically the listing of enumeration areas Registers of farms based on administrative sources and associated aggregated data from the census. Random such as business registrations or tax collections. This samples of enumeration areas are selected and screened process is used in some developed countries. It offers the for households from which subsamples are selected for advantages of the registers from the agricultural census, but household surveys—a two-stage sampling process. Some again, it needs to be updated regularly. A disadvantage of the countries use their administrative structure of counties, administrative sources is that they may not include the total townships, and villages as their framework for the census, population, especially units below a threshold required to be with the village becoming the enumeration area. Villages are registered or pay taxes. In other words, while they will be in- also used as a �rst-stage sampling unit in countries where the clusive of commercial farms, they are not likely to include village is where the farm households are generally located. small-scale farms and subsistence farming units. Household registers from the population census. Area sample frames. An area sample frame is the land Countries with statistical capacity are able to develop a regis- mass of the country or the space within a country containing ter of all households included in the population census. The the populations of interest. Both maps and satellite images list of population households is the sample frame used for are used to divide the country into administrative areas such household surveys. One problem is that the list of households as provinces, districts, and so forth. Satellite imagery can becomes out-of-date with households changing or dissolving be used to subdivide the administrative areas into land-use and new households being formed. Unless administrative categories such as cropland, rangeland, woodlands, urban data or other means are used to keep the population register areas, and so on. Sampling units of segments of land with up-to-date, survey results contain an increasing coverage identi�able boundaries can be formed, or each land-use bias over time. stratum can be divided into square grids with a sample Agricultural census enumeration areas. In many countries, of points becoming the sampling units. During the data- the cartographic materials and data from the population cen- collection process, rules of association are used to connect sus are used for the agricultural census. The sampling frame farm holdings or households to the segments or points. An consists of enumeration areas and aggregated data from the area frame is suitable for obtaining information about vari- census data collection. As in the population census, random ables associated with land such as crops, livestock, forests, samples of enumeration areas are selected and screened and water. Depending on the process used, area frames for farms or agricultural holdings for agricultural production can be costly and time consuming to construct. However, surveys. recent innovations using satellite imagery and two-stage sampling of points have reduced both the cost and time. An Registers of farms from the agricultural census. As in advantage of an area frame is that the frame does not go the household registers, countries with the capacity can use out-of-date; it is complete in its coverage, and provides a E C O N O M IC AND S E CT OR WORK 38 EX A MPLES OF SA MPLE FRA MES US ED FOR A GR ICULTUR A L S TATIS TIC S basis to georeference survey data with the underlying land Multiple frames. A combination of the above frames is use. It also provides ground truth useful for classifying satel- used, often involving the use of an area frame in conjunc- lite imagery by land cover. The primary disadvantage of area tion with one of the list frames, to take advantage of the frames is that the sampling is based on land use and not strengths and weaknesses of each. The United Nations Food on the size and type of agricultural holding. Sampling vari- and Agricultural Organization (FAO 1998) provides an over- ability becomes a problem if there is a large range in size of view of multiple-frame sampling. This is appropriate where the agricultural holdings. A summary of the methodology of there is a large variation in the sizes and types of agricultural area-frame sampling is provided by Gallego (1995). Another holdings with a subset of large commercial farms. The list of disadvantage is that data-collection costs exceed those commercial farms can be strati�ed by size and type, and the based on registers where telephone or mail can be used area frame ensures the population is completely covered by instead of personal interviews. providing coverage of the small and subsistence farms. GLOBA L STR ATEGY TO IMPR OVE A GR ICULTURA L A ND R UR A L S TATIS TIC S REFERENCES 39 REFERENCES Carletto, Gero, 2009. Improving the Availability, Quality and Policy Norton, George and Jeffrey Alwang, 2004. “Measuring the Bene�ts Relevance of Agricultural Data: The Living Standards Measurement of Policy-Oriented Social Science Research: Evidence from Survey—Integrated Surveys on Agriculture. Development Developing Countries,� chapter 10 in Pardey, P. and V. Smith Research Group, the World Bank, Washington, DC. (eds.), Whatís Economics Worth?: Valuing Policy Research. International Food Policy Research Institute. Baltimore: Johns FAO, 1998. Multiple Frame Agricultural Survey, Volume 2, Agricultural Hopkins University Press, 2004. Survey Programs Based on Area Frame or Dual Frame (Area and List) Sample Designs, Food and Agricultural Organization, Rome, PARIS21, 2009. “Assessing the Inclusion of the Agricultural Sector Italy. in National Strategies for the Development of Statistics,� ISI Satellite Meeting on Agricultural Statistics, Maputo, Mozambique, FAO, 2005a. Land Cover Classi�cation System, Classi�cation August 2009. Concepts and User Manual, Software version 2, Food and Agriculture Organization, Rome, Italy. PARIS21, 2007. “A Guide to Designing a National Strategy for the Development of Statistics,� Partnership in Statistics in the 21st, FAO, 2005b. System of Integrated Agricultural Censuses and Century, PARIS21, PARIS21 Secretariat. Surveys, World Programme for the Census of Agriculture 2010, Food and Agriculture Organization, Rome, Italy. Steiner, Michael, 2005. Sample Design for Agricultural Surveys in China, Proceedings of the 55th Conference of the International FAO, 2006. Independent External Evaluation of the FAO, Inception Statistical Institute, Sydney, Australia. Report submitted to the Council Committee for the Independent External Evaluation (CC-IEE) by the Independent External United Nations 1994. Fundamental Principles of Of�cial Statistics, Evaluation Core Team, May 2006.Food and Agricultural U.N. Statistics Division, New York. Web site: unstats.un.org/ Organization, Rome, Italy (www.fao.org). unsd/methods/statorg/fp. FAO, 2008a. Independent External Evaluation of FAOís Role and United Nations, 2007. The Wye Group Handbook on Rural Work in Statistics, Prepared by an Evaluation Team and Expert Households Livelihood and Well-Being, jointly produced by the Panel following the Independent External Evaluation of the FAO, U.N. Economic Commission for Europe, Eurostat, FAO, OECD, Food and Agriculture Organization, Rome Italy, www.fao.org. and the World Bank. FAO, 2008b. Terrestrial Essential Climate Variables for Climate United Nations, 2008. Wye Group on Statistics on Rural Development Change Assessment, Mitigation, and Adaption, Natural and Agricultural Household Income—Web site: unstats.un.org/ Resources Management and Environment Department, Food unsd/methods/citygroup/wye. and Agriculture Organization, Rome, Italy. World Bank, 2008a. World Development Report, Agriculture Gallego, F. J., 1995. Sampling Frames of Square Segments, Report for Development, Of�ce of the Publisher, the World Bank, EUR 16317, Of�ce of Of�cial Publications of the European Washington, DC. Communities, Joint Research Center, Ispra, Italy. World Bank, 2008b. Tracking Results in Agriculture and Rural Gardner, Bruce, 2004. “Returns to Policy-Related Social Science Development in Less-than-Ideal Conditions—A Source Book Research in Agriculture,� chapter 9 in Pardey, P. and V. Smith of Indicators for Monitoring and Evaluation, the Global Donor (eds.), Whatís Economics Worth?: Valuing Policy Research. Platform for Rural Development, the FAO, and World Bank, Sector International Food Policy Research Institute. Baltimore: Johns Manager, Agriculture and Rural Development Department, the Hopkins University Press, 2004. World Bank, Washington, DC. E C O N O M IC AND S E CT OR WORK A G R I C U L T U R A L A N D R U R A L D E V E L O P M E N T This is a joint publication with: Agriculture and Rural Development (ARD) 1818 H Street, NW Washington, D.C. 20433 USA Telephone: 202-477-1000 Internet: www.worldbank.org/ard United Nations