A G R I C U LT U R E A N D E N V I R O N M E N TA L S E R V I C E S D I S C U S S I O N PA P E R 0 3 83177 FISH TO 2030 Prospects for Fisheries and Aquaculture WORLD BANK REPORT NUMBER 83177-GLB DECEMBER 2013 A G R I C U LT U R E A N D E N V I R O N M E N TA L S E R V I C E S D I S C U S S I O N PA P E R 0 3 FISH TO 2030 Prospects for Fisheries and Aquaculture WORLD BANK REPORT NUMBER 83177-GLB © 2013 International Bank for Reconstruction and Development / International Development Association or The World Bank 1818 H Street NW Washington DC 20433 Telephone: 202-473-1000 Internet: www.worldbank.org This work is a product of the staff of The World Bank with external contributions. The findings, interpretations, and conclusions expressed in this work do not necessarily reflect the views of The World Bank, its Board of Executive Directors, or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Rights and Permissions The material in this work is subject to copyright. Because The World Bank encourages dissemination of its knowledge, this work may be reproduced, in whole or in part, for noncommercial purposes as long as full attribution to this work is given. Any queries on rights and licenses, including subsidiary rights, should be addressed to the Office of the Publisher, The World Bank, 1818 H Street NW, Washington, DC 20433, USA; fax: 202-522-2422; e-mail: pubrights@worldbank.org. Cover Photos: Shutterstock, LLC. CONTENTS iii CONTENTS Foreword . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix Acronyms and Abbreviations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xi Executive Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .xiii Chapter 1: Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .1 1.1. Motivations and Objective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .1 1.2. Lessons from Fish to 2020 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .4 1.3. Strategy of Improving Modeling Framework in Fish to 2030 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .7 1.4. Policy Research Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .8 Chapter 2: Preparing IMPACT Model for Fish to 2030 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 2.1. Basics of IMPACT Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 2.2. IMPACT Model Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 2.3. Data Used and Parameter Specification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16 2.4. Establishing a Consistent Base-Year Picture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21 2.5. Assessing the Quality of Projections . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24 2.6. Issues and Discussions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31 Technical Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33 Chapter 3: IMPACT Projections to 2030 under the Baseline Specification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39 3.1. Production . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39 3.2. Consumption . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44 3.3. Trade and Prices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46 3.4. Fishmeal and Fish Oil . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47 Technical Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50 Chapter 4: IMPACT Projections to 2030 under Selected Scenarios . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55 4.1. Scenario 1: Faster Aquaculture Growth . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55 4.2. Scenario 2: Expanded Use of Fish Processing Waste in Fishmeal and Fish Oil Production . . . . . . . . . . 57 A G R I C U LT U R E A N D E N V I R O N M E N TA L S E R V I C E S D E PA R T M E N T D I S C U S S I O N PA P E R iv CONTENTS 4.3. Scenario 3: A Major Disease Outbreak in Shrimp Aquaculture in Asia . . . . . . . . . . . . . . . . . . . . . . . 60 4.4. Scenario 4: Accelerated Shift of Consumer Preferences in China . . . . . . . . . . . . . . . . . . . . . . . . . . 63 4.5. Scenario 5: Improvement of Capture Fisheries Productivity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66 4.6. Scenario 6: Impacts of Climate Change on the Productivity of Capture Fisheries . . . . . . . . . . . . . . . 68 Chapter 5: Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71 5.1. Main Findings from the Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71 5.2. Discussion and Possible Future Directions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77 FIGURES Figure 1.1: Comparison of Fish to 2020 Projections and FAO Data for Global Food Fish Supply . . . . . . . . . . . . . . . . . . . . .4 Figure 1.2: Evolution of World Food Fish Production, 1984–2009. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .5 Figure 1.3: Average Annual Growth Rates of Capture and Aquaculture Production, 1960–2009 . . . . . . . . . . . . . . . . . . . .5 Figure 1.4: Trends of Real Prices of Selected Seafood . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .6 Figure 1.5: Global Fishmeal Use . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .7 Figure 2.1: Definition of Aggregate Regions for Results Reporting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 Figure 2.2: Schematic of Links of Fish with Fishmeal and Fish Oil in the IMPACT Model . . . . . . . . . . . . . . . . . . . . . . . . . 16 Figure 2.3: Key Data Relationships Used to Balance Fish Data in IMPACT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 Figure 2.4: Computational Steps and Sequence of the Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24 Figure 2.5: Comparison of Projections and Data for Global Fish Production, 2000–08. . . . . . . . . . . . . . . . . . . . . . . . . . 25 Figure 2.6: Comparison of Projections and Data for Global Fish Utilization, 2000–06 . . . . . . . . . . . . . . . . . . . . . . . . . . 25 Figure 2.7: Comparison of Projections and Data for Global Fishmeal Production, 2000–08 . . . . . . . . . . . . . . . . . . . . . . 25 Figure 2.8: Comparison of Projections and Data for Regional Capture Fisheries Production, 2008 . . . . . . . . . . . . . . . . . . 26 Figure 2.9: Comparison of Projections and Data for Capture Fisheries Production by Species, 2008 . . . . . . . . . . . . . . . . . 26 Figure 2.10: Comparison of Projections and Data for Regional Aquaculture Production, 2008 . . . . . . . . . . . . . . . . . . . . 26 Figure 2.11: Comparison of Projections and Data for Aquaculture Production by Species, 2008 . . . . . . . . . . . . . . . . . . . 27 Figure 2.12: Comparison of Projections and Data for Regional Per Capita Food Fish Consumption, 2006 . . . . . . . . . . . . . 27 Figure 2.13: Comparison of Projections and Data for Regional Total Food Fish Consumption, 2006 . . . . . . . . . . . . . . . . . 27 Figure 2.14: Comparison of Regional Projections and Data for Net Fish Export, 2006 . . . . . . . . . . . . . . . . . . . . . . . . . . 28 Figure 2.15: Comparison of Projections and Data for World Fish Prices, 2000–08. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28 Figure 2.16: Comparison of Projections and Data for World Prices of Fishmeal, 1984–2011 . . . . . . . . . . . . . . . . . . . . . . 29 Figure 2.17: Comparison of Projections and Data for World Prices of Fish Oil, 1984–2011 . . . . . . . . . . . . . . . . . . . . . . . 29 Figure 3.1: Global Fish Production: Data and Projections, 1984–2030 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39 F I S H TO 2 030: P R O S P E C T S F O R F I S H E R I E S A N D A Q UA C U LT U R E CONTENTS v Figure 3.2: Volume and Share of Capture and Aquaculture Production in Global Harvest . . . . . . . . . . . . . . . . . . . . . . . 39 Figure 3.3: Average Annual Growth Rates of Capture and Aquaculture Production, 1960–2029 . . . . . . . . . . . . . . . . . . . 40 Figure 3.4: Projected Global Fish Supply by Species . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42 Figure 3.5: Projected Global Aquaculture Fish Supply by Species. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42 Figure 3.6: Projected Change in Real Prices between 2010 and 2030 by Commodities . . . . . . . . . . . . . . . . . . . . . . . . . 47 Figure 3.7: Projected Production and Fishmeal Use in Global Fed Aquaculture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49 Figure 3.8: Projected Average Feed Conversion Ratio for Fishmeal in Global Fed Aquaculture . . . . . . . . . . . . . . . . . . . . 50 Figure 3.9: Comparison of IMPACT and OECD-FAO Projections for Global Fish Supply . . . . . . . . . . . . . . . . . . . . . . . . . 51 Figure 3.10: Comparison of IMPACT and OECD-FAO Projections for Global Aquaculture Production . . . . . . . . . . . . . . . . 51 Figure 3.11: Comparison of IMPACT and OECD-FAO Projections for Global Food Fish Consumption . . . . . . . . . . . . . . . . 52 Figure 3.12: Comparison of IMPACT and OECD-FAO Projections for Global Fishmeal Supply . . . . . . . . . . . . . . . . . . . . . 52 Figure 3.13: Comparison of IMPACT and OECD-FAO Projections for Implied FCR for Fishmeal . . . . . . . . . . . . . . . . . . . . 53 Figure 4.1: Projected Increase in Fishmeal Production due to Usage of Whole Fish and Fish Processing Waste in 2030 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59 Figure 4.2: Global Shrimp Supply under Baseline and Disease Scenarios . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60 Figure 4.3: Impact of Disease Outbreak on Shrimp Aquaculture Production. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61 Figure 4.4: Projected Net Exports of Fish by Region under Accelerated Consumer Preference Shift in China . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65 Figure 4.5: Projected Change in Total Food Fish Consumption in 2030 by Region under Accelerated Consumer Preference Shift in China . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66 Figure 4.6: Global Fish Supply under Improved Productivity, 2000–30 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67 Figure 4.7: Projected Changes in Capture Production in 2030 under Capture Productivity Improvement Scenario . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67 Figure 4.8: Projected Changes in Fish Consumption in 2030 under Capture Productivity Improvement Scenario . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67 TABLES Table E.1: Summary Results under Baseline Scenario (000 tons) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xv Table E.2: Summary Results for Year 2030 under Baseline and Alternative Scenarios . . . . . . . . . . . . . . . . . . . . . . . . . .xvi Table 2.1: Non-Fish Commodities Included in the IMPACT Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 Table 2.2: Fish Products Included in the IMPACT Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12 Table 2.3: Abbreviation Code for Aggregate Regions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 Table 2.4: Summary of Key Variables in the IMPACT Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 Table 2.5: Number of Countries Covered in FAO Fishmeal and Fish Oil Dataset, 1976–2009 . . . . . . . . . . . . . . . . . . . . . . 19 Table 2.6: List of Key Parameters in the IMPACT Model and Their Sources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 A G R I C U LT U R E A N D E N V I R O N M E N TA L S E R V I C E S D E PA R T M E N T D I S C U S S I O N PA P E R vi CONTENTS Table 2.7: List of Countries Used to Define World Prices for Base Year . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34 Table 2.8: Estimated Reduction Ratios . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36 Table 2.9: Waste Ratios Used in the IMPACT Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37 Table 3.1: Projected Total Fish Production by Region . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40 Table 3.2: Projected Aquaculture Production by Region . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41 Table 3.3: Projected Capture Fisheries Production by Region . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42 Table 3.4: Projected Species Shares in Aquaculture Production . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43 Table 3.5: Projected Top Three Fish Producing Regions by Species . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43 Table 3.6: Income and Population Growth Assumptions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44 Table 3.7: Projected Per Capita Fish Consumption by Region . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45 Table 3.8: Projected Total Food Fish Consumption by Region . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45 Table 3.9: Projected Net Exports of Fish by Region . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46 Table 3.10a: Projected Top Three Net Fish Exporting Regions by Species. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46 Table 3.10b: Projected Top Three Net Fish Importing Regions by Species . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47 Table 3.11: Projected Total Fishmeal Production by Region . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48 Table 3.12: Projected Fishmeal Use by Region . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49 Table 3.13: Projected Net Exports of Fishmeal by Region (000 tons) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50 Table 4.1: Projected Effects of Faster Aquaculture Growth on Aquaculture Supply by Region . . . . . . . . . . . . . . . . . . . . 56 Table 4.2: Projected Effects of Faster Aquaculture Growth on Aquaculture Supply and Commodity Prices . . . . . . . . . . . . 57 Table 4.3: Projected Amount of Fish Processing Waste Used in Fishmeal Production by Region (000 tons) . . . . . . . . . . . . 58 Table 4.4: Projected Effects of Expanded Use of Fish Processing Waste in Fishmeal Production on Aquaculture Supply and Commodity Prices. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59 Table 4.5: Projected Impact of Disease Outbreak on Shrimp Aquaculture Production (000 tons) . . . . . . . . . . . . . . . . . . . 61 Table 4.6: Comparison of Projected Net Exports of Shrimp by Region with and without Disease Outbreak (000 tons) . . . . . 62 Table 4.7: Projected Changes in the Food Fish Consumption in China due to Accelerated Preference Shift . . . . . . . . . . . . 64 Table 4.8: Projected Aquaculture Production in 2030 under Accelerated Consumer Preference Shift in China . . . . . . . . . . 65 Table 4.9: Projected Capture Production in 2030 under Baseline and Climate Change Scenarios . . . . . . . . . . . . . . . . . . 69 Table 4.10: Projected Fish Supply in 2030 under Baseline and Climate Change Scenarios . . . . . . . . . . . . . . . . . . . . . . . 69 Table 4.11: Projected Food Fish Consumption in 2030 under Baseline and Climate Change Scenarios . . . . . . . . . . . . . . . 70 F I S H TO 2 030: P R O S P E C T S F O R F I S H E R I E S A N D A Q UA C U LT U R E FOREWORD v ii FOREWORD Feeding an expected global population of 9 billion by 2050 is a daunting challenge that is engaging researchers, technical experts, and leaders the world over. A relatively unappreciated, yet promising, fact is that fish can play a major role in satisfying the palates of the world’s growing middle income group while also meeting the food security needs of the poorest. Already, fish represents 16 percent of all animal protein consumed globally, and this proportion of the world’s food basket is likely to increase as consumers with rising incomes seek higher- value seafood and as aquaculture steps up to meet increasing demand. Aquaculture has grown at an impressive rate over the past decades. It has helped to produce more food fish, kept the overall price of fish down, and made fish and seafood more accessible to consumers around the world. That’s why greater investment is needed in the indus- try—for new and safer technologies, their adaptation to local conditions, and their adoption in appropriate settings. But supplying fish sustainably—producing it without depleting productive natural resources and without damaging the precious aquatic environment—is a huge challenge. We continue to see excessive and irresponsible harvesting in capture fisheries and in aquaculture. Disease outbreaks, among other things, have heavily impacted production—most recently with early mortality syndrome in shrimp in Asia and America. At the World Bank, we hear from the heads of major seafood companies that they want to secure access to reliable and environmentally sustainable supply chains. Matching growing market demand with this private sector interest in reliable and sustainable sourcing presents a major opportunity for developing countries prepared to invest in improved fisheries management and environmentally sustainable aqua- culture. By taking up this opportunity, countries can create jobs, help meet global demand, and achieve their own food security aspirations. There is substantial potential for many developing countries to capitalize on the opportunity that the seafood trade provides. This study em- ploys state-of-the-art economic models of global seafood supply and demand that can be used to analyze the trends and the extent of such opportunities. The insights gained here can inform developing and developed countries alike of the importance and urgency of improved capture fisheries and aquaculture management, so that seafood demand is met in an environmentally and economically sustainable way. Working alongside partners like IFPRI and FAO, the World Bank can support developing countries in their efforts to manage their fish produc- tion sustainably through tailored and innovative solutions that work. Juergen Voegele Director Agriculture and Environmental Services Department World Bank A G R I C U LT U R E A N D E N V I R O N M E N TA L S E R V I C E S D E PA R T M E N T D I S C U S S I O N PA P E R ACKNOWLEDGMENTS ix ACKNOWLEDGMENTS This report was made possible through the contributions and support of many experts. Siwa Msangi (Senior Research Fellow, International Food Policy Research Institute) led the modeling work to add a fisheries component to IFPRI’s IMPACT model and drafted the main text of the report. Mimako Kobayashi (Natural Resources Economist, World Bank) coordinated with the IFPRI team to improve and validate model output and drafted and edited the report. Miroslav Batka (Research Analyst, International Food Policy Research Institute) processed data that went into the model, performed model simulations, and generated figures and tables for the report. Stefania Vannuccini (Fishery Statistician, Fishery Information and Statistics Branch, Food and Agriculture Organization of the UN) constructed the primary raw data set from which the model baseline was built and provided input into the commodity definitions. Madan M. Dey (Professor, University of Arkansas at Pine Bluff ) provided key input in conceptualizing the model structure in the initial stages, contributed to the supply side modeling, and edited the report. James L. Anderson (Adviser, World Bank) provided overall guidance to the report preparation with his expertise in the global seafood market. The authors wish to thank and acknowledge the invaluable contributions by Keiran Kelleher (the former Fisheries Team Leader, World Bank) for initiating the project; the Fishery Information and Statistics Branch of FAO for providing data; participants of the “think-shop” in Rome (FAO), March 7–8, 2011, to build a framework for analysis and elaborate an action plan (James Muir, Richard Grainger, Rohana Subasinghe, Audun Lem, Xiaowei Zhou, Koji Yamamoto, Junning Cai, Sachiko Tsuji, Matthias Halwart); Kehar Singh (University of Arkansas at Pine Bluff ), James Muir (University of Stirling), Pierre Charlebois (Economist, FAO consultant), and Randall Brummett (World Bank) for expert input that helped model specification; and the peer reviewers—Rebecca Lent (U.S. Marine Mammal Commission), Vikas Choudhary (World Bank), and Xavier Vincent (World Bank)—for providing useful comments that helped to improve the report draft. Ann Gordon, Cambria Finegold, Charles C. Crissman, and Alan Pulis (WorldFish) drafted a companion report, “Fish Production, Consumption, and Trade in Sub-Saharan Africa: A Review Analysis.” The report is available upon request. This is the result of a collaborative effort between IFPRI, FAO, the University of Arkansas at Pine Bluff, and the World Bank. Preparation and publication of the report was supported by the Global Program on Fisheries (PROFISH) Multi-Donor Trust Fund, the Trust Fund for Environmentally & Socially Sustainable Development (TFESSD), and the Japan Policy and Human Resources Development (PHRD) Fund Staff Grant. A G R I C U LT U R E A N D E N V I R O N M E N TA L S E R V I C E S D E PA R T M E N T D I S C U S S I O N PA P E R A C R O N Y M S A N D A B B R E V I AT I O N S xi ACRONYMS AND ABBREVIATIONS AFR Sub-Saharan Africa CAPRI Common Agricultural Policy Regionalised Impact Modeling System CHN China CobSwf aggregate of cobia and swordfish CSE consumer subsidy equivalent EAP East Asia and the Pacific, including Mongolia and developed nations, excluding Southeast Asia, China, and Japan ECA Europe and Central Asia, including developed nations EelStg aggregate of eels and sturgeon EEZ exclusive economic zone EwE Ecopath with Ecosim FAO Food and Agriculture Organization of the United Nations FAOSTAT FAO food and agriculture statistics FBS food balance sheets (produced by the FAO) FCR feed conversion ratio FIPS FAO Fisheries Information and Statistics Branch FishStat FAO Fisheries and Aquaculture Statistics FPI Fishery Performance Indicators GAMS General Algebraic Modeling System GDP gross domestic product GDP/c gross domestic product per capita HS Harmonized System IFFO International Fishmeal and Fish Oil Organisation IFPRI International Food Policy Research Institute IMPACT International Model for Policy Analysis of Agricultural Commodities and Trade IND India ISA infectious salmon anemia JAP Japan LAC Latin America and Caribbean MDemersals major demersal fish MM marketing margin MNA Middle East and North Africa MSY maximum sustainable yield A G R I C U LT U R E A N D E N V I R O N M E N TA L S E R V I C E S D E PA R T M E N T D I S C U S S I O N PA P E R xii A C R O N Y M S A N D A B B R E V I AT I O N S NAM North America (United States and Canada) OCarp silver, bighead, and grass carp OECD Organization for Economic Co-Operation and Development OFresh freshwater and diadromous species, excluding tilapia, Pangasius/catfish, carp, OCarp, and EelStg OMarine other marine fish OPelagic other pelagic species Pangasius/catfish Pangasius and other catfish PSE producer subsidy equivalent ROW rest of the world, including Greenland, Iceland, and Pacific small island states RR reduction ratio SAR South Asia, excluding India SEA Southeast Asia WBG World Bank Group WCO World Customs Organization WDR World Development Report All tons are metric tons unless otherwise indicated. All dollar amounts are U.S. dollars unless otherwise indicated. F I S H TO 2 030: P R O S P E C T S F O R F I S H E R I E S A N D A Q UA C U LT U R E EXECUTIVE SUMMARY x iii EXECUTIVE SUMMARY CONTEXT The World Bank Group (WBG) Agriculture Action Plan 2013–151 summarizes critical challenges facing the global food and agriculture sector. Global population is expected to reach 9 billion by 2050, and the world food-producing sector must secure food and nutrition for the grow- ing population through increased production and reduced waste. Production increase must occur in a context where resources necessary for food production, such as land and water, are even scarcer in a more crowded world, and thus the sector needs to be far more efficient in utilizing productive resources. Further, in the face of global climate change, the world is required to change the ways to conduct economic activities. Fisheries and aquaculture must address many of these difficult challenges. Especially with rapidly expanding aquaculture production around the world, there is a large potential of further and rapid increases in fish supply—an important source of animal protein for human consumption. During the last three decades, capture fisheries production increased from 69 million to 93 million tons; during the same time, world aquaculture production increased from 5 million to 63 million tons (FishStat). Globally, fish2 currently represents about 16.6 percent of animal protein supply and 6.5 percent of all protein for human consumption (FAO 2012). Fish is usually low in saturated fats, carbohydrates, and cholesterol and provides not only high-value protein but also a wide range of essential micronutrients, including various vitamins, minerals, and polyunsaturated omega-3 fatty acids (FAO 2012). Thus, even in small quantities, provision of fish can be effective in addressing food and nutritional security among the poor and vulnerable populations around the globe. In some parts of the world and for certain species, aquaculture has expanded at the expense of natural environment (for example, shrimp aquaculture and mangrove cover) or under technology with high input requirements from capture fisheries (for example, fishmeal). However, some aquaculture can produce fish efficiently with low or no direct input. For example, bivalve species such as oysters, mussels, clams, and scallops are grown without artificial feeding; they feed on materials that occur naturally in their culture environment in the sea and lagoons. Silver carp and bighead carp are grown with planktons proliferated through fertilization and the wastes and leftover feed materials for fed species in multispecies aquaculture systems (FAO 2012). While the proportion of non-fed species in global aquaculture has declined relative to higher trophic-level species of fish and crustaceans over the past decades, these fish still represent a third of all farmed food fish production, or 20 million tons (FAO 2012). Further, production efficiency of fed species has improved. For example, the use of fishmeal and fish oil per unit of farmed fish produced has declined substantially as reflected in the steadily declining inclusion levels of average dietary fishmeal and fish oil within compound aquafeeds (Tacon and Metian 2008). Overall, a 62 percent increase in global aquaculture production was achieved when the global supply of fishmeal declined by 12 percent during the 2000–08 period (FAO 2012). 1 It builds on the World Development Report (WDR) 2008 and the subsequent WBG Agriculture Action Plan 2010–12. WDR 2008 reaffirmed that “promoting agriculture is imperative for meeting the Millennium Development Goal of halving poverty and hunger by 2015 and reducing poverty and hunger for several decades thereafter.” It redefined “how agriculture can be used for development, taking account of the vastly different context of opportunities and challenges that has emerged.” The subsequent WBG Agriculture Action Plan 2010–12 provided strategies to operationalize the findings of the WDR 2008. 2 Throughout this report, fish is considered in a broad sense that includes finfish, mollusks, and crustaceans. A G R I C U LT U R E A N D E N V I R O N M E N TA L S E R V I C E S D E PA R T M E N T D I S C U S S I O N PA P E R xiv EXECUTIVE SUMMARY Many of the fishers and fish farmers in developing countries are smallholders. The Food and Agriculture Organization (FAO) estimates that 55 million people were engaged in capture fisheries and aquaculture in 2010, while small-scale fisheries employ over 90 percent of the world’s capture fishers (FAO 2012). To these small-scale producers fish are both sources of household income and nutrients, and sustainable production and improved efficiency would contribute to improve their livelihoods and food security. Sustainably managing marine and coastal resources, including fish stock and habitat, would also help building and augmenting resilience of coastal communities in the face of climate change threats. One important feature of this food-producing sector is that fish is highly traded in international markets. According to the FAO (2012), 38 percent of fish produced in the world was exported in 2010. This implies that there are inherent imbalances in regional supply and regional demand for fish, and international trade—through price signals in markets—provides a mechanism to resolve such imbalances (Anderson 2003). Therefore, it is important to understand the global links of supply and demand of fish to discuss production and consumption of fish in a given country or a region, while understanding the drivers of fish supply and demand in major countries and regions is essential in making inferences about global trade outcomes. Developing countries are well integrated in the global seafood trade, and flow of seafood exports from developing countries to developed countries has been increasing. In value, 67 percent of fishery exports by developing countries are now directed to developed countries (FAO 2012). This report offers a global view of fish supply and demand. Based on trends in each country or group of countries for the production of capture fisheries and aquaculture and those for the consumption of fish, driven by income and population growth, IFPRI’s newly improved International Model for Policy Analysis of Agricultural Commodities and Trade (IMPACT model) simulates outcomes of interactions across countries and regions and makes projections of global fish supply and demand into 2030. Projections are generated under different as- sumptions about factors considered as drivers of the global fish markets. This report reflects a collaborative work between the International Food Policy Research Institute (IFPRI), the FAO, the University of Arkansas at Pine Bluff, and the World Bank. This work builds on the publica- tion Fish to 2020 by Delgado and others (2003). Throughout the report, the discussions are centered around three themes: (1) health of global capture fisheries, (2) the role of aquaculture in filling the global fish supply-demand gap and potentially reducing the pressure on capture fisheries, and (3) implications of changes in the global fish markets on fish consumption, especially in China and Sub-Saharan Africa. FINDINGS AND IMPLICATIONS This study employs IFPRI’s IMPACT model to generate projections of global fish supply and demand. IMPACT covers the world in 115 model regions for a range of agricultural commodities, to which fish and fish products are added for this study. As is the case with most global modeling work, an important value that IMPACT brings to this study is an internally consistent framework for analyzing and organizing the underlying data. However, there are known data and methodology issues that arise from choices made by the key researchers for the pur- poses of maintaining computational tractability, internal analytical consistency, and overall simplicity. These are summarized in section 2.6. BASELINE SCENARIO After demonstrating that the model successfully approximates the dynamics of the global fish supply and demand over the 2000–08 period, the outlook of the global fish markets into 2030 is projected under the scenario considered most plausible given currently observed trends (see table E.1 for key results). The model projects that the total fish supply will increase from 154 million tons in 2011 to 186 million tons in 2030. Aquaculture’s share in global supply will likely continue to expand to the point where capture fisheries and aquaculture will be contributing equal amounts by 2030. However, aquaculture is projected to supply over 60 percent of fish destined for direct human F I S H TO 2 030: P R O S P E C T S F O R F I S H E R I E S A N D A Q UA C U LT U R E EXECUTIVE SUMMARY xv TABLE E.1: Summary Results under Baseline Scenario (000 tons) TOTAL FISH SUPPLY FOOD FISH CONSUMPTION DATA PROJECTION DATA PROJECTION 2008 2030 2006 2030 Capture 89,443 93,229 64,533 58,159 Aquaculture 52,843 93,612 47,164 93,612 Global total 142,285 186,842 111,697 151,771 Total broken down by region as follows ECA 14,564 15,796 16,290 16,735 NAM 6,064 6,472 8,151 10,674 LAC 17,427 21,829 5,246 5,200 EAP 3,724 3,956 3,866 2,943 CHN 49,224 68,950 35,291 57,361 JAP 4,912 4,702 7,485 7,447 SEA 20,009 29,092 14,623 19,327 SAR 6,815 9,975 4,940 9,331 IND 7,589 12,731 5,887 10,054 MNA 3,518 4,680 3,604 4,730 AFR 5,654 5,936 5,947 7,759 ROW 2,786 2,724 367 208 Source: IMPACT model projections. Note: ECA = Europe and Central Asia; NAM = North America; LAC = Latin America and Caribbean; CHN = China; JAP = Japan; EAP = other East Asia and the Pacific; SEA = Southeast Asia; IND = India; SAR = other South Asia; MNA = Middle East and North Africa; AFR = Sub-Saharan Africa; ROW = rest of the world. consumption by 2030. It is projected that aquaculture will expand substantially, but its growth will continue to slow down from a peak of 11 percent per year during the 1980s. The global production from capture fisheries will likely be stable around 93 million tons during the 2010–30 period. Looking across regions, China will likely increasingly influence the global fish markets. According to the baseline model results, in 2030 China will account for 37 percent of total fish production (17 percent of capture production and 57 percent of aquaculture production), while accounting for 38 percent of global consumption of food fish.3 Given the continued growth in production projection, China is expected to remain a net exporter of food fish (net importer of fish if fishmeal is considered). Fast supply growth is also expected from aquaculture in South Asia (including India), Southeast Asia, and Latin America. Per capita fish consumption is projected to decline in Japan, Latin America, Europe, Central Asia, and Sub-Saharan Africa. In particular, per capita fish consumption in Sub-Saharan Africa is projected to decline at an annual rate of 1 percent to 5.6 kilograms during the 2010–30 period. However, due to rapid population growth, which is estimated at 2.3 percent annually during the 2010–30 period, total food fish consumption demand would grow substantially (by 30 percent between 2010 and 2030). On the other hand, projected production increase is only marginal. Capture production is projected to increase from an average of 5,422 thousand tons in 2007–09 to 5,472 thousand tons in 2030, while aquaculture is projected to increase from 231 thousand tons to 464 thousand tons during the same period. While the region has 3 Reduction into fishmeal and fish oil and inventory are the two other fish utilization categories considered in this study. A G R I C U LT U R E A N D E N V I R O N M E N TA L S E R V I C E S D E PA R T M E N T D I S C U S S I O N PA P E R x vi EXECUTIVE SUMMARY been a net importer of fish, under the baseline scenario, its fish imports in 2030 are projected to be 11 times higher than the level in 2000. As a result, the region’s dependency on fish imports is expected to rise from 14 percent in 2000 to 34 percent in 2030. Looking across species, the fastest supply growth is expected for tilapia, carp, and Pangasius/catfish. Global tilapia production is expected to almost double from 4.3 million tons to 7.3 million tons between 2010 and 2030. The demand for fishmeal and fish oil will likely become stronger, given the fast expansion of the global aquaculture and sluggishness of the global capture fisheries that supply their ingredients. During the 2010–30 period, prices in real terms are expected to rise by 90 percent for fishmeal and 70 percent for fish oil. Nonetheless, with significant improvements anticipated in the efficiency of feed and management practices, the projected expansion of aquaculture will be achieved with a mere 8 percent increase in the global fishmeal supply during the 2010–30 period. In the face of higher fishmeal and fish oil prices, species substitution in production is also expected, where production of fish species that require relatively less fish-based feed is preferred.4 SCENARIO ANALYSIS Besides the baseline (most plausible) scenario, six additional scenarios are implemented to investigate potential impacts of changes in the drivers of global fish markets (table E.2). TABLE E.2: Summary Results for Year 2030 under Baseline and Alternative Scenarios BASELINE SCENARIO 1 SCENARIO 2 SCENARIO 3 SCENARIO 4 SCENARIO 5 SCENARIO 6 FASTER GROWTH WASTE DISEASE CHINA CAPTURE GROWTH CCa CCb Total fish supply 186.8 194.4 188.6 186.6 209.4 196.3 184.9 185.0 (million tons) Capture supply 93.2 93.2 93.2 93.2 93.2 105.6 90.2 90.2 (million tons) Aquaculture supply 93.6 101.2 95.4 93.4 116.2 90.7 94.7 94.8 (million tons) Shrimp supply 11.5 12.3 11.5 11.2 17.6 11.6 11.5 11.4 (million tons) Salmon supply 5.0 5.4 5.1 5.0 6.1 5.0 4.8 4.8 (million tons) Tilapia supply 7.3 9.2 7.4 7.3 7.4 7.2 7.3 7.3 (million tons) Fishmeal price 1,488.0 13% –14% –1% 29% –7% 2% 2% ($/ton; % to baseline) Fish oil price 1,020.0 7% –8% –0% 18% –6% 3% 3% ($/ton; % to baseline) CHN per capita consumption 41.0 43.3 41.5 40.9 64.6 42.2 40.7 40.7 (kg/capita/year) AFR per capita consumption 5.6 5.9 5.8 5.6 5.4 6.4 5.5 5.5 (kg/capita/year) Source: IMPACT model projections. Note: CC-a = climate change with mitigation, CC-b = climate change without drastic mitigation, CHN = China, AFR = Sub-Saharan Africa. 4 This substitution pattern, however, cannot be confirmed in the model results at the aggregate level. This is likely due to the fact that fishmeal-intensive species, such as shrimp and salmon, tend to have higher income elasticities of demand than low-trophic species and effects of output demand growth likely overweigh the effects of higher input costs. F I S H TO 2 030: P R O S P E C T S F O R F I S H E R I E S A N D A Q UA C U LT U R E EXECUTIVE SUMMARY x v ii Scenario 1 addresses the case where all aquaculture is able to grow faster than under the baseline scenario by 50 percent between 2011 and 2030. In particular, the scenario assumes faster technological progress such that aquaculture would be able to supply a given amount at a lower cost (supply curves would shift outward), but it assumes the same feed requirements per unit weight of aquaculture production. Technical progress may include genetic improvement, innovations in distribution, improvements in disease and other management practices, control of biological process (life cycle) for additional species, and improvements in the condition of existing production sites and expansion of new production sites. While these technical changes are implicit in the baseline parameters, this scenario accelerates the changes by 50 percent. At the global level, the model predicts that aquaculture production in 2030 would expand from 93.2 million tons under the baseline case to 101.2 million tons under this scenario. The model predicts that the faster growth in all aquaculture would stress the fishmeal market and this effect would dictate which species and which regions would grow faster than the others. Under this scenario, tilapia production in 2030 would be 30 percent higher than in the baseline case; production of mollusks, salmon, and shrimp in 2030 would be higher by more than 10 percent. As a result, relative to the baseline scenario, all fish prices in 2030 in real terms would be lower by up to 2 percent, except for the price of the other pelagic category, which is used as an ingredient of fishmeal and fish oil. Fishmeal price in 2030 would be 13 percent higher than in the baseline case, while fish oil price would be higher by 7 percent. Scenario 2 investigates how expanded use of fish processing waste in fishmeal and fish oil production might affect the market of these fish-based products, where, in addition to the baseline countries, all countries that produce fishmeal or fish oil are now assumed to have the option to use waste in their production starting in 2011. Aquaculture expansion has relied in large part on improvements surrounding feed, including feed composition for nutrition and digestibility as well as cost effectiveness, genetics of fish, and feeding techniques and practices. While anticipated continuation of these improvements is already incorporated in the baseline parameters, this scenario addresses possible expansion of feed supply by utilizing more fish processing waste in the production of fishmeal and fish oil. The model indicates that fishmeal production in 2030 would increase by 12 percent and fishmeal price would be reduced by 14 percent relative to the 2030 results in the baseline case. This would boost the aquaculture production of freshwater and diadromous fish, salmon, and crustaceans. Although cost is involved in selection, collection, and reduction of fish waste, use of the additional feedstock represents a great opportunity to increase fishmeal and fish oil production, especially where organized fish processing is practiced. For example, 90 percent of the ingredients used in fishmeal produced in Japan come from fish waste (FAO data).5 Globally, about 25 percent of fishmeal is produced with fish processing waste as ingredient (Shepherd 2012). Increased use of fish waste would reduce the competition for small fish between fishmeal production (that is, indirect human consumption) and direct human consumption. Scenario 3 introduces a hypothetical major disease outbreak that would hit shrimp aquaculture in China and South and Southeast Asia and reduce their production by 35 percent in 2015. The model is used to simulate its impact on the global markets and on production in affected and unaffected countries between 2015 and 2030. Results suggest that countries unaffected by the disease would increase their shrimp production initially by 10 percent or more in response to the higher shrimp price caused by the decline in the world shrimp supply. However, since Asia accounts for 90 percent of global shrimp aquaculture, the unaffected regions would not entirely fill the supply gap. The global shrimp supply would contract by 15 percent in the year of the outbreak. However, with the simulated recovery, the projected impact of disease outbreak on the global aquaculture is negative but negligible by 2030. Scenario 4 is a case where consumers in China expand their demand for certain fish products more aggressively than in the baseline case. The scenario is specified such that Chinese per capita consumption of high-value shrimp, crustaceans, and salmon in 2030 would be three times higher than in the baseline results for 2030 and that of mollusks double the baseline value. These are higher-value commodities 5 Since domestic production is insufficient, Japan imports fishmeal, mainly from Peru. A G R I C U LT U R E A N D E N V I R O N M E N TA L S E R V I C E S D E PA R T M E N T D I S C U S S I O N PA P E R x viii EXECUTIVE SUMMARY relative to other fish species and, except for mollusks, they require fishmeal in their production. Under this scenario, global aquaculture pro- duction could increase to more than 115 million tons by 2030. This scenario would benefit the producers and exporters of these high-value products, such as Southeast Asia and Latin America. While overall fish consumption in China in 2030 would be 60 percent higher relative to the baseline case, all other regions would consume less fish by 2030. For Sub-Saharan Africa, per capita fish consumption in 2030 would be reduced by 5 percent under this scenario, to 5.4 kilograms per year. Fishmeal price in 2030 in real terms would increase by 29 percent and fish oil price by 18 percent relative to the baseline case. Over 300 thousand tons more of fishmeal would be produced, by reducing additional 1 million tons of fish otherwise destined for direct human consumption. Scenario 5 simulates the impacts of productivity increase of capture fisheries in the long run where fisheries around the globe let the fish stocks recover to the levels that permit the maximum sustainable yield (MSY). In The Sunken Billions (Arnason, Kelleher, and Willmann 2009), it is estimated that effectively managed global capture fisheries are assumed to sustain harvest at 10 percent above the current level. In this scenario, a gradual increase in the global harvest is assumed, achieving this augmented level in 2030. If this scenario were to be realized, the world would have 13 percent more wild-caught fish by 2030, relative to the baseline projection. In this scenario the resulting increase in the production of small pelagic and other fish for reduction into fishmeal and fish oil would reduce the pressure on the feed market, which results from the rapid expansion of aquaculture production that is expected to continue. Fishmeal price is expected to be lower by 7 percent than under the baseline case. Production in all regions would benefit from this scenario. In particular, Sub-Saharan Africa would achieve fish consumption in 2030 that is 13 percent higher than under the baseline scenario. This is because increased harvest is likely to be consumed within the region, rather than being exported. Distributional implications of the scenario would be even higher if stock recovery process is accompanied by efforts to substantially reduce inefficiency often prevalent in the harvest sector. Though confounded with losses due to lower-than-MSY yield, the cost of inefficient harvest sector is estimated to amount to $50 billion each year at the global level (Arnason, Kelleher, and Willmann 2009). On the other hand, relative abundance of fish would dampen fish prices so that aquaculture production in 2030 would be reduced by 3 million tons relative to the baseline case. Scenario 6 considers the impacts of global climate change on the productivity of marine capture fisheries. Changes in the global fish markets are simulated based on the maximum catch potentials (maximum sustainable yield, MSY) predicted by Cheung and others (2010) under two scenarios: one with mitigation measures in place so that no further climate change would occur beyond the year 2000 level and the other with continuing trend of rising ocean temperature and ocean acidification. Their mitigation scenario yields a 3 percent reduction in the global marine capture fisheries production in 2030 relative to the baseline scenario, while no-mitigation scenario would result in mar- ginal additional harm to the capture fisheries at the global level (reduction of harvest by 0.02 percent in 2030). While the aggregate impact is negligible, distribution of the expected changes in catches widely varies across regions. In principle, high-latitude regions are expected to gain while tropical regions lose capture harvest (Cheung and others 2010). The highest gains are expected in the Europe and Central Asia (ECA) region (7 percent) and the largest losses in the Southeast Asia (SEA; 4 percent) and East Asia and Pacific (EAP; 3 percent) regions. The model predicts that market interactions will attenuate the impact of the changes in the capture harvest and its distribution. Aquaculture will likely increase its production to offset the small loss in the capture harvest. Imports and exports will likely smooth the additional supply- demand gap caused by the changes in capture harvest, and fish consumption levels in 2030 are not expected to change in any region due to climate change. The simulated loss in global catches is relatively small in part because this study provides medium-term projections into 2030, whereas climate change is a long-term phenomenon. Given the structure of the IMPACT model, many small island states were grouped together in the “rest of the world” (ROW) model region. Therefore, this simulation exercise is unable to analyze the impact of climate changes in small island states, including Pacific island countries and territories. F I S H TO 2 030: P R O S P E C T S F O R F I S H E R I E S A N D A Q UA C U LT U R E EXECUTIVE SUMMARY x ix OVERALL LESSONS We have developed a rigorous analytical tool that is capable of making projections on the implications of the ongoing shifts on global fish production and reallocation of fish supply through international trade. The model, though with known limitations, is successfully calibrated and employed to evaluate different policies and alternative events and to illustrate likely evolution of the global seafood economy. From the modeling exercise and scenario analyses, it is clear that aquaculture will continue to fill the growing supply-demand gap in the face of rapidly expanding global fish demand and relatively stable capture fisheries. While total fish supply will likely be equally split between capture and aquaculture by 2030, the model predicts that 62 percent of food fish will be produced by aquaculture by 2030. Beyond 2030, aquaculture will likely dominate future global fish supply. Consequently, ensuring successful and sustainable development of global aqua- culture is an imperative agenda for the global economy. Investments in aquaculture must be thoughtfully undertaken with consideration of the entire value chain of the seafood industry. Policies should provide an enabling business environment that fosters efficiency and further technological innovations in aquaculture feeds, genetics and breeding, disease management, product processing, and marketing and distribution. The same is true for capture fisheries—developing enabling environment through governance reforms and other tools represents the first step toward recovery of overharvested fish stock and sustainability of global capture fisheries. A G R I C U LT U R E A N D E N V I R O N M E N TA L S E R V I C E S D E PA R T M E N T D I S C U S S I O N PA P E R CHAPTER 1 — INTRODUC TION 1 Chapter 1: INTRODUCTION 1.1. MOTIVATIONS AND OBJECTIVE harvested, and there is little prospect for significant increase in the Two notable manuscripts were published in 2003 on global fisheries, supply of these species from capture fisheries. aquaculture, and seafood trade. The Fish to 2020 study by Delgado In contrast, the rapid expansion of global aquaculture produc- and others (2003) provided a comprehensive global overview of tion has continued with no sign of peaking. During the past three the food fish supply and demand balance and trends observed decades, global aquaculture production expanded at an average during the 1970s through 1990s. These analyses formed the basis annual rate of more than 8 percent, from 5.2 million tons in 1981 of forward-looking projections to 2020 using IFPRI’s IMPACT model. to 62.7 million tons in 2011 (FishStat). Aquaculture’s contribution to The International Seafood Trade by Anderson (2003) drew insights total food fish supply grew from 9 percent in 1980 to 48 percent in into global seafood trade by discussing it in the context of broader 2011 (FAO 2013). The estimated number of fish farmers also grew food commodity trade. Focusing on key fish species and major play- from 3.9 million in 1990 to 16.6 million in 2010. The rapid and mas- ers in the markets, Anderson further analyzed the factors that drove sive growth of aquaculture production has contributed significantly global seafood trade and prices. to increased production of species whose supply would be other- wise constrained given the lack of growth in capture fisheries pro- Three key observations motivated these two publications: stagnant duction. As a result, the prices of these species (for example, salmon global capture fisheries, rapid expansion of aquaculture, and the and shrimp) declined, especially during the 1990s and in the early rise of China in the global seafood market. Separately and using 2000s (FAO 2012). different approaches and methods, both studies analyzed these trends and their implications for the global seafood economy. Ten Seafood demand from China, the single largest market for seafood, years later, many of these trends and the associated concerns and has grown substantially, and its influence on the global fish markets challenges have continued, as documented in the FAO’s State of and trade has intensified. China’s per capita fish consumption grew World Fisheries and Aquaculture 2012. According to the FAO report, to 33.1 kilograms per year in 2010, at an annual rate of 6 percent since the mid-1990s, production from global capture fisheries has between 1990 and 2010. So far, due particularly to growth in aqua- stabilized around 90 million tons, with marine fisheries contribut- culture production, fish production in China has kept pace with ing around 80 million tons. This represents a substantial increase the growth in consumption demand from population and income from 18.7 million tons in 1950, of which 16.8 million tons were from growth. While Asia accounted for 88 percent of world aquaculture marine waters, but the expansion of marine capture fisheries was production by volume in 2011, China alone accounted for 62 per- achieved in part at the cost of deteriorating regional fish stocks. cent. Aquaculture now represents more than 70 percent of the Since the beginning of FAO stock assessments, the proportion of total fish produced in China. With the rapid growth in production, overexploited stocks has steadily increased from 10 percent in 1974 China’s share in the global fish production grew from 7 percent in to 26 percent in 1989 and, with a slowing trend, to 30 percent in 1961 to 35 percent in 2011. Notwithstanding that China consumes 2009. Furthermore, most of the stocks of the 10 key fish species 34 percent of global food fish supply, it is still a net exporter of food (which represent 30 percent of marine capture production) are fully fish. Nevertheless, China is both an importer and exporter of fish. A G R I C U LT U R E A N D R U R A L D E V E LO P M E N T D I S C U S S I O N PA P E R 2 CHAPTER 1 — INTRODUC TION China became the third-largest fish-importing country by value in and Brazil), and more recently infectious salmon anemia (ISA) virus 2011 after Japan and the United States. Part of the fish imports is raw in Chile (OECD 2010). Early mortality syndrome, a disease recently material to be reexported after processing. China now represents found in farmed shrimp, has caused substantial losses in China, 13 percent of world fish export in value, amounting to $17.1 billion Vietnam, Malaysia, and Thailand (Leaño and Mohan 2012). These in 2011 and $18.2 billion in 2012 (FAO 2013). outbreaks provide a warning to other rapidly expanding aquacul- ture sectors of the importance of disease management and adop- Thus, the factors that that motivated Delgado and others (2003) and tion of best practices (Bondad-Reantaso and others 2005). Global Anderson (2003) are still very relevant after a decade and will likely climate change will likely exacerbate the susceptibility of aquacul- continue to shape the evolution of the global seafood economy ture to disease (see, for example, Leung and Bates 2013). In addi- through international trade. According to the FAO (2012), seafood is tion, climate change will cause further changes in the oceans and among the most heavily traded food commodities, with 38 percent aquatic ecosystems and therefore pose threats to fish populations of all fish produced being exported in 2010. Fish and fish prod- and the economies that depend on them (World Bank 2013b). ucts account for 10 percent of agricultural exports in value terms. Nominally, world trade of fish and fish products increased from $8 Despite the overall growth of fish consumption and trade that has billion in 1976 to $128 billion in 2012, which in real terms translates occurred in much of the world, a decline in per capita fish consump- into an average annual growth rate of 4.0 percent. Developing tion has been observed in some Sub-Saharan African countries, such countries are well integrated in the global seafood trade, with more as Gabon, Malawi, South Africa, and Liberia (FAO 2012), as well as than 54 percent of all fishery exports by value and more than 60 per- in some developed countries, such as Japan and the United States. cent by quantity (in live weight equivalent) coming from develop- Per capita fish consumption in the Africa region is roughly half of ing countries. The FAO (2012) attributes the growing participation of the global average (FAO 2012). The decline in per capita fish con- developing countries in the global fish trade to, at least in part, the sumption has far-reaching consequences for the intake of protein generally low import tariffs on fish and fish products imposed by and micronutrients important for human growth and development developed countries, which are dependent on fish imports (and do- (Oken and others 2008, USDA 2012). mestic aquaculture) due to sluggish capture fisheries. Furthermore, the FAO (2012) argues, while growing demand, trade liberalization Opportunity policies, globalization, and technological innovations have led to an In the face of these concerns and challenges, however, there has increase in global fish trade, improvements in processing, packing, been an important and promising shift in the approach to global and marketing and distribution have altered the way fish products fisheries and aquaculture challenges. The shift is driven in part by are prepared and marketed. the growing body of knowledge on the key drivers of change within the global fish markets and the understanding of management Given these observations, this report addresses the following suite and governance of capture fisheries and aquaculture. For example, of questions: how will global seafood trade evolve further in the the Sunken Billions study by the World Bank and the FAO (Arnason, next 15–20 years? In particular, to what extent and at what rate will Kelleher, and Willmann 2009) endeavored to quantify, in a clear and aquaculture be able to continue to expand as fish demand growth convincing way, the extent of economic losses due to the poor keeps rising in China and elsewhere? How will increased fish supply management of the global marine fisheries; the estimated losses from aquaculture be distributed across regions? amount to $50 billion each year. A recent article by Costello and Along with these continuing trends, some new and renewed others (2012) illustrated how sustainable levels of stocks in marine concerns have arisen. With the rapid growth of aquaculture, there fisheries can be regained if appropriate management changes are have been major disease outbreaks within the aquaculture sec- undertaken, especially in those fisheries that are less studied and tor in various countries, including white-spot syndrome virus in have not been as closely monitored or assessed for their stock levels. shrimp (global), infectious myonecrotic virus in shrimp (Indonesia Similarly, Gutiérrez, Hilborn, and Defeo (2011) demonstrated that F I S H TO 2 030: P R O S P E C T S F O R F I S H E R I E S A N D A Q UA C U LT U R E CHAPTER 1 — INTRODUC TION 3 community-based comanagement of aquatic resources is the only improve the health of the oceans and the performance and sustain- realistic solution for the majority of the world’s fisheries when strong ability of global fisheries. For example, the Rio+20: UN Conference community leadership is present and is combined with effective on Sustainable Development in June 2012 affirmed “the necessity resource management tools such as quotas and marine protected to promote, enhance and support more sustainable agriculture, areas. including . . . fisheries and aquaculture” and stressed “the crucial role of healthy marine ecosystems, sustainable fisheries and sustainable The Fish to 2020 study raised concerns regarding environmental im- aquaculture for food security and nutrition and in providing for the pacts of aquaculture expansion, including massive changes in land livelihoods of millions of people.” use, pollution of neighboring waters with effluent, and the spread of disease among fish farms. While these concerns remain today, there has been considerable discussion surrounding sustainable In Fish to 2030 aquaculture (Ward and Phillips 2008, Brummett 2013). Sustainability In this new and dynamic context, the Fish to 2030 study extends has now been recognized as the principal goal of aquaculture gov- the projections of the global supply, demand, and trade of fish and ernance by many governments (FAO 2012). Furthermore, the Fish fish products to 2030, incorporating lessons learned in Fish to 2020 to 2020 study projected that the demand for fishmeal and fish oil and new developments in the global fish markets since it was pub- would continue to increase with aquaculture expansion and that lished. Linking the expertise of the IMPACT modeling team at IFPRI, this could have serious implications for the viability and health the sectoral knowledge of the fisheries group at the World Bank, of the capture fisheries of those species used in feed production. and comprehensive data provided by the Fisheries and Aquaculture While it could lead to an overexploitative outcome as predicted by Department of the FAO, the fish component of IMPACT is sub- Delgado and others (2003), increasing demand and the associated stantially improved over the model version used in Fish to 2020. In rise in price for those fish species could also offer a remarkable op- particular, the fish species and global regions are much more disag- portunity for fisheries to implement appropriate management and gregated in the new version so that more targeted analyses are pos- utilize the resources profitably and sustainably. Moreover, driven sible. Chapter 2 presents the details of how the model is modified by the high cost of protein, especially fishmeal, aquaculture has and how more recent data and other information are incorporated achieved substantial innovation in feeds and efficiency improve- into the analysis. ment in feeding practices in recent years (Rana, Siriwardena, and Hasan 2009). In trying to reduce dependence on fishmeal, research Using the model, this study generates a series of projections of fish institutions and the aquaculture feed industry have conducted nu- supply, demand, and trade into 2030. The first set of results is based merous studies, which have led to an “impressive reduction in the on the “baseline scenario” that reflects the currently observed trends average inclusion of fishmeal in compound feeds for major groups of supply and demand, and thus it is deemed the “most plausible” of farmed species” (FAO 2012). As a result, according to FAO data, a case. These are presented in chapter 3. 62 percent increase in global aquaculture production was achieved Once a plausible baseline level of fish production, consumption, when the global supply of fishmeal declined by 12 percent during and trade is established out to 2030, the model is used to examine the 2000–08 period. The use of fish processing waste in fishmeal the implications of several alternative scenarios that are designed to production has also increased (Chim and Pickering 2012, FAO 2012, illustrate how shocks and changes to the production or consump- Shepherd 2012). tion side of the world fish economy trigger market responses. One Building on knowledge and lessons learned from a variety of ex- scenario introduces impacts of climate change on the health and periences, there is considerable momentum toward combined productivity of marine ecosystems, in particular through change in efforts of public and private stakeholders, donors and recipients, ocean temperatures and levels of acidity in ocean waters (Brown development organizations such as the World Bank and the FAO, and others 2010, Cheung and others 2010). The results for six alter- civil society organizations, and advanced research institutions to native scenarios are presented in chapter 4. A G R I C U LT U R E A N D R U R A L D E V E LO P M E N T D I S C U S S I O N PA P E R 4 CHAPTER 1 — INTRODUC TION 1.2. LESSONS FROM FISH TO 2020 FIGURE 1.1: Comparison of Fish to 2020 Projections and FAO Data for Global Food Fish Supply In this section, we revisit the Fish to 2020 study (Delgado and others FAO total food fish supply Fish to 2020 total fish supply 2003) and review the IMPACT model’s predictive power by compar- 120 ing their projections to the actual evolution of global markets since it 115 was published. Further, we summarize their key findings and discuss how the model can be improved to better address the same issues. 110 The discussions here will lead to the two subsequent sections: the strategy to improve the IMPACT fish component (section 1.3) and 105 the policy research questions addressed in this study (section 1.4). Million tons 100 Global Food Fish Production 95 The Fish to 2020 study used the IMPACT model to generate projec- tions of global food fish production, consumption, and trade for 90 the period 1997 to 2020. The study addressed only fish destined for human consumption (food fish), and thus not all species pro- 85 duced by capture fisheries or aquaculture were included. Among 80 the excluded species were those “reduced” into fishmeal and fish 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 oil and not used for human food. The model projected that the Sources: Delgado and others 2003 and FAO Fisheries Information and Statistics Branch (FIPS) food balance sheets (FBS) data. global food fish supply would grow from 93.2 million tons in 1997 Note: As is detailed in section 2.3, data on global fish consumption were available through 2007 at the time of model preparation. to 130.1 million tons by 2020. Of those, capture fisheries was pro- jected to grow from 64.5 million tons in 1997 to 76.5 million tons in 2020, to represent 59 percent of the total projected supply in 2020. but in reality aquaculture grew at the rate of 7.1 percent per year on Aquaculture production, on the other hand, was projected to grow average. Given that the model incorporated the best available infor- from 28.6 million tons in 1997 up to 53.6 million tons by 2020, rep- mation, the divergence between the model projections and actual resenting the remaining 41 percent of the projected supply in 2020. data implies that aquaculture expanded much more rapidly than According to the projections, by 2020 developing countries would the expectations of experts. A better representation of aquaculture be responsible for 79 percent of world food fish production, while expansion trends is warranted in the current modeling exercise in 77 percent of global fish consumption would occur in developing Fish to 2030. countries. The next two figures portray the context in which Fish to 2020 study A decade later, one can assess the quality of the Fish to 2020 model was prepared. Figure 1.2 illustrates the evolution of the world food projections by contrasting them with actual data. In figure 1.1, a fish production since 1980s. The rise of aquaculture in the past reasonably close congruence is seen between the Fish to 2020 pro- three decades is clear in the figure. Aquaculture accounted for only jections and actual evolution of the global food fish supply for the 12 percent of world food fish production in 1984, while it accounted first decade of the projection period (that is, 1997–2007). However, for 46 percent in 2009. In fact, aquaculture is one of the most rap- the results suggest an overprediction of capture production and idly growing food sectors globally (FAO 2012), which is in a sharp an underprediction of aquaculture production (not shown in the contrast with stagnant capture fisheries production. The stagnation figure). Over the 1997–2007 period, global capture fisheries were of capture fisheries is also seen in figure 1.3, where even a nega- projected to grow at 0.8 percent annually, while they actually grew tive growth (decline) is recorded for 2000–09. The figure also illus- at the average annual rate of 0.5 percent. The projection of aqua- trates how rapidly aquaculture grew during the 1980s and 1990s. culture growth rate in the Fish to 2020 study was 3.4 percent a year, After growing at an average annual rate of more than 10 percent, F I S H TO 2 030: P R O S P E C T S F O R F I S H E R I E S A N D A Q UA C U LT U R E CHAPTER 1 — INTRODUC TION 5 FIGURE 1.2: Evolution of World Food Fish Production, urbanization were considered to drive the demand for fish as well 1984–2009 as for livestock products in developing countries. By contrast, fish Farmed Wild Total consumption in developed countries was projected to increase by 140 about only 4 percent, from 28.1 million tons in 1997 to 29.2 million 120 tons in 2020. 100 Million tons 80 The Fish to 2020 study concluded that the projected fish consump- 60 tion trajectories to 2020 could not be met by capture fisheries alone 40 and would only be feasible if aquaculture continued to grow ag- 20 gressively. Rapid expansion of aquaculture was also discussed in 0 the context of its implications for dietary diversification and food 84 87 90 93 96 99 02 05 08 19 19 19 19 19 19 20 20 20 Source: FishStat. security among the poor in developing countries. It was expected that aquaculture could augment fish supply and reduce prices, as FIGURE 1.3: Average Annual Growth Rates of Capture and observed for low-value freshwater fish in Asia, and could benefit Aquaculture Production, 1960–2009 poor households in food insecure parts of the world. Capture Aquaculture 12 The report projected stagnant fish consumption in Sub-Saharan 10 Africa and, as discussed earlier, the FAO (2012) reports declining per capita fish consumption in some Sub-Saharan African nations dur- 8 ing the 2000s. The reported food fish consumption in Africa in 2009 6 was 9.1 kilograms per capita per year (FAO 2012). Whether this will Percent grow in the future with affordable, aquaculture-sourced fish supply 4 is an important policy research question we will keep coming back 2 to throughout this study. 0 1960–69 1970–79 1980–89 1990–99 2000–09 −2 Global Food Fish Trade and Prices Source: FishStat. One of the policy research questions addressed in the Fish to 2020 study was whether the growth patterns for fish demand would growth in world aquaculture production has slowed. Nevertheless, continue along the same trends that have been observed in de- aquaculture continued to grow at more than 6 percent annually veloped and developing regions, and how price and trade patterns during the 2000s. Adequately capturing this rapid growth trajectory would develop over time to determine fish distribution across these of aquaculture constitutes one of the major objectives in the effort regions. Given the relatively stable capture fisheries production for to improve the modeling framework in Fish to 2030. the past decades, the only sector with the ability to grow seemed to be the aquaculture sector, even though the capacity for growth Global Food Fish Consumption might not uniformly exist in all regions. If aquaculture were to grow The Fish to 2020 study projected that developing countries would to meet the growing consumption demand, what would be the consume a much greater share of the world’s fish in the future implications for trade in food fish and the prices of those products? and that trade in fish commodities would also increase. The report The Fish to 2020 study projected, except under their “faster aquacul- projected that fish consumption in developing countries would ture expansion” scenario, that fish prices would rise more dynami- increase by 57 percent, from 62.7 million tons in 1997 to 98.6 million cally than prices for any other food product. This implied that there tons in 2020. Rapid population growth, increasing affluence, and would be regional imbalances in fish supply and demand and that A G R I C U LT U R E A N D R U R A L D E V E LO P M E N T D I S C U S S I O N PA P E R 6 CHAPTER 1 — INTRODUC TION international trade would respond by reallocating supplies from FIGURE 1.4: Trends of Real Prices of Selected Seafood more productive, surplus regions to those regions that tend to fall Wild cod, fillet, frozen Farmed salmon, 2–3 lb fillet, fresh Farmed catfish, fillet, frozen in deficit of food fish. In other words, the model projected that ris- 8.00 ing fish prices would be a catalyst to stimulate further international 7.00 trade of fish to correct for regional imbalances. $/pound in constant 2000 US$ 6.00 The IMPACT model price projections, however, need to be interpreted 5.00 with caution. As figure 22 in the State of the World Fisheries and 4.00 Aquaculture (FAO 2012) indicates, at the aggregate level, average 3.00 fish prices declined in real terms during the 1990s and, even with an 2.00 increase during the 2000s, fish prices in 2010 were still lower than 1.00 the 1990 levels. In general, falling prices were observed for those 0.00 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 species that achieved rapid expansion of aquaculture production Source: USDA, Urner-Barry Publications, U.S. Department of Commerce (USDC)/ (FAO 2012), which in turn was driven by technological advances in National Marine Fisheries Service (NMFS). Reproduced from Anderson 2012. the animal genetics and in the production and utilization of feeds (Brummett 2003). Those species with falling prices include shrimp, incorporating such dynamisms of global aquaculture, this poses a salmon, and some fish species farmed in freshwater. challenge for the model improvement effort in this study as well. The difficulties the IMPACT model faced in addressing species- specific trends in production and prices originated in part from their Aquaculture-Capture Interactions species aggregation strategy. In the Fish to 2020 study, the species The Fish to 2020 study addressed the sustainability of marine cap- were aggregated into four commodities: low-value food fish, high- ture fisheries in the face of rapid expansion of global aquaculture value finfish, crustaceans, and mollusks. However, more importantly, and associated strong growth in fishmeal demand. An “ecological there seem to be structural limitations in the extent to which the collapse” scenario simulated the market-level impacts of a gradual IMPACT model could represent the dynamics of world fish prices. but catastrophic collapse of the marine fisheries (a decline of food The IMPACT model was originally developed for commodities fish capture fisheries at the annual rate of 1 percent was applied). whose international markets are relatively more established and As expected, the results were striking in terms of price increases for mature (for example, grains and meat), while aquaculture is a rela- fishmeal and fish oil as well as for fish commodities that use these tively new and far more dynamic industry, with new technological products in production. advances being made continuously for existing and new spe- The study also emphasized the role of technology, especially tech- cies as well as for processing, packaging, and distribution. Some nology to increase the efficiency with which fishmeal and fish oil are aquaculture species, such as salmon, have already gone through a converted into farmed fish. A scenario where such feed conversion series of substantial technological changes and their markets have efficiency would improve twice as fast relative to the baseline sce- fairly matured. During the process, the production cost has been nario resulted in reduced prices of fishmeal and fish oil, but practi- substantially reduced—and world salmon price has dropped con- cally no change was projected for the levels of fish prices or fish siderably (figure 1.4). For those relatively new aquaculture species, supply. The latter results seem counterintuitive and warrant further such as tilapia and Pangasius, technological advances have only investigation. begun and similar downward trends of real prices are expected in the near future. Aquaculture species that are not yet commercially The Fish to 2020 study has identified production and use of fishmeal farmed may become commercially viable in the future, and they will and fish oil as one of the key interactions between aquaculture and likely follow similar paths of market maturity as other species (Asche capture fisheries and between fisheries (capture or aquaculture) 2011; Brummett 2003, 2007). If the model indeed had difficulties in and the natural environment. However, the study only addressed F I S H TO 2 030: P R O S P E C T S F O R F I S H E R I E S A N D A Q UA C U LT U R E CHAPTER 1 — INTRODUC TION 7 FIGURE 1.5: Global Fishmeal Use fish species enables better representation of the dynamics of high- Other Aquaculture Poultry Swine value markets, such as shrimp and salmon, and to separate them 2% 4% 2% from other important and fast-growing aquaculture species, such 100% 10% as tilapia and Pangasius. The production side of the newly modified 90% 80% IMPACT model contains 16 fish species groups, which is aggregated 48% 70% into nine commodities for the modeling of consumption and trade. 60% 50% 73% This disaggregation, based mainly on feeding requirements of fish 50% species, is also much less arbitrary than the previous classification, 40% as the same species may be of higher or lower value depending 30% 50% 5% on value-added processes (for example, tuna fillets in an expensive 20% 36% 10% 20% restaurant versus canned tuna) or depending on countries and 0% regions. For example, relatively high-value export species in Sub- 1960 1980 2010 Saharan Africa, such as Nile perch, were categorized as low value. Source: Shepherd 2012. Note also that the new version of the model includes production of all types of fish, including those destined for fishmeal and fish oil food fish, and fish species that are used for production of fishmeal production as well as for direct human consumption and other uses. and fish oil were left out. Thus the scope and the depth of the This makes possible the modeling of explicit links between fishmeal analyses on the aquaculture-capture interactions through feed fish and fish oil production and their use in aquaculture. were limited. Furthermore, as illustrated in figure 1.5, the utiliza- tion of global fishmeal has evolved over the past half century. The importance of aquaculture as user of fishmeal has grown substan- Realistic Treatment of Capture Production Growth tially as a result of the industry’s rapid growth since the 1980s. From As seen in the previous section, the Fish to 2020 study tended to proj- this observation, it is likely that fishmeal-intensive segments of the ect overly optimistic growth of capture fisheries and underestimate aquaculture industry (for example, salmon and shrimp) increasingly the growth of aquaculture in relation to the actual data between affect the dynamics of fishmeal and fish oil markets. In this context, 1997 and 2007. We suspect that the rising fish prices in the model, modeling of demand for fishmeal and for fish oil as a function of combined with capture supply that was specified overly sensitive aquaculture production seems essential. to fish prices, drove the results. Consequently, the projected capture supply increased more than the actual data indicated, crowding out the growth of aquaculture in the model. 1.3. STRATEGY OF IMPROVING MODELING FRAMEWORK IN FISH TO 2030 In response to these shortcomings, this study treats the growth From the review of Delgado and others (2003), we have identified of capture fisheries as entirely exogenous—that is, no supply re- several directions in which the IMPACT model can be improved for sponse to price changes is modeled for capture fisheries. In terms use in the Fish to 2030 analysis. In this section we summarize the of modeling price responses of supply, we maintain a solid focus strategy in adding a fisheries component to the existing IMPACT on aquaculture. The rationale behind this decision is that, given model. Detailed descriptions of the model and data are provided relatively stable capture fisheries in the last decades and the fact in chapter 2. that dynamic biological processes determine the amount of fish stock available for harvest, modeling of price-responsive capture Expansion of Fish Product Category supply in a static sense seems unrealistic. The open-access nature of The four broad classifications of fish used in the Fish to 2020 study many capture fisheries also further complicates the representation (low-value food fish, high-value finfish, crustaceans, and mollusks) of fish supply behavior (Arnason, Kelleher, and Willmann 2009). Thus, are expanded in this study. A more disaggregated representation of rather than allowing capture supply to respond freely to increasing A G R I C U LT U R E A N D R U R A L D E V E LO P M E N T D I S C U S S I O N PA P E R 8 CHAPTER 1 — INTRODUC TION or decreasing fish prices, in this study we exogenously specify the spent identifying the causes of data problems and devising ways to behavior of capture fisheries based on the observed trends and ac- reconcile those problems. The exercise is detailed in section 2.4 and cording to alternative scenarios. However, results on the final distri- calibration results are presented in section 2.5 of chapter 2. bution of capture fisheries production will depend on relative prices and demand in each country. 1.4. POLICY RESEARCH QUESTIONS Incorporating Fishmeal and Fish Oil Markets Using the modified IMPACT model, this study takes on the chal- In the new version of IMPACT, production and utilization of fishmeal lenge of inferring the short- to medium-run picture of the global and fish oil are modeled explicitly. The utilization of lower-value, fish markets. The discussion begins with the baseline scenario, smaller pelagic and other species for “reduction” and the resulting which reflects the trends that are currently observed and is deemed production of fishmeal and fish oil are now endogenized in a much most plausible given the current knowledge. Subsequently, the fol- more complete way. That is, the supply of fishmeal and fish oil and lowing six illustrative scenarios are introduced: the demand for the ingredient fish species are now determined § Scenario 1: Faster aquaculture growth in the model as a result of responses to the price of those com- § Scenario 2: Expanded use of fish processing waste in fish- modities. The demand for the fish-based feed is also modeled as meal and fish oil production price responsive. Further, feed demand is now calculated for each § Scenario 3: A major disease outbreak in shrimp aquaculture species based on the biological requirements and their trends. The in Asia new treatment of fishmeal and fish oil production and utilization is § Scenario 4: Accelerated shift of consumer preferences in China entirely parallel to the way that plant-based oil and meal produc- § Scenario 5: Improvement of capture fisheries productivity tion and utilization have been treated in the IMPACT model, where § Scenario 6: Impacts of climate change on the productivity of capture fisheries oil-bearing crops (soybean and oilseeds) are used as input into the production of vegetable oil and the meal produced as by-product is The results from the baseline and the six scenarios are used to un- used as input for livestock production. derstand how the trends in the factors considered as key drivers of change actually drive the model output. For example, we posit that In addition to small fish from capture fisheries, the newly modified the growth of demand for fish products is based on trends in regional IMPACT model also accounts for the use of fish processing waste income and population growth. However, even without the model, in production of fishmeal and fish oil. While this is an area that is we predict that the growth in regional demand for fish would not not well documented, an increase in the importance of waste use be proportional to the population growth or income growth. There is suggested (Chim and Pickering 2012, FAO 2012, Shepherd 2012). will be limitations in the extent that the global fish supply can grow Incorporation of fish processing waste enables a comprehensive and fish prices will adjust to the extent that the demand grows faster analysis of the links of the global fish markets through fishmeal and than supply. Production in some countries and regions will grow fish oil and implicitly through fish processing. faster than in others, and accordingly there will be regional gaps in fish supply and demand and the global fish trade market will balance Model Calibration those regional gaps. The elasticities of demand incorporated in the In this study, considerable effort is taken to calibrate the model pro- IMPACT model translate the strength of income growth and price jections to observed data. Such an exercise was not conducted in changes into consumption growth for each country. the Fish to 2020 study. By ensuring that the near-term projections closely align with the most recent data, further confidence about In the scenario analyses, special attention is placed on the cases of the future projections is gained. The calibration exercise also has China and Sub-Saharan Africa. China is one massive market that can enabled us to appreciate the issues and problems regarding the influence the dynamics of the global fish supply and demand. China available fisheries statistics, and, as a result, much time has been currently accounts for 35 percent of global fish production and 30 F I S H TO 2 030: P R O S P E C T S F O R F I S H E R I E S A N D A Q UA C U LT U R E CHAPTER 1 — INTRODUC TION 9 percent of global fish consumption and is a net exporter of fish, the supply of fish products with their demands in all regions of the although different commodities are imported or exported. While world. Thus, none of the effects of these localized supply shocks can one scenario specifically addresses China’s consumption trend be completely isolated. The expected differences in the way differ- (scenario 4), all other scenarios affect China’s fish supply-demand ent regions will be able to cope with such shocks are of interest to balance in important ways, which in turn influences the rest of the this study. world through the global fish markets. How such repercussions in the global markets affect fish supply balances in Sub-Saharan Africa Another channel of global links in the fish markets is through fish- is one of the key research questions of this study. As seen earlier, per meal and fish oil. The rapid growth of aquaculture across various capita fish consumption in this region is on a declining trend. The fish species leads to considerable pressure on supplies of fish-based region is a net importer of fish in volume and, with the projected feed. Scenario 1 directly intensifies such pressure while scenario 2 population growth at an annual rate of 2.3 percent between 2010 reduces it. Other scenarios also indirectly affect the supply or de- and 2030 (UN 2011), the region’s dependence on imports for fish mand of fish-based feed. Given the fish species disaggregation in consumption is expected to rise. the new version of the IMPACT model, the effects of changes in fishmeal and fish oil supply on aquaculture can now be examined Scenarios 3 and 6 introduce supply shocks to aquaculture through at the species level. disease outbreak and capture fisheries through climate change, respectively. Although the direct impacts of these shocks may be Scenario 5 offers a picture of the potential outcome of global ef- local, their impacts extend globally. Links exist at all levels of the fish forts to restore capture fisheries around the world. Growing global sector. Though not explicitly incorporated, crucial biophysical links interests in oceans agenda are expected to accelerate and scale up include the spread of fish diseases through waterways, the migra- such global efforts and bring the state of the capture fisheries closer tion pathways of fish that determine spatial stock distributions, and to their potentials as described in The Sunken Billions (Arnason, any other connectedness brought about by contiguity of ocean Kelleher, and Willmann 2009). This study offers illustrations of how waters. Further interconnectedness occurs through value chains of each region may benefit from improved capture fisheries in terms fish products and, in particular, international trade, which connects of gains in fish production and consumption. A G R I C U LT U R E A N D R U R A L D E V E LO P M E N T D I S C U S S I O N PA P E R C H A P T E R 2 — P R E PA R I N G I M PA C T M O D E L F O R F I S H T O 2 0 3 0 11 Chapter 2: PREPARING IMPACT MODEL FOR FISH TO 2030 2.1. BASICS OF IMPACT MODEL TABLE 2.1: Non-Fish Commodities Included in the IMPACT Model IFPRI’s International Model for Policy Analysis of Agricultural CATEGORIES DESCRIPTION Commodities and Trade, or IMPACT (Rosegrant and others 2001) Livestock products Beef and buffalo meat continues to serve as the “workhorse” of the analyses in this book. All poultry meat (chicken and ducks, primarily) Sheep and goat meat (small ruminants) IMPACT is a global, multimarket, partial equilibrium economic mod- Pig meat el that covers a wide range of agricultural products, such as cereals, Eggs All liquid and solid milk products from large and small ruminants oilseeds, roots and tubers, pulses, livestock products, and now the Cereals Aggregate of all grains (rice, wheat, maize, and other coarse grains) new addition of fish products. While the model has undergone a Roots and tubers Aggregate of all roots and tuber crops (Irish and sweet potatoes, number of extensions since when the Fish to 2020 study was con- yams, cassava, and other roots/tubers) ducted, the basic architecture of the model has remained true to its Pulses Principally chickpea and pigeon pea origins. The main objective of IMPACT is to provide forward-looking Sugar crops Aggregate of sugar cane and sugar beet projections of supply, demand, and trade for various agricultural Soybean Soybean, with soybean oil and meal as by-products Temperate oilseeds Rapeseed (canola), sunflower and safflower seeds, with their oil and products. Projections are typically generated under baseline specifi- meal by-products cations and under alternative scenarios. Tropical oilseeds Groundnut, coconut, palm, and other tropical oil-bearing crops, with their oil and meal by-products Fruits and vegetables Aggregate of fruit and vegetable categories Commodities Cotton Lint cotton For the purpose of this study, as discussed in chapter 1 (section 1.3), Other Aggregate of other miscellaneous agricultural crops the fish category is expanded relative to the Fish to 2020 specifica- tion. Accordingly, a total of 17 fish products are included in the finfish, crustaceans, and mollusks). In section 2.3, we will further de- newer version of IMPACT. Some of the existing non-fish commodities scribe the logic and data sources that underlie the choice of these are aggregated in order to reduce the overall “size” of the model and fish product classifications. In brief, the consumption category is the number of variables; however, this aggregation of non-fish com- more aggregated than the production category due to lack of dis- modities does not change the model results in any way. This helps to aggregated consumption data. On the production side, available make the model run faster and allows a focus on the commodities of highly disaggregated data by species are aggregated to ensure particular interest, namely fish and fish-based products (fishmeal and model tractability while maintaining a sufficient level of disaggrega- fish oil). Table 2.1 summarizes the aggregated non-fish commodities tion to allow flexibility in analysis. that are incorporated in the current IMPACT model, while table 2.2 lists the added fish products. These cover the range of commodities Regions that are important for global food consumption and nutrition. The earlier version of the IMPACT model used in the Fish to 2020 The addition represents a significant expansion of the fish category study divided the world into 36 regions. In contrast, the latest version from the Fish to 2020 classification (low-value food fish, high-value of the IMPACT model now contains 115 regions. This is the degree of A G R I C U LT U R E A N D R U R A L D E V E LO P M E N T D I S C U S S I O N PA P E R 12 C H A P T E R 2 — P R E PA R I N G I M PA C T M O D E L F O R F I S H T O 2030 TABLE 2.2: Fish Products Included in the IMPACT Model CONSUMPTION CATEGORY PRODUCTION CATEGORY DESCRIPTION SPECIES GROUP SPECIES GROUP ABBREVIATION Shrimp Shrimp Shrimp Shrimp and prawns Crustaceans Crustaceans Crustaceans Aggregate of all other crustaceans Mollusks Mollusks Mollusks Aggregate of mollusks and other invertebrates Salmon Salmon Salmon Salmon, trout, and other salmonids Tuna Tuna Tuna Tuna Freshwater and diadromous Tilapia Tilapia Tilapia and other cichlids Pangasius and other catfish Pangasius/catfish Pangasius and other catfish Carp Carp Major carp and milkfish species Other carp OCarp Silver, bighead, and grass carp Eel and sturgeon EelStg Aggregate of eels and sturgeon Other freshwater and diadromous OFresh Aggregate of other freshwater and diadromous species Demersals Major demersals MDemersal Major demersal fish Mullet Mullet Mullet Pelagics Cobia and swordfish CobSwf Aggregate of cobia and swordfish Other pelagics OPelagic Other pelagic species Other marine Other marine OMarine Other marine fish Fishmeal Fishmeal Fishmeal Fishmeal from all species Fish oil Fish oil Fish oil Fish oil from all species spatial disaggregation used in this study. In fact, these 115 regions include productivity and efficiency gains in agricultural production. are mainly countries, with some smaller nations grouped together These drivers essentially shift the intercepts of the supply curves to form regions. For the definition of the IMPACT 115 regions, see over time. On the other hand, changes in supply in response to price the model description by Rosegrant and the IMPACT Development changes are treated endogenously in the model using supply func- Team (2012). While the IMPACT model generates results for each of tions, which embed price elasticities. In this study, exogenous trends the 115 regions, for the purpose of this study, results are presented are the only determinants of supply growth in capture fisheries pro- for 12 aggregate regions. These 12 aggregate regions are defined duction. In contrast, the growth of aquaculture supply in the model in figure 2.1. Table 2.3 contains the abbreviation code for each ag- is regulated by both price responses and exogenous trends in pro- gregate region. Major fishing/aquaculture nations in Asia—namely duction and efficiency surrounding feed and feeding practices. China (CHN), Japan (JAP), and India (IND)—are separated from their In the Fish to 2020 study, the model started its simulations in the year corresponding regions to give special consideration in the analysis. 1997 and carried out projections into 2020. In the current study, the model begins its projections in the year 2000 and carries forward Dynamics to 2030. Due to data constraints for some of the existing IMPACT The IMPACT model finds a global market equilibrium in each period model components (such as land cover, irrigation maps, and some (typically the time step is a year) and continues sequentially over hydrology measures), the model base year is set at year 2000 even the projected time horizon. To introduce dynamics, the IMPACT though more recent commodity data are available (see section 2.3). model incorporates trends in the drivers of change for demand and Nonetheless, we make use of the early years of projection as a com- supply, and these are specified exogenously. Key drivers of demand parison period for calibration purposes. We use those years to evalu- are income and population growth. On the supply side, exogenous ate the fit of the model projections to the existing data so as to gain drivers are those outside the supply response to price changes; they further confidence in the medium-term projections to 2030. This F I S H TO 2 030: P R O S P E C T S F O R F I S H E R I E S A N D A Q UA C U LT U R E C H A P T E R 2 — P R E PA R I N G I M PA C T M O D E L F O R F I S H T O 2 0 3 0 13 FIGURE 2.1: Definition of Aggregate Regions for Results Reporting AFR CHN EAP ECA IND JAP LAC MNA NAM SAR SEA ROW TABLE 2.3: Abbreviation Code for Aggregate Regions fish-related components of the model. Additional details on the REGION ABBREVIATION NOTE non-fish components of the model can be found in the model de- ECA Europe and Central Asia, including developed nations scription by Rosegrant and the IMPACT Development Team (2012). NAM North America (United States and Canada) LAC Latin America and Caribbean Single Global Market and Single World Price EAP East Asia and the Pacific, including Mongolia and developed The basic modeling approach of IMPACT is a partial equilibrium nations, excluding Southeast Asia, China, and Japan CHN China representation of perfect, competitive world agricultural markets JAP Japan for crops, livestock, and fish. Supply and demand relationships SEA Southeast Asia for those commodities are linked to each other within a relatively SAR South Asia, excluding India simple representation of world trade, where all countries export to IND India and import from a single, integrated world market for each com- MNA Middle East and North Africa modity (Rosegrant Agcaoili-Sombilla and Perez 1995, Rosegrant AFR Sub-Saharan Africa and others 2001). The model reaches equilibrium in each market ROW Rest of the world, including Greenland, Iceland, Pacific small island states by solving for the single world price that balances the net exports and imports for all countries so that the market effectively clears globally. At the country level, the supply and demand of each com- kind of comparison was not conducted explicitly in the Fish to 2020 modity adjust according to price movements; the adjustment is analysis, and this represents a significant improvement in this work. regulated by commodity-specific price elasticities of supply and demand.6 At the country level, there can be either a net surplus 2.2. IMPACT MODEL STRUCTURE Here we describe the IMPACT model in more detail so that the reader can gain a deeper understanding of its structure and in- 6 Price elasticities represent the percentage change in demand or supply that occurs as a result of a unit percentage change in price of the good ner workings. Given the purpose of this report, we focus on the in question. A G R I C U LT U R E A N D R U R A L D E V E LO P M E N T D I S C U S S I O N PA P E R 14 C H A P T E R 2 — P R E PA R I N G I M PA C T M O D E L F O R F I S H T O 2030 or deficit, which is to be reconciled on the global market through by changes in profitability. Thus, no price elasticities of supply are international trade. specified for capture fisheries. Given the way the model handles international trade, in the pre- As a way to introduce dynamic trends in the supply relationships, sentation of the projection results, trade for each country or region exogenous growth rates are specified for each commodity in each is expressed in terms of net export. A net import is expressed in a country and multiplied to both capture and aquaculture supply negative value. The model structure does not permit a separate functions. This operation is essentially equivalent to shifting sup- identification of countries that are both importers and exporters of ply curves to the right or left according to some exogenous trends. a particular commodity. Neither does the model identify bilateral Typically, these exogenous growth factors represent an increase in trade flows. While some other trade models may represent differ- productivity over time that comes from improvements in technol- ential preferences for imported and home-produced goods, there ogy and technical efficiency so that more output is obtained at the are no explicit functions for export supply or import demand in same cost. How the growth rates are estimated for capture fisheries IMPACT. In effect, each commodity in the model is assumed to be of and aquaculture production is discussed in the next section. homogenous quality. And thus it is assumed that consumer prefer- ences differentiate commodities in terms of the broad categories Food Consumption Demand described in tables 2.1 and 2.2 but not in terms of their origin. The consumption of agricultural food commodities in a country is expressed as the product of per capita consumption and the total Supply Functions population. For both existing commodities and newly added fish In contrast with a model that seeks to explicitly optimize the alloca- products, reduced-form demand functions regulate per capita tion of resources to the production of various goods (for example, consumption demand in the model. Demand for a good in this maximizing total welfare), the IMPACT model uses reduced-form model is a function of the good’s own price, prices of other food supply functions. Specified for each commodity in each country, products, and the person’s income level. The own- and cross-price supply functions determine the optimal amount of goods to pro- elasticities of demand represent the preferences for increasing duce given the profitability (as represented by input and output consumption in response to a more favorable consumer price or prices) and existing resource constraints. for substituting toward other goods as their prices become rela- tively more favorable. Income elasticity of demand represents the The supply of crops and livestock products is represented as the tendency of consumers to consume more or less of a product as product of yield and planted area or animal numbers. In the case of their incomes rise or fall. Dynamics of consumption demand in a fish supply, there are two supply functions for a given fish species: country are introduced through exogenous trends in total popula- one that regulates supply from capture fisheries and the other from tion and income growth, which shift the demand curves (usually to aquaculture. The sum of the two determines the total supply in a the right) over time. country. Supply functions for aquaculture portray the expansion or contraction of production in response to changes in aquaculture profitability. Aquaculture profitability in the model is characterized Crush/Reduction Demand for Oil Extraction and Meal Production by the price of the fish species, prices of other species, and prices Another type of demand represented in the model is the demand of fishmeal and fish oil (critical inputs to aquaculture).7 Accordingly, for commodities to be used in oil extraction. In the IMPACT model, price elasticities of supply with respect to own price are positive, soybeans and two groups of oilseeds are “crushed” for oil extraction while those with respect to fishmeal and fish oil prices are negative. (table 2.1). Oil extraction from these oil-bearing crops produces by- In contrast, supply of capture fisheries is assumed not to be affected products (meal) that become important animal feed. As a new ad- dition in this study, the model includes “reduction” of small pelagic 7 At this point, soybean meal price is not included in the aquaculture sup- ply functions. and other fish for production of fishmeal and fish oil. These are also F I S H TO 2 030: P R O S P E C T S F O R F I S H E R I E S A N D A Q UA C U LT U R E C H A P T E R 2 — P R E PA R I N G I M PA C T M O D E L F O R F I S H T O 2 0 3 0 15 important animal feed products, and the use of fish oil for direct Only own price enters a feed demand function for aquaculture consumption as nutritional supplements has increased in recent production, thus denying the possibility of price-driven substitution years (Shepherd 2012). The demand for raw commodities used for among feed items.8 Instead, substitution among aquaculture feeds oil and meal production purposes is also represented in reduced- (especially between fishmeal and soybean meal) is exogenously in- form demand functions. In the model, the crush/reduction demand troduced in the form of trends in feed use coefficients (feed conver- depends on the price of the oil-bearing commodity and the prices sion ratios, or FCRs), which enter feed demand functions as param- of oil and meal. The demand decreases with higher price of the oil- eters. Trends in FCRs also reflect feeding efficiency improvements bearing commodity and the demand increases with higher price of that occur over time. In aquaculture, constant innovation in feeds oil and meal. and feeding practices has contributed importantly to the dramatic expansion that the sector has witnessed over the past decades For the production of fishmeal and fish oil, the model allows the use (Rana, Siriwardena, and Hasan 2009). Thus, FCR trends represent an of fish processing waste, and the model includes a simple demand important exogenous demand shifter in the model. The definition function for processing waste. We assume no explicit market for of FCRs and their specifications are discussed in greater detail in the processing waste so the price for waste does not exist. The demand next section. function for fish processing waste contains only fishmeal price as its argument. Other Demand The reduction demand for whole fish or processing waste has no All other types of demand for fish are simply treated as an exog- exogenous driver of change in the model. Conversion of units from enous amount. While the base-year level of “other demand” is de- primary production (oil-bearing crops and fish) to products (oil and termined according to the data, it is subsequently assumed to grow meal) is regulated by crush (reduction) ratios. Derivation of these in proportion to total demand. ratios for fishmeal and fish oil is discussed in the next section. Links of Supply and Demand Bringing together the supply and demand relationships that we Feed Demand have described thus far, we can portray the way fish products are In livestock and aquaculture production, feed is the single most modeled in IMPACT as in figure 2.2. This parallels the way in which important input. Producers may choose a least-cost feed ration to non-fish commodities are modeled (figure not shown). The pro- achieve a certain production target, which could be represented duction is differentiated between capture and aquaculture, with in a model by a cost-minimization problem subject to explicit con- capture production growth being treated as completely exogenous straints, such as minimum nutrient requirements. In this study, as while aquaculture supply is price responsive. On the demand side, with supply and consumption and processing demands, we adopt human (direct) consumption accounts for most fish use, while a reduced-form approach that allows for price-driven adjustments lower-value species are demanded for reduction into fishmeal and in feed demand according to specified price elasticities. fish oil. Much of the fishmeal and fish is used in aquaculture produc- A feed demand function is defined for each feed commodity. For tion. Thus, the capture and aquaculture segments of the sector are each of the feed commodities used in aquaculture (fishmeal, fish connected both in food fish markets and feed markets. oil, and soybean meal), two separate functions are specified: one for aquaculture production and one for livestock production. For Price Transmission aquaculture production, a feed demand function contains the price All prices in the model are keyed to the world prices that clear global of the feed (fishmeal, fish oil, or soybean meal) and the levels of markets for each traded commodity. However, at the country level, aquaculture production. Since the latter is determined through the supply function, feed demand is also indirectly affected by prices of 8 Prices of other feed items enter in feed demand functions for livestock fish products. production. A G R I C U LT U R E A N D R U R A L D E V E LO P M E N T D I S C U S S I O N PA P E R 16 C H A P T E R 2 — P R E PA R I N G I M PA C T M O D E L F O R F I S H T O 2030 FIGURE 2.2: Schematic of Links of Fish with Fishmeal and Fish Oil in the IMPACT Model Fish Scenario-specific Fish oil & meal Capture growth Aquaculture growth T T S S Fish oil & meal Aquaculture T Production T Capture (exog) production Trade Processing waste Conversion Trade R equilibrium technology equilibrium RS Price Price balance balance Reduction demand R R T Other demand S T Other demand S S Demand Fish oil & meal T T GDP growth Food demand demand Food demand Population growth Feed demand Linked to aquaculture production growth producer and consumer prices9 are allowed to deviate from the world crop side (such as irrigated/rain-fed area and water availability) have prices as a result of policy-driven factors (that is, subsidies and taxes) not yet been updated beyond year 2000. Consequently, we initi- and marketing and transaction costs. Deviations from the world prices ate the model in 2000 in all simulation runs while taking advantage are embodied in three country-specific parameters: the subsidy equiv- of the period for which actual data are available in calibrating the alents for producers and consumers (PSE, CSE) and the marketing model parameters (see section 2.5). margin (MM). Marketing margins reflect a variety of infrastructural and The code of the IMPACT model is written in the General Algebraic market imperfections at the country level that add to the prices that Modeling System (GAMS) programming language. The model is consumers pay for imports (or what producers lose in export value). solved as a system of simultaneous equations. Exact solutions are possible because the problem is set up such that the number of Summary of Model Structure equations matches the number of free variables. The current model Table 2.4 summarizes the key variables in the IMPACT model and solves for an equilibrium solution across 42,267 endogenous vari- indicates which variables are exogenously given and endogenously ables in total. determined in the model. Solution Procedure 2.3. DATA USED AND PARAMETER SPECIFICATION In this study, the base year of 2000 is used as the starting point of the Given the introduction in section 2.2 of the structure of the IMPACT simulations. Although more recent data are available for most com- model and the key variables and parameters incorporated in the modities from the FAO databases, some important data used on the model, in this section we describe the data used this study. In particular, the data are important to establish a consistent picture 9 The terms producer and consumer here are applied in relation to the of the global fish markets in the base (initial) year of the projection prices that enter, respectively, into the supply or demand equations, for a specific product in question. For example, an aquaculture producer (section 2.4) as well as for calibration purposes (section 2.5). This sees the producer price for the fish commodity, while the consumer section also describes model parameter specifications and the es- price affects the demand for it. However, the fishmeal that is used as an input into aquaculture production enters as an intermediate price. timation procedure for some of the parameters. These parameter F I S H TO 2030: P R O S P E C T S F O R F I S H E R I E S A N D A Q UA C U LT U R E C H A P T E R 2 — P R E PA R I N G I M PA C T M O D E L F O R F I S H T O 2 0 3 0 17 TABLE 2.4: Summary of Key Variables in the IMPACT Model CROPS LIVESTOCK FISH NOTES Area n.a. n.a. Price responsive with exogenous growth Yield n.a. n.a. Price responsive with exogenous growth n.a. Numbers n.a. Price responsive with exogenous growth n.a. Yield n.a. Completely exogenous n.a. n.a. Total supply Capture: completely exogenous Aquaculture: price responsive with exogenous growth Food demand Food demand Food demand Price and income responsive with exogenous growth Crush demand n.a. Reduction demand For oil-bearing crops and whole fish/fish processing waste Price responsive, no exogenous growth Feed supply n.a. Feed supply Crop: coarse grains and meals produced from oil-bearing crops Fish: fishmeal and fish oil Supply determined by crush/reduction demand multiplied by fixed crush/reduction ratios n.a. Feed demand Feed demand By livestock and aquaculture production Price responsive and dependent on livestock and aquaculture production volumes; with exogenous growth in feed conversion ratios Separate feed demand functions for livestock and aquaculture Biofuel demand n.a. n.a. With exogenous growth (according to policy scenario) Other demand Other demand Other demand Changes in strict proportion to sum of other (endogenous) demand categories Trade Difference between supply and demand forced to balance globally Prices World prices: endogenously determined to balance global trade Country prices: linked to world prices with producer/ consumer subsidy equivalents and marketing margins Note: n.a. = not applicable. values form the baseline scenario, whose results will be presented § Production (Capture + Aquaculture) in chapter 3. The discussions here focus on the data and param- § Meals Input eters for fish products that are newly added to the IMPACT model § Other Non-Food Use for this study. Data and parameters in the larger IMPACT model are § Exports for Human Consumption described in the model description by Rosegrant and the IMPACT § Imports for Human Consumption Development Team (2012). 10 § Total Food Fish Supply § Per Capita Food Fish Supply In this study, we use three broad sets of fish-related data provided § Stock Variation by the FAO (consumption-trade, production, and fishmeal–fish oil)11 § Population as well as price data from multiple sources. The Meals Input series form the basis of the crush/reduction de- Consumption and Trade Data mand functions discussed in the previous section, while the Other For apparent consumption and trade data, this study relies on FAO Non-Food Use series correspond to the other demand category in FIPS FBS of fish and fishery products.12 Data are available for 226 IMPACT. The Stock Variation series represent the residuals between countries or areas for the following domains: supply, demand, and trade each year for each country. Since the products in FAO. These data are disseminated through the FAO Yearbook: 10 Some parameters used in this study are based on the older version of Fisheries and Aquaculture Statistics and through FAOSTAT at http:// IMPACT (see Rosegrant and others 2001). faostat.fao.org/site/617/default.aspx#ancor. However, it is important to 11 FAO data were received from the FIPS of the FAO Fisheries and Aqua- highlight that notwithstanding the same source and final results in terms culture Department in fall 2011. of supply, data in the two domains are presented according to a different 12 FAO FIPS is responsible for the calculation of FBS of fish and fishery methodology related to the treatment of non-food commodities. A G R I C U LT U R E A N D R U R A L D E V E LO P M E N T D I S C U S S I O N PA P E R 18 C H A P T E R 2 — P R E PA R I N G I M PA C T M O D E L F O R F I S H T O 2030 data on the actual level of stock are unavailable, IMPACT does not The conversion process is a complex one and, as a result, a substan- explicitly model stock-holding behavior. To accommodate the tial degree of error can arise. Thus, we proceed with caution in using existence of this category in the dataset, the IMPACT model run is consumption and trade data when we seek internal balance for the initiated with the value of stock change for the base year (2000) and IMPACT model between production, consumption, and trade for progressively reducing it to zero over the first 5 to 10 years of the the base year (section 2.4). simulation. The correspondence of other series in the FAO FIPS FBS At the time of model preparation, these data series were available dataset with IMPACT variables is self-explanatory. to the team for the years 1976–2007. The fact that newer data series For fish and fishery products, the FAO calculates FBS for eight were not available for consumption and trade data limits the scope 13 groups of similar biological characteristics. In order to achieve a of calibration exercises for these variables. In the assessment of the more consistent link between fishery trade and consumption data quality of model output in section 2.5, comparisons are made be- and production statistics for the use in the IMPACT model, the ex- tween model projections and FAO FIPS data for the years 2000–06 isting groups for fishery trade and consumption data have been for variables related to consumption and trade. (The data are pre- modified through the creation of ad hoc categories. However, due sented in three-year moving averages. Thus the data for 1999–2007 to the limitation of raw trade data availability for selected species, in are used in the comparisons.) particular for freshwater fish, it was not possible to establish a fully comparable one-to-one link with production series. FAO fishery Production Data trade data reflect the national classifications used by the countries For data on primary fish production, this study relies on the FAO fish- to collect and report their trade. These classifications are generally eries databases available through FishStat.15 The data series available based on the Harmonized System (HS) classification of the World in FishStat include primary production by systems (aquaculture and Customs Organization (WCO), which is used as a basis for the col- capture) and trade (import/export, frozen/chilled/processed).16 All lection of customs duties and international trade statistics by more series are available in both volume (tons) and, with the exception of than 200 countries. Only starting with the new version, entered into capture fisheries, value (in U.S. dollars). At the time of model prepa- 14 force on January 1, 2012, selected freshwater species, including ration, these data were available for the years 1984–2009. Thus, for tilapia, catfish, and carps, are identified in the HS, while in previous production-related series, the comparison between data and model versions only live carps had a separate code. projections is provided for the years 2000–08 in section 2.5. (Again, the data are presented in three-year moving averages, and the data In the IMPACT model, the consumption and trade series are avail- for 1999–2009 are used in the comparisons.) able for nine aggregate fish commodities. In contrast, production series come in much more disaggregated categories (FishStat da- The primary production data in FishStat are highly disaggregated by tabase, see next). Thus, these nine categories form the lowest com- fish species (over 2,000 species or groups of species). For tractability mon denominators in defining the fish product categories for this in the IMPACT model, the fish species are aggregated in this study. study. The nine fish commodities are shown in the consumption In doing so, we maintain consistency between the aggregated category in table 2.2. production series and consumption and trade data series, while al- lowing a certain degree of disaggregation so that some key policy The FAO FIPS series expresses the volume of various types of seafood in terms of live weight equivalent or the unit of primary production. 15 See this link for details on FishStat: http://www.fao.org/fishery/statistics/ software/fishstat/en. 13 The eight groups are freshwater and diadromous fish, demersal fish, pe- 16 Trade data are available in both FAO FIPS FBS and FishStat. The trade lagic fish, marine fish unspecified, crustaceans, cephalopods, mollusks series in FishStat is measured in terms of product weight, rather than other than cephalopods, and other aquatic animals. live weight equivalent as in FAO FIPS FBS. Since FAO FIPS FBS is the only 14 HS 2012 reflects the FAO joint proposal to the WCO for the revision of source of apparent fish consumption data at the world level, which are the codes related to agriculture, forestry, and fishery products, with a measured in live weight equivalent, the trade series from FAO FIPS FBS resulting improved breakdown of selected fishery species. are used in this study. F I S H TO 2 030: P R O S P E C T S F O R F I S H E R I E S A N D A Q UA C U LT U R E C H A P T E R 2 — P R E PA R I N G I M PA C T M O D E L F O R F I S H T O 2 0 3 0 19 research questions can be addressed using the projections. The TABLE 2.5: Number of Countries Covered in FAO Fishmeal species aggregation is shown as production category in table 2.2. and Fish Oil Dataset, 1976–2009 PRODUCTION IMPORT EXPORT When selecting the aggregation rule of fish species, special consid- Fishmeal 86 202 159 eration was given to their diet, since it relates to the types of feed Fish oil 54 201 143 aquaculture production uses. For instance, we wished to separate Source: Compilation of data from FishStat, Oil World, and the IFFO. those fish species that require a relatively high percentage of animal protein in their diets (for example, salmon, tuna, and demersals such composition is captured in net export only to the degree that as snapper, cod, halibut, and flounder) from those that can be grown there is a difference in the volume of importation and exportation. mainly on a plant-based diet (for example, tilapia, Pangasius and Accordingly, in the process of obtaining the consistent base-year other catfish, carps and other cyprinids, and milkfish). Furthermore, picture of the global fish markets, we had to ensure that only these are distinct from other species that are not usually fed directly countries that produce fishmeal or fish oil could have positive net but are grown in fertilized bonds, such as silver and grass carp, or export. The details of base-year establishment are presented in those that live off of detritus and/or plankton, such as mollusks and section 2.4. other invertebrates. In the Fish to 2020 study shrimp, prawns, and Fishmeal and fish oil are produced by reducing whole fish caught other crustaceans were combined in a single category. They are for that purpose and bycatch and/or other low-value species, as well separated in this study, given that shrimp is an important commod- as waste from poor postharvest handling or from the processing of ity by itself in the world seafood market and that data of relatively fish into fillets and other value-added products. It is estimated that good quality are available for shrimp. currently about 25 percent of fishmeal produced globally uses fish processing waste as ingredient (Shepherd 2012). The FAO FIPS FBS Fishmeal and Fish Oil Data dataset includes the series “Meals Input” that represents the volume The dataset for fishmeal and fish oil provided by the FAO contains of various types of whole fish used for reduction into fishmeal and series for production, imports, and exports for the years 1976–2009. fish oil. However, country-level data on the volume of processing The dataset is a compilation of data from FishStat, Oil World,17 waste used for reduction are not available. The use of fish process- and the International Fishmeal and Fish Oil Organisation (IFFO).18 ing waste is imputed for 2000, as a part of the base-year establish- The number of countries represented in the dataset is shown in ment. (See section 2.4 and technical appendix C to this chapter.) table 2.5. Finally, since the datasets provided by the FAO do not contain series A series of data preparation tasks were necessary in order to achieve on the use of aquaculture feed (fishmeal, fish oil, and soybean meal), consistency between the reported volume of fishmeal and fish oil these are also imputed such that they balance with production and production, reported amount of whole fish used in reduction, and trade on a country and global level. See section 2.3’s subsection reported volume of fishmeal and fish oil exported by each country. Feed Conversion Ratios and FCR Growth and the technical appendix C First, table 2.5 indicates that there are more exporter countries of to this chapter for details on the imputation procedure. fishmeal/fish oil than producers in the data. Given the treatment of international trade in the IMPACT model, all fishmeal and fish oil World Prices are homogenous and trade is expressed in terms of net exports. World prices series are generated in a manner consistent with the As a result, for example, importation of a product for the purpose definition of world prices specific to the IMPACT model. World of reexportation or exportation after reformulation of the product prices for traded commodities are derived in three steps. First, for each commodity, a group of countries is defined that, combined, 17 http://www.oilworld.biz. constitute the bulk of world exports (see technical appendix A to 18 http://www.iffo.net/. this chapter for the procedure). Second, the world price is calculated A G R I C U LT U R E A N D R U R A L D E V E LO P M E N T D I S C U S S I O N PA P E R 20 C H A P T E R 2 — P R E PA R I N G I M PA C T M O D E L F O R F I S H T O 2030 as the weighted average of export unit value (FishStat) in each of Capture Growth the above countries. The weight used is the export share of each Exogenous trends of capture fisheries incorporated in the model are country, in live weight equivalent, in the sum of all the selected estimated using the time series data from the FAO FishStat data- major players in the market. Third, the world prices calculated for base. The estimates are based on the best fit to observed data over individual fish species are aggregated to consumption category us- the 2000–08 period and plausible trajectories beyond 2008. ing trade volume as weight. For fishmeal and fish oil, price series Aquaculture Growth provided by the IFFO are used. More careful estimation of exogenous growth rates is conducted Parameter Specification for aquaculture production using FAO FishStat data (1984–2009) Lastly, we discuss the specification of the parameters of the IMPACT for each commodity in each country. Three patterns are observed model. Given the sheer number of parameters used in the model, it in the data and, for each case, the following approach is used to is not possible to describe them all here. While further explanations specify growth rates. First, when the production is trending upward, are provided for some of the parameters, key groups of parameters a logistic growth curve is fitted to the data using least squares are listed in table 2.6 together with their sources. method. Second, when there is a downward trend, the projection TABLE 2.6: List of Key Parameters in the IMPACT Model and Their Sources PARAMETER DESCRIPTION DATA SOURCE ELASTICITIES Area elasticity Own- and cross-price elasticities of crop area Modified from values in Appendix B, Rosegrant and others 2001 Livestock elasticity Own- and cross-price elasticities of livestock numbers Yield elasticity Own-price elasticities of crop yield FoodDmd elasticity Own- and cross-price elasticities of food demand IncDmd elasticity Income elasticities of food demand Feed elasticity Price elasticities of demand for feed commodities Fish elasticity Own- and cross-price elasticities of aquaculture supply From the model used in the Fish to 2020 model (Delgado and others 2003), modified to cover additional fish categories and regions and to match the FoodDmd elasticity [Fish] Own- and cross-price elasticities of food fish demand observed supply growth in the 2000–08 period IncDmd elasticity [Fish] Income elasticities of food fish demand Crush elasticity Price elasticities for crush/reduction demand Own estimates, adjusted in calibration process Crush elasticity for waste Price elasticities for reduction demand for fish processing waste Own estimates, adjusted in calibration process OTHERS PSE, CSE Producer and consumer subsidy equivalents From the model used in the Fish to 2020 model (Delgado and others 2003), modified to cover additional fish categories and regions MM Marketing margin FCR (feed conversion ratio) Amount of feed required per unit of livestock and aquaculture production (defined [Fish] Estimated based on Tacon and Metian 2008 (see text and technical for each of fishmeal, fish oil, and soybean meal) appendix C to this chapter) Reduction ratio Amount of fishmeal and fish oil produced (in product weight) per unit of whole [Fish] Estimated based on FAO data and Jackson 2010 (see text and technical fish used (in live weight equivalent) appendix C to this chapter) Waste ratio Amount of fish processing waste generated from a unit of whole fish (in live From various sources (see text and technical appendix C to this chapter) weight equivalent) EXOGENOUS GROWTH Population growth rate Exogenous growth rates of human population UN Medium Variant Population projections (UN 2011) Income growth rate Exogenous growth rates of income (gross domestic product, or GDP) World Bank Global Economic Prospects projections and data (World Bank 2012) Capture growth rate Exogenous growth rates of capture fisheries production Estimated using historical data (see text) Aquaculture growth rate Exogenous growth rates of aquaculture production Estimated using historical data (see text) FCR growth rate Exogenous growth rates of feed conversion ratio Estimated using the imputed FCRs F I S H TO 2 030: P R O S P E C T S F O R F I S H E R I E S A N D A Q UA C U LT U R E C H A P T E R 2 — P R E PA R I N G I M PA C T M O D E L F O R F I S H T O 2 0 3 0 21 is fixed at the latest observation (2009) for the rest of the projection production data for 2000, we estimate the most plausible amount of period (2010–30). Third, when the recent trend exhibits fluctuations feed that must have been used per unit of livestock and aquaculture around a stable average, a mean is calculated for the appropriate output produced. The process is repeated for 2009, and FCR growth duration and used as the projected value for 2010–30. rates are calculated based on the two sets of FCR estimates for fish- meal and fish oil. Except for a few cases, the calculated growth rates For salmon aquaculture in Chile, to account for ISA outbreak during of fishmeal and fish oil FCRs are negative, supporting the general 2007–10, a negative growth rate of 3.9 percent per year is imposed tendency that their usage per unit of aquaculture production is for the 2005–10 period, after which the recovery is assumed to start decreasing. See technical appendix C to this chapter for additional at the rate estimated for the 2000–05 period from the fitted logistic detail of the estimation process. curve and to continue at the subsequent rates. Reduction Ratio and Waste Ratio Feed Conversion Ratios and FCR Growth Reduction ratio and waste ratio are used in the model to calculate A feed conversion ratio (FCR) represents the quantity of feed re- the amount of feed produced from whole fish or fish processing quired per unit of livestock or aquaculture production. This set of pa- waste. While waste ratios are derived from the literature (see techni- rameters is used in the model to calculate the amount of total feed cal appendix C to this chapter for the sources), reduction ratios are used for livestock and aquaculture production. For aquaculture, we estimated in the process of establishing the base-year picture. See define FCR for three feed items: fishmeal, fish oil, and soybean meal. section 2.4 and technical appendix C to this chapter for the proce- Thus, the use of the term FCR in this study is slightly different from dure of reduction ratio estimation. the conventional sense in the literature, where the concept is usually applied to the total volume of feed used to produce unit volume of We have described the specification of key parameters that are meat or fish. In this study, starting with initial FCR values in the base added in this version of the IMPACT model. In principle, these pa- year (2000), they are allowed to evolve over time. Given the defini- rameter values form the basis of the baseline specification of the tion of the FCR adopted in this study, the FCR evolution implies the model, whose results are presented in chapter 3. However, it must composite of two separate effects: (1) substitution between fish- and be noted here that these parameter values are further fine-tuned plant-based feedstock and (2) efficiency improvement in feed use. individually in order to achieve consistency across data series in the base year and to obtain reasonable model results under the baseline As discussed earlier, there is constant innovation in the aquacul- scenario. The definition of reasonable is one of expert judgment and ture feed industry, such that lower-cost plant-based alternatives has been subject to much discussion and adjustment in the course are increasingly used in feed formulation, substituting away from of constructing the baseline results for this report. Such adjustment higher-cost fishmeal and fish oil (Barnes and others 2012; Hardy efforts and their outcomes are presented in the next two sections. 1996, 2003; Naylor and others 2009; Rana, Siriwardena, and Hasan 2009). The quality of feeds has also been improved in terms of 2.4. ESTABLISHING A CONSISTENT their digestibility, while at the same time, efficiency has increased BASE-YEAR PICTURE through better feeding methods and farm management in general In this study, all simulations are initiated in the base year of 2000. The (Tacon, Hasan, and Metian 2011). Therefore, the general tendency next step in model preparation is to establish a credible picture of around the globe is a reduction in FCRs for fishmeal and fish oil. fish supply, demand, and trade for the base year. Since all projections However, in a country where aquaculture feeds were not intensively for the subsequent years depend on this estimate, appropriately used previously, improvement in feeding practices may imply an representing the base-year picture of the global agricultural com- increase in any or all of the three FCRs. modity markets is extremely important in the use of the IMPACT For the purposes of this study, aquaculture FCRs and their growth model. In assembling a number of basic components of the model, rates are estimated using available information. Using the FAO the challenge is to obtain internal consistency in data series across A G R I C U LT U R E A N D R U R A L D E V E LO P M E N T D I S C U S S I O N PA P E R 22 C H A P T E R 2 — P R E PA R I N G I M PA C T M O D E L F O R F I S H T O 2030 those components. Since a consistent base-year picture has been other countries in the model, causing an imbalance elsewhere. The established for the non-fish commodities prior to the inclusion of effect ripples through all of the price-quantity-trade relationships fish products in the model, the discussions here focus on such tasks and throws off the balance such that the model cannot reproduce for fish products. the observed data in the base year of the simulation. The difficulty in obtaining data consistency is further exacerbated in this model Balance of Supply, Demand, and Trade at Country Level version by the endogenous treatment of fishmeal and fish oil pro- Simply put, for each fish commodity in each country in a given year, duction and their use in aquaculture production. the following condition must hold for consistency: Figure 2.3 depicts the data links and the “adding-up conditions,” or (1) Food Fish Demand + Reduction Demand + Other Demand + conditions for market clearance (equilibrium) in the fish part of the Export = Production + Import. model. (See technical appendix B to this chapter for the complete However, since production and consumption-trade data come from set of adding-up conditions for the IMPACT model.) As a funda- different datasets, this equality is not guaranteed to hold in the raw mental requirement, the balance of supply and demand must be data for most country-commodity combinations. reached for each commodity in each country as in equation (1). At the same time, the global sum of net trade (exports minus import Adding-Up Conditions at the Global Level flows) must equal zero for each commodity. In other words, every Due to the complex links in the model as depicted in figure 2.2, an ton of exports from any country must be imported by some other imbalance in this relationship in a given country will be “exported” to country, with no residual left on the market. FIGURE 2.3: Key Data Relationships Used to Balance Fish Data in IMPACT Country-level supply and demand balance Production + Net imports = Demand (Aquaculture, Capture) (Food, Feed, Crush, Other) Waste from fish processing Global-level trade balance Feed conversion ratios Fish ∑ Global Net imports = 0 Crush ratios Country-level supply and demand balance Fish meal & oil Demand + Net exports = Production (Whole fish, waste) Global-level trade balance ∑ Global Net exports = 0 F I S H TO 2 030: P R O S P E C T S F O R F I S H E R I E S A N D A Q UA C U LT U R E C H A P T E R 2 — P R E PA R I N G I M PA C T M O D E L F O R F I S H TO 2030 23 This is the case for trade in fishmeal and fish oil as well. As a re- In relation to fishmeal and fish oil production: sult, the link between fishmeal, fish oil, and aquaculture results § By estimating reduction ratios, the levels of fish processing in two more adding-up conditions. First, the fishmeal and fish oil waste used in fishmeal and fish oil production are imputed. demand quantities have to be consistent with the production § In doing so, in order to achieve internal consistency, the levels of aquaculture, such that there is sufficient feed demand for amount of whole fish used in fishmeal and fish oil produc- fishmeal/fish oil to justify the quantities of aquaculture produc- tion is allowed to deviate from “Meals Input” data series. tion. Second, the demand for whole fish for reduction (plus the § As a result, the levels of food fish consumption demand and imputed value of processing waste) has to be consistent with the fish trade volume are also allowed to deviate from the FAO production quantities of fishmeal and fish oil that are reflected in data. § The amount of fish supplied (by capture fisheries or aquacul- the data. ture) is never allowed to deviate from the data. Inconsistency across Datasets In relation to fishmeal and fish oil utilization: As discussed in section 2.3, the addition of fish products to the § By estimating FCRs and based on the aquaculture produc- IMPACT model for this study relies on three FAO datasets: tion data, the levels of fishmeal and fish oil use are imputed. § In doing so, in order to achieve internal consistency, the § Dataset I: Contains disaggregated data for fish production levels of trade and production of fishmeal and fish oil are data (FishStat). allowed to deviate from the data. § Dataset II: Contains aggregated consumption and trade data (FAO FIPS FBS). A consistent base-year picture is obtained with a help of a GAMS § Dataset III: Contains fishmeal/fish oil production and trade data. program developed specifically for the purposes. The program estimates reduction ratios and FCRs, while fixing the value of fish The three datasets contain data obtained from different sources and production variables to the levels indicated by data and penalizing domains and are not specifically prepared for the purpose of being the deviations of other variables from the data. This ensures the used together. Therefore, it is not surprising that the data series do base-year values of production, consumption, and trade are inter- not satisfy the adding-up conditions of figure 2.3. nally consistent and as close to the original FAO data as possible. See technical appendix C to this chapter for the procedure and as- Establishing a Consistent Picture for 2000 sumptions used in the establishment of the base-year picture of the As a result, the task here is to find a set of values for the IMPACT global fish markets for 2000. variables that are internally consistent and satisfy the adding-up conditions for the base year. The basic strategy we take is to re- Setting the “Intercepts” of Supply and Demand Functions construct a consistent picture for the year 2000 by adhering to the Once the conditions in figure 2.3 are satisfied and a consistent available data as much as possible while allowing some variables to picture is obtained for the base year, initial values of the “inter- deviate from the levels indicated by the data. Note that the data are cepts” of the supply and demand curves are determined. These presented in three-year moving averages. And thus we define the intercepts regulate the position of the supply and demand curves “base” to be the average picture of the global agricultural market for and thus also serve to “scale” the values of model output. Since the years 1999–2001. they are derived from the consistent initial values of the variables, The data inconsistency originates mainly from the addition of fish- using these intercept levels in the model result in the perfect meal and fish oil to the model. As a result, deviations are allowed replication of the base-year picture for 2000. For the subsequent to a larger extent for variables related to fishmeal and fish oil. More years in the simulation, these intercept values are changed ac- specifically, establishing the base-year picture involves the follow- cording to the exogenous growth rates specified for each of the ing related processes. supply and demand functions. Figure 2.4 depicts the sequence A G R I C U LT U R E A N D R U R A L D E V E LO P M E N T D I S C U S S I O N PA P E R 24 C H A P T E R 2 — P R E PA R I N G I M PA C T M O D E L F O R F I S H TO 2030 FIGURE 2.4: Computational Steps and Sequence of the Model datasets are also relevant here. For these reasons, it is impossible for Read-in key data from model database. the model to reproduce all of the variables at the levels indicated Check for proper balance and adding-up by the data during the calibration period. The calibration objective is thus to generate projections that are closer to the actual data for Initialize intercepts of supply and demand equations with base year 2000 data relatively more important commodities for relatively more impor- tant players in those markets. Solve model for supply, demand & trade that are consistent with prices In doing so, differential priorities are assigned to series from the and balanced global trade three datasets according to the confidence given to the series and Go to next year datasets. Priorities are given to the data series for which no unit Shift intercepts of supply and demand functions according to conversion is necessary. The first priority is given to the production exogenous growth trajectory data from FishStat (Dataset I). The series are measured in live weight, which is the common unit throughout the model. Second, we pri- Update growth rates for population, income, area oritize production and trade data for fishmeal and fish oil (Dataset expansion & productivity gains III). These are measured in product weight. However, in the produc- Save solution values and do tion process, fish (whole or processing waste), which is measured additional post-simulation in live weight (or its equivalent), is converted to products, which is aggregation & calculations measured in product weight. Thus, a larger degree of discrepancy is expected for these series than for primary production. The largest discrepancy is expected for fish utilization (consumption, reduction, of model initialization, solution, and information updating within and other) and trade from FAO FIPS FBS (Dataset II), as these involve the IMPACT model. unit conversion of final processed consumption products back to live weight equivalent. 2.5. ASSESSING THE QUALITY OF PROJECTIONS In the next subsection, we discuss the results of the calibrated model. The final step in model preparation is to adjust parameter values so that subsequent model projections are sufficiently close to the Global Projections observed data for the calibration period (2000–08 for production Figure 2.5 compares the projections of global capture and aqua- series and 2000–06 for consumption and trade series). Projections culture production (represented by lines) to the FAO data (squares) under the baseline scenario are used for calibration exercises. over the 2000–08 period. The figure shows that, at the global level, Baseline projections for the years beyond the calibration period will the IMPACT model generates production projections that are very be discussed in chapter 3. close to the actual data. Note that, since fish production series are not allowed to deviate from the data in the base-year establishment Calibration is implemented by sequentially and manually fine- process, the projections coincide with the data for the year 2000. tuning model parameters, rather than using some algorithm to calibrate all parameters at once. Adjusted parameters are mostly Figure 2.6 plots the projections of the three categories of fish uti- elasticities, but in some cases exogenous growth rates of aquacul- lization at the global level against the FAO data over the 2000–06 ture production are also adjusted. Exogenous growth rates for food period. The utilization categories of food consumption, reduction, consumption demand are never adjusted. As depicted in figures 2.2 and other use are described in section 2.2, and they all add up to and 2.3, the model is built on extremely complex links of supply-de- the total demand. The IMPACT model output for total demand mand relationships for 115 model regions as well as global adding- matches the data very closely. In contrast, there is a slight gap be- up conditions. The issues of data quality and inconsistency across tween the model results and the data for the food and reduction F I S H TO 2 030: P R O S P E C T S F O R F I S H E R I E S A N D A Q UA C U LT U R E C H A P T E R 2 — P R E PA R I N G I M PA C T M O D E L F O R F I S H TO 2030 25 FIGURE 2.5: Comparison of Projections and Data for Global FIGURE 2.6: Comparison of Projections and Data for Global Fish Production, 2000–08 Fish Utilization, 2000–06 160 160 Projection total Projection total 140 140 Data total Data total 120 120 Projection food Projection capture Data food 100 Million tons 100 Million tons Data capture Projection 80 reduction Projection 80 aquaculture Data reduction 60 60 Data aquaculture Projection other 40 40 Data other 20 20 0 0 00 01 02 03 04 05 06 07 08 00 01 02 03 04 05 06 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 Sources: FishStat and IMPACT model projections. Sources: FAO FIPS FBS and IMPACT model projections. categories. As discussed in the previous section, the projections for FIGURE 2.7: Comparison of Projections and Data for Global reduction demand, and accordingly also those for food fish con- Fishmeal Production, 2000–08 sumption, are allowed to deviate from the data in order to achieve 8,000 Projection Data internal consistency across series for the base year. The difficulty 7,000 with data inconsistency continues into the calibration period, and 6,000 deviations from data are allowed according to the priority rule presented earlier. Here, deviations persist, especially for the food Thousand tons 5,000 and reduction demand series. However the deviations for the 4,000 two series balance out, and a good overall fit is obtained for total demand. 3,000 2,000 The overprojection of reduction demand is associated with the con- sistent overprojection of fishmeal production during the calibration 1,000 period (2000–08) relative to the available data (figure 2.7). Fishmeal 0 and fish oil demand are determined in the model as aquaculture and 00 1 2 03 04 05 6 7 08 0 0 0 0 20 20 20 20 20 20 20 20 20 livestock production times the corresponding FCRs. Thus, the pro- Sources: Compilation of data from FishStat, Oil World, and the IFFO and IMPACT model jected levels of aquaculture and livestock production largely drive projections. the projections of fishmeal and fish oil supply and demand. While FCRs derived from the literature reflect biological requirements of feed in aquaculture production, in many cases fishmeal and fish oil Projections by Region and Species availability in a country (production plus imports), as indicated by Figure 2.8 shows projections of capture fisheries production across FAO data, is insufficient to support the observed aquaculture pro- different regions of the world against FAO data for the year 2008. duction levels. According to the dataset prioritization rule, therefore, The match between the two series is very close for all major re- fishmeal and fish oil production projections are allowed to deviate gions except for the Latin America and Caribbean (LAC) region. The from the data, presuming that the reported feed supply is a likely overprediction of LAC capture fisheries for 2008 originates from the underestimate of what was actually used. overprediction of the reduction demand and fishmeal production. A G R I C U LT U R E A N D R U R A L D E V E LO P M E N T D I S C U S S I O N PA P E R 26 C H A P T E R 2 — P R E PA R I N G I M PA C T M O D E L F O R F I S H TO 2030 FIGURE 2.8: Comparison of Projections and Data for Regional FIGURE 2.9: Comparison of Projections and Data for Capture Capture Fisheries Production, 2008 Fisheries Production by Species, 2008 ECA Shrimp 2008 Data 2008 Projection Crustaceans NAM Mollusks Salmon LAC Tuna EAP Tilapia Pangasius /catfish CHN Carp JAP OCarp EelStg SEA OFresh MDemersal SAR Mullet IND CobSwf OPelagic MNA OMarine AFR - 0 00 00 00 00 00 00 00 00 ,0 ,0 ,0 ,0 ,0 ,0 ,0 5, 2008 Data 2008 Projection 15 20 35 10 25 30 40 ROW Thousand tons - 0 0 0 0 00 00 00 00 00 00 00 00 00 00 ,0 ,0 ,0 ,0 ,0 ,0 Sources: FishStat and IMPACT model projections. 2, 4, 6, 8, 10 12 14 16 18 20 Note: Pangasius/catfish = Pangasius and other catfish; OCarp = silver, bighead, Thousand tons and grass carp; EelStg = aggregate of eels and sturgeon; OFresh = freshwater and diadromous species (excluding tilapia, Pangasius/catfish, carp, OCarp, and EelStg); Sources: FishStat and IMPACT model projections. MDemersal = major demersal fish; CobSwf = aggregate of cobia and swordfish; Note: ECA = Europe and Central Asia; NAM = North America; LAC = Latin America OPelagic = other pelagic species; OMarine = other marine fish. and Caribbean; CHN = China; JAP = Japan; EAP = other East Asia and the Pacific; SEA = Southeast Asia; IND = India; SAR = other South Asia; MNA = Middle East and North Africa; AFR = Sub-Saharan Africa; ROW = rest of the world. Since fish from capture origin (mostly other pelagic category) is the FIGURE 2.10: Comparison of Projections and Data for major source of fishmeal ingredients and LAC is the largest fishmeal- Regional Aquaculture Production, 2008 producing region, the discrepancy between the projection and the ECA 2008 Data 2008 Projection data is exacerbated in this region. However, one could also argue NAM that small pelagics are subject to huge variations due to El Niño and LAC La Niña, as well as decadal oscillations. Without explicit modeling EAP of these oscillations, calibrated and simulated harvest behavior of CHN small pelagics likely deviate actual observation. JAP We also see a good fit of projections with the data for capture pro- SEA duction by species (figure 2.9). The fit in 2008 is fairly close for most SAR species, with a slightly larger deviation in the OPelagic category. IND MNA Again, this category of fish is most heavily used for the production AFR of fishmeal and fish oil through reduction, and the overprojection ROW in reduction demand and fishmeal production contributes to the - 0 00 00 00 00 00 00 00 overprojection of this category. For the same reason, but to a lesser 00 ,0 ,0 ,0 ,0 ,0 ,0 ,0 5, 15 20 40 10 25 30 35 degree, OMarine production is overpredicted, as they are also used Thousand tons for fishmeal and fish oil production. Sources: FishStat and IMPACT model projections. Note: ECA = Europe and Central Asia; NAM = North America; LAC = Latin America In figure 2.10, a very good fit between data and projections in 2008 and Caribbean; CHN = China; JAP = Japan; EAP = other East Asia and the Pacific; SEA = Southeast Asia; IND = India; SAR = other South Asia; MNA = Middle East and is confirmed for aquaculture production across all regions. North Africa; AFR = Sub-Saharan Africa; ROW = rest of the world. F I S H TO 2 030: P R O S P E C T S F O R F I S H E R I E S A N D A Q UA C U LT U R E C H A P T E R 2 — P R E PA R I N G I M PA C T M O D E L F O R F I S H TO 2030 27 FIGURE 2.11: Comparison of Projections and Data for FIGURE 2.12: Comparison of Projections and Data for Aquaculture Production by Species, 2008 Regional Per Capita Food Fish Consumption, 2008 Data 2008 Projection 2006 Shrimp 2006 Data 2006 Projection Crustaceans Mollusks ECA Salmon NAM Tuna Tilapia LAC Pangasius/catfish Carp EAP OCarp CHN EelStg OFresh JAP MDemersal SEA Mullet CobSwf SAR OPelagic IND OMarine MNA - 0 0 0 0 00 00 00 00 00 00 00 00 ,0 ,0 ,0 ,0 2, 4, 6, 8, AFR 10 12 14 16 Thousand tons ROW Sources: FishStat and IMPACT model projections. - 10 20 30 40 50 60 70 Note: Pangasius/catfish = Pangasius and other catfish; OCarp = silver, bighead, and grass carp; EelStg = aggregate of eels and sturgeon; OFresh = freshwater and kg / capita / year diadromous species (excluding tilapia, Pangasius/catfish, carp, OCarp, and EelStg); Sources: FAO FIPS FBS and IMPACT model projections. MDemersal = major demersal fish; CobSwf = aggregate of cobia and swordfish; Note: ECA = Europe and Central Asia; NAM = North America; LAC = Latin America OPelagic = other pelagic species; OMarine = other marine fish. and Caribbean; CHN = China; JAP = Japan; EAP = other East Asia and the Pacific; SEA = Southeast Asia; IND = India; SAR = other South Asia; MNA = Middle East and North Africa; AFR = Sub-Saharan Africa; ROW = rest of the world. Looking across species, the overall fit of 2008 aquaculture projection FIGURE 2.13: Comparison of Projections and Data for to the data is fair (figure 2.11). The largest divergence is observed Regional Total Food Fish Consumption, 2006 for MDemersal (overprediction by 22 percent) and Pangasius/catfish 2006 Data 2006 Projection (underprediction by 19 percent). The prediction errors in 2008 for ECA other species range between 1 and 13 percent in absolute terms. NAM Turning to the regional calibration of the demand side of the model, LAC figure 2.12 shows a comparison between the model projections and EAP FAO data for per capita food fish demand for 2006. JAP was by far the CHN JAP largest consumer of food fish per capita in 2006, followed by other SEA Asian regions: EAP, SEA, and CHN. The fit of the model projections with SAR the data in 2006 is reasonably good for most regions, with largest de- IND viations observed for JAP (overprediction by 7 kilograms), EAP (under- MNA prediction by 9 kilograms), and ROW (overprediction of 11 kilograms).19 AFR However, the deviations between projections and data in 2006 are ROW - 0 00 00 00 00 00 00 00 relatively small in terms of total food fish demand obtained as per 00 ,0 ,0 ,0 ,0 ,0 ,0 ,0 5, 15 20 25 35 10 30 40 capita demand times the population (figure 2.13). Thousand tons Sources: FAO FIPS FBS and IMPACT model projections. Note: ECA = Europe and Central Asia; NAM = North America; LAC = Latin America and Caribbean; CHN = China; JAP = Japan; EAP = other East Asia and the Pacific; 19 ROW includes a wide range of countries and this group was not the SEA = Southeast Asia; IND = India; SAR = other South Asia; MNA = Middle East and focus of calibration exercise. North Africa; AFR = Sub-Saharan Africa; ROW = rest of the world. A G R I C U LT U R E A N D R U R A L D E V E LO P M E N T D I S C U S S I O N PA P E R 28 C H A P T E R 2 — P R E PA R I N G I M PA C T M O D E L F O R F I S H TO 2030 FIGURE 2.14: Comparison of Regional Projections and Data FIGURE 2.15: Comparison of Projections and Data for World for Net Fish Export, 2006 Fish Prices, 2000–08 2006 Data 2006 Projection 120 Data (based on EX) Data (based on IM) Price index with 2000 value 115 Projection ECA 110 fixed to 100 NAM 105 LAC 100 EAP 95 CHN 90 JAP 85 2000 2001 2002 2003 2004 2005 2006 2007 2008 SEA Sources: FishStat and IMPACT model projections. SAR Note: EX = export data; IM = import data. IND prices are defined. Two fish price series are constructed based on MNA trade data. Construction of the first series is based on export data AFR and follows the procedure described in section 2.3, where the world ROW prices of traded species in the model are determined as weighted 00 00 00 00 00 0 0 0 0 0 0 00 00 00 00 00 averages of unit values faced by “dominant” exporters. The second ,0 ,0 ,0 ,0 ,0 1, 2, 3, 4, 5, –5 –4 –3 –2 –1 fish price series is based on import data such that the world price Thousand tons Sources: FAO FIPS FBS and IMPACT model projections. of each species is constructed as weighted average of import Note: ECA = Europe and Central Asia; NAM = North America; LAC = Latin America and Caribbean; CHN = China; JAP = Japan; EAP = other East Asia and the Pacific; unit value (from FishStat) by EU-15, Japan, and the United States. SEA = Southeast Asia; IND = India; SAR = other South Asia; MNA = Middle East and North Africa; AFR = Sub-Saharan Africa; ROW = rest of the world. Construction of the latter series follows the FAO Fish Price Index, detailed in Tveterås and others (2012). The third series in the figure represents the world prices projected by the model. For each of the In figure 2.14, we see a comparison between the model projections three series, world prices of different species are aggregated into and data for regional net exports of fish. The fit between the projec- a single fish price by weighting them according their net export tions and data for 2006 is fairly close for ECA and North America volumes generated within the model. For ease of presentation, the (NAM), but farther apart in LAC and Asian regions (EAP, CHN, JAP, three series are presented in the form of indices, where the price SEA, SAR, and IND). The deviation results from the fact that the levels for 2000 are scaled to 100. FAO datasets do not enforce the same balancing of trade across all species and regions as done in the IMPACT model. Following the The two indices based on fish trade data indicate that the aggregate data priority rule presented earlier, trade projections are allowed to fish price fluctuated, but it rose overall some 3 to 4 percent between deviate from the levels indicated by the data to a larger extent than 2000 and 2008. On the other hand, the index based on projected production. Furthermore, larger deviations are allowed for trade price series indicates a steady increase for a total of 17 percent than for consumption series. In all cases but for EAP, however, the during the same period. The model appears to overestimate the direction of trade—that is, whether a region is an overall net fish aggregate fish price. In fact, this originates from overestimation of exporter or net importer—agrees with the data. prices for the shrimp, crustaceans, freshwater and diadromous, and demersals categories (individual results not shown in the figure). As Price Projections discussed in section 1.2, the IMPACT model seems to have structural We now turn to the comparison of model projections of world limitations in representing the dynamics of world fish prices. The prices with the available data. In this study all prices are presented Fish to 2020 study also overestimated fish prices relative to observed in real terms (price levels in constant 2000 U.S. dollars). Figure 2.15 data. Throughout the rest of the study, therefore, price projections shows aggregate prices of all nine fish commodities for which world are interpreted with caution. F I S H TO 2 030: P R O S P E C T S F O R F I S H E R I E S A N D A Q UA C U LT U R E C H A P T E R 2 — P R E PA R I N G I M PA C T M O D E L F O R F I S H TO 2030 29 FIGURE 2.16: Comparison of Projections and Data for World FIGURE 2.17: Comparison of Projections and Data for World Prices of Fishmeal, 1984–2011 Prices of Fish Oil, 1984–2011 1,400 1,400 Data - IFFO Peru Data - FAO Data - IFFO Rotterdam Data - IFFO Hamburg Projection Data - FAO 1,200 1,200 Projection 1,000 1,000 800 800 US$ / ton US$ / ton 600 600 400 400 200 200 – – 84 86 88 90 92 94 96 98 00 02 04 06 08 10 84 86 88 90 92 94 96 98 00 02 04 06 08 10 19 19 19 19 19 19 19 19 20 20 20 20 20 20 19 19 19 19 19 19 19 19 20 20 20 20 20 20 Sources: Compilation of data from FishStat, Oil World, and the IFFO and IMPACT model Sources: Compilation of data from FishStat, Oil World, and the IFFO and IMPACT model projections. projections. A comparison between projections and data for fishmeal world that identify specific links between origins and destinations of trade. price shows a fairly close alignment, as seen in figure 2.16. While The FAO does not compile such data at the level of species detail the real prices shown in the FAO and IFFO data exhibit steeper rise needed for this study, and modeling bilateral trade would depend in the second half of 2000s, the projection reproduces the trend. on the quality of data and the structure of model. Thus, by allowing international trade to be the mechanism that reconciles supply and The steadily increasing trend in real world prices of fish oil is similar demand over time, this study focuses on capturing the overall driv- to that of fishmeal and is shown in figure 2.17. Again, the model ers of supply and demand growth. underpredicts the steep rise of fish oil price after 2005. Accordingly, the structure of world markets in IMPACT is simple and Conclusions of Calibration Exercise assumes a single market-clearing price for each good, across all re- As is the case with most global modeling work, a value that IMPACT gions. This, together with the absence of bilateral trade flows and the brings to the study of global fish supply and demand is an internally assumed homogeneity of quality in the goods coming from differ- consistent framework for analyzing and organizing the underlying ent regions, makes for only a crude approximation to how prices may data, which is drawn from disparate and often inconsistent sources. actually be formed on the basis of imperfect substitution between As in many food and agriculture studies, the detailed level of un- domestic and imported goods at the country-level—that is, the clas- derstanding of what happens on the supply (production) side is sic Armington assumption (Armington 1969), or the fact that market often not matched on the demand side for a given country—that concentration and imperfect competition might cause for a more is, consistent links between supply, demand (across the various complex process of price formation. There are classes of general and utilization categories), and international trade are not typically un- partial equilibrium multimarket models that handle these issues derstood. Although IMPACT does represent the net flows of trade better than IMPACT, such as the computable general equilibrium into or out of each country/region, it does not model bilateral flows models developed under the framework of the Global Trade Analysis A G R I C U LT U R E A N D R U R A L D E V E LO P M E N T D I S C U S S I O N PA P E R 30 C H A P T E R 2 — P R E PA R I N G I M PA C T M O D E L F O R F I S H TO 2030 Project (GTAP) modeling consortium (Hertel 1997); the World Bank’s elasticities and growth rates of the model are solved simultaneously Linkage model (van der Mensbrugghe 2005); or partial equilibrium with the market equilibrium calculations and projections, such models with spatial trade, such as the Global Biosphere Management that a closest alignment with observed data (for historical data) is model (Havlík and others 2011) of International Institute for Applied achieved. This is computationally intractable except for a highly sim- Systems Analysis (IIASA) or the regional Common Agricultural Policy plified system and would not guarantee that projections into the Regionalised Impact Modelling System (CAPRI model) (Britz 2005). future would be consistent with expert judgment. Some regional However, none of them can handle the detail of fish species, their models—such as the CAPRI model (Britz 2005) that is applied to the feed requirements, and their links with the rest of the agricultural European Union region—adjust elasticities estimated from the liter- sector in the way we have incorporated into the model. We felt that ature such that they are consistent with the theoretical constraints this was a more important feature of long-term growth that should implied by the functional forms of the model and with the other be captured in our analysis, and accepted some sacrifices on the parameters of the model (Jansson and Kempen 2006). Because the detail brought to market structure and price links. CAPRI model is used for static policy analysis, however, it does not have to deal with the issue of how to calibrate to observed market In this study elasticities and exogenous supply shifters were ad- changes over time, as is the case with those models applied to me- justed to obtain a good alignment with observed data (or to con- dium- and long-term market projections. form to expert opinion) in the projections. This is part of the normal process of calibration that many medium- and long-term projection A key issue of how models such as IMPACT can best reflect ob- models do—for example, OECD-FAO’s AgLINK-CoSiMo or the U.S. served past (or expected future) changes in market conditions Department of Agriculture’s Partial Equilibrium Agricultural Trade depends on the degree to which a standard market equilibrium Simulation (PEATSim) model (Somwaru and Dirkse 2012)—neces- modeling structure can endogenously adapt itself to changing sitated by the disparate sources of demand and supply elasticities micro-level market conditions and whether it can be applied to in the literature, as elasticities are usually estimated outside of a commodities that are highly dynamic in their growth and market market equilibrium framework. As observed by Blanco-Fonseca development, as opposed to more “mature” markets—such as for (2010) for the OECD-FAO and USDA agricultural baseline pro- grains, livestock, and other commodities that are relatively well cesses, the results of the first runs of the models are subjected to developed and have stable market structures—for which consid- an extensive review process, in which considerable expert judg- erable data exists on observed, past behavior. IMPACT, like other ment is reflected in model adjustments, until a consensus on the models that rely on the solution of a multicommodity, multiregion final set of projections is reached.20 The revisions of the IMPACT market equilibrium problem, has to predefine which regions and projections on fish within the project team reflected a similar ap- commodities will be analyzed throughout the projection period, proach, although with a much smaller group of experts and much and define a starting point for consumption, production, and trade simpler iteration processes involved. patterns, from which the projected market outcomes will evolve. An alternative to model calibration through parameter adjustments Gradual changes can be introduced into the model structure to is to do a structural estimation of model parameters, whereby the account for growth in population and income or gradual improve- ments in production technology, which are achieved through sequential shifts in the intercepts of demand and supply curves. 20 In the case of the OECD-FAO process, there are a series of annual ques- Gradual changes in consumption preferences can also be captured tionnaires that are sent to participating countries and are translated into the AgLink database by country experts within the OECD secretariat. through changes in the marginal expenditure propensities (that These are then combined with projections from the country modules managed by the FAO within the CoSiMo model such that a common is, elasticities) according to patterns that have been observed to baseline can be produced. The results are further reviewed by staff at occur in consumption of well-known commodities. More dramatic both institutions as well as by country experts at the OECD commodity working group (Adenäuer 2008). market transformations, however, present a challenge to market F I S H TO 2 030: P R O S P E C T S F O R F I S H E R I E S A N D A Q UA C U LT U R E C H A P T E R 2 — P R E PA R I N G I M PA C T M O D E L F O R F I S H TO 2030 31 equilibrium models, given that they may entail the introduction market pressures will result in an overall upward or a downward (or disappearance) of production or consumption of commodities trend in prices. in regions for which the model was initialized, and entails the in- troduction (or removal) of equations and variables from the model 2.6. ISSUES AND DISCUSSIONS structure in the middle of the simulation horizon. One of the well- In this section, issues in the data and methodology are summarized. known properties of the Armington (1969) approach to modeling This study relies on data compiled and made available by different trade is that it is not possible to introduce new trade flows into groups within the FAO. While these data are widely used in fisher- the model solution that did not exist at the initialization of the ies analyses, it is known that availability and quality of data are the simulations (Plassmann 2004, Jansson and Kempen 2006). This major constraint to any modeling exercise of global fish supply or presents a problem when dealing with fast-developing markets demand. The following list summarizes the data issues encountered for fish species such as tilapia, which saw a tenfold increase in ex- in the present modeling exercise: ports over 10 years and which we observed to change very rapidly § No single source of data or database exists for fish produc- even over the first five to six years of the IMPACT model calibration tion, consumption, and trade (import and export) for coun- period. Enormous growth in tilapia production has been observed tries/regions represented in the model. in mainland China (dwarfing Taiwan, China), and new producers, § Since data are drawn from disparate sources, for a given such as Ecuador, have emerged. Even though we prepared the country, the data on fish production, the consumption and model structure such that it could allow for this rapid growth dur- trade of fish, and the production and trade of fishmeal and fish oil are typically inconsistent. This is a common challenge ing the calibration period, it is not possible for the model to create that also arises when modeling other agricultural markets new growth where it does not already exist for any period after and commodities besides fish. that initial calibration window. This is a problem that is inherent § Fish production data are available in much more detail in in the current approaches to multicommodity, multiregional mar- terms of species disaggregation than fish consumption or ket equilibrium models, and will continue to present a challenge trade data, so detailed information on fish production is lost when trying to model fish markets at the level of species detail in the process of species aggregation. that we have done. An obvious solution to this problem would be § FAO data on bilateral fish trade are unavailable at the to adopt a higher level of commodity aggregation, as is done for preferred species disaggregation level or in the unit used in study (live weight equivalent). the OECD-FAO analysis of aquaculture. But this would come at the § Trade volume and value are well documented, but the expense of being unable to undertake a detailed analysis at a more correctness of conversion factors for processed fish into live disaggregate species level, which we see as a key advancement of weight equivalent is uncertain (for example, whole crab our approach. versus crab meat). § Trade data do not capture exports of processed products In general, we have focused more on the quantities of supply and based on imported raw material. demand, compared to the prices, in obtaining a close fit between § Consumption data used in this study are based on the projections and available data. Given that prices observed in the difference between production, non-food uses, and trade, data are the result of interactions and processes that occur within and thus the quality of consumption data depends on the quality of each of the original components. a much more complicated value chain than the one captured in § Production, apparent consumption (use), and trade data for this simplified model, we are not able to obtain as close a fit for fishmeal and fish oil are even less reliable and they are incon- price data as for quantity data. As a result, more confidence is given sistent with production data for small pelagic and other fish to the quantity projections that come from the model than to the from which fishmeal and fish oil are produced. price projections. In this study, only qualitative interpretations are § No data exist on the amount of fish (in whole or chopped) provided for price projections, as an indication of whether future directly used as feed in fish farming. A G R I C U LT U R E A N D R U R A L D E V E LO P M E N T D I S C U S S I O N PA P E R 32 C H A P T E R 2 — P R E PA R I N G I M PA C T M O D E L F O R F I S H TO 2030 Further, even for production series, which are considered more reli- § IMPACT modeling is based on historical data; it predicts able than consumption or trade data series, the following issues are changes in the production and consumption of various well known: commodities in countries that are already producing or con- suming the specific commodities. In other words, the model § For capture fisheries, overall catch levels are underestimated does not predict emergence of new participants—produc- due to unreliable data on bycatch, discards, and illegal, unre- ers or consumers—in the market. This poses a limitation to ported, and unregulated (IUU) fishing. the model’s applicability in highly dynamic markets, such § Catch data are organized by flag states of fishing vessels, and as those for fisheries and especially aquaculture, where they do not necessarily reflect the catch levels in waters of continual technological innovations allow farming of new each coastal states. species and create new market opportunities. § Given the focus of the IMPACT model on agricultural Therefore, any sophisticated model of the global seafood market markets, several small island states and regions such as is constrained in its construction by data availability; one must in- Greenland and Iceland are grouped together in a “rest of the terpret model results with caution, with these constraints and data world” (ROW) model region, as these areas are neither major quality issues in mind. Availability and quality of data dictate quality producers nor major consumers of agricultural products. of conclusions and decisions made based on them. IMPACT is no Some countries in the ROW group are, however, key actors in exception. Much of the modeling effort in the present Fish to 2030 fisheries and aquaculture. Nevertheless, the model structure does not allow for changing of the makeup of the model study focused on organizing available data from multiple sources regions for different commodities. and reconciling across them in order to obtain a consistent picture of the global seafood market. It became clear to us that in order for Further, several issues with model parameter specifications can be these predictive models to generate definitive conclusions, higher- summarized as follows: quality data are necessary. Investment in fisheries and aquaculture § With 115 model regions (individual and grouped countries) data—from collection to compilation across various stages and represented in IMPACT, the sheer size of the model pre- aspects of the market—should be promoted. Collaboration with vented us from examining every input parameter and every private sector players in the seafood value chain may be an effective output variable for each country. Use of country-specific direction for improving data collection and compilation. information from existing country studies and microdata would improve the quality of model output. Besides data issues, application of IMPACT model to the global fish- § For the most part, model parameters are drawn from the eries and aquaculture sector involved several challenges that lead relevant literature and existing data. However, future trends of to rigidity and limitation in the capacity of the model: parameters, which need to be specified by the researcher, are § IMPACT does not model bilateral trade flows, so specific links an additional source of error. A set of most important trend between origins and destinations of trade cannot be ana- parameters are exogenous growth parameters for aquaculture, lyzed. This also implies that IMPACT assumes homogenous which reflect anticipated technological change in aquaculture, quality of imported and exported commodities (that is, they especially in feed efficiency. Rigorous analysis to predict the are treated as perfect substitutes). extent and the timing of technological innovations for various aquaculture fish species is beyond the scope of this study. § Instead, the structure of world markets in IMPACT is simple and assumes a single market-clearing price for each good Given these observations, the model does not precisely replicate across all regions. This leads to simplified representation the realized outcomes of the global fish markets represented in of price formation, which in reality may be influenced by multiple (and inconsistent) series of data. Nonetheless, substantial a range of factors, including commodity stock/inventory, price expectations, and imperfections in market structure efforts were made to fine-tune the model to make its projections such as market concentration and the power exerted by as close as possible to the observed data. In doing so, we focused large firms. on the data series to which we give more confidence (for example, F I S H TO 2 030: P R O S P E C T S F O R F I S H E R I E S A N D A Q UA C U LT U R E C H A P T E R 2 — P R E PA R I N G I M PA C T M O D E L F O R F I S H TO 2030 33 capture harvest and aquaculture production series) rather than fish consumption data mainly relies on the quality of the those for which the accuracy is known to be limited (for example, original data from which they derive, as discussed earlier, consumption series). We also focused our calibration efforts on perfect adherence of model output to consumption data is infeasible. countries and regions that are important fish producers, and some § Finally, we place least confidence on model output for important aquaculture species (for example, salmon, shrimp, tilapia, world fish prices. In general, the dynamics of the global fish and Pangasius/catfish) received special attention. and seafood market is such that the price of major farmed species has declined and fish has become more affordable. The following is a summary observation on the general quality of However, due to the overly simplified representation of the model output. international trade markets in IMPACT, the model cannot § The model closely replicates aggregate fish supply trends at replicate such fish price trends. the global level for 2000–08. Nevertheless, the study represents a careful examination of § Confidence on model results declines as one examines fish data and parameters and rigorous modeling of fish supply and supply results at the country level,21 especially for smaller countries. However, when the results are aggregated at the demand. To our knowledge, the present model is the most com- regional level, the fit between model output and data is fairly prehensive economic model of global fish and seafood market in close. terms of the treatment of fish species, countries and regions, and § For supply of small pelagic fish in Latin America, there is activities related to fisheries and aquaculture production, utiliza- a gap between model projections and corresponding tion, and trade. data. These are major input used for fishmeal production. The deviation occurs because fishmeal requirements are calculated in the model for each aquaculture species based TECHNICAL APPENDIX on the coefficients found in the literature and the calculated A. Definition of Major Market Players for Deriving World Prices global fishmeal requirements are larger than what the data indicate on global fishmeal use. As a result, model forecasts The groups of major market players are selected according to a more small fish to be destined for fishmeal production than combination of approaches. the data indicate, and parameters are adjusted so that small § Largest exporters indicated in FishStat for the 1999–2001 pelagic fish production in Latin America picks up much of time period the difference. However, the harvest of small pelagic fish is § Expert opinion subject to huge variations because of El Niño and La Niña § For each commodity, the combined exports of the selected as well as decadal oscillations, and calibration to data from countries must add up to at least 75 percent of total world particular years is probably not appropriate. exports in the year 2000. § Even at the regional level, model results for per capita fish consumption, and accordingly total fish consumption, Note that silver, bighead, and grass carp are considered non-traded exhibit deviation from available data in the 2000s. The de- goods as the vast majority is consumed domestically in China and viation for per capita consumption is largest in Japan, East no sizable international market exists. Therefore, no world price is Asia and the Pacific, Southeast Asia, and ROW (a group of small countries that are not categorized in any region in calculated for this commodity. Table 2.7 lists the major market play- IMPACT). The deviation for total consumption (per capita ers for individual species and their combined share in the global consumption times the population) is largest in China market. and Southeast Asia. Note that, however, as the accuracy of B. Adding-up Conditions for the IMPACT Model 21 There are other multicommodity fish sector models (for example, the The complete set of adding-up conditions for the IMPACT Model is AsiaFish model) that work better for country-specific fish supply and de- mand projections (Dey, Briones, and Ahmed 2005; Dey and others 2008). as follows: A G R I C U LT U R E A N D R U R A L D E V E LO P M E N T D I S C U S S I O N PA P E R 34 C H A P T E R 2 — P R E PA R I N G I M PA C T M O D E L F O R F I S H TO 2030 TABLE 2.7: List of Countries Used to Define World Prices for Base Year PRODUCTION COUNTRY COMBINED SHARE IN CATEGORY (MAJOR MARKET PLAYER) GLOBAL TRADE Shrimp Thailand, China, Denmark, India, Indonesia, Netherlands, Norway, Vietnam, Greenland, Canada, Iceland, Malaysia, Mexico, United 75% Kingdom Crustaceans China, Canada, Thailand, United States, Indonesia, Russian Federation, United Kingdom, Myanmar, Vietnam, Australia, India, Denmark, 76% Mexico, Ireland, Republic of Korea Mollusks China, Republic of Korea, Argentina, Spain, Thailand, New Zealand, Morocco, United States, Vietnam, Netherlands, Denmark, United 75% Kingdom, Taiwan Province of China, Canada, Italy Salmon Norway, Chile, Denmark, United States, Canada 77% Tuna Thailand, Taiwan Province of China, Spain, France, Indonesia, Philippines, Ecuador, Côte d’Ivoire, Republic of Korea, Colombia, Seychelles 76% Tilapia Taiwan Province of China, Honduras, United States 99% Pangasius and other catfish Taiwan Province of China, United States 100% Carp Czech Republic, Taiwan Province of China, China, Belgium, Hungary 78% Other carp n.a. n.a. Eel and sturgeon China, Taiwan Province of China 89% Other freshwater and Tanzania, Indonesia, Uganda, Kenya, Belgium, Canada, Netherlands 75% diadromous Major demersals Norway, Russian Federation, United States, Iceland, Denmark, Germany, Netherlands, New Zealand, Canada, Spain, Argentina, Republic 77% of Korea, United Kingdom Mullet United States, New Zealand, Taiwan Province of China 100% Cobia and swordfish Taiwan Province of China, Spain, Portugal 89% Other pelagic Norway, Russian Federation, United Kingdom, Netherlands, Chile, Denmark, Namibia, Spain, United States, Sweden, Germany, Morocco, 78% Ireland, Latvia, Thailand, Poland, Canada Other marine China, Thailand, Russian Federation, India, Indonesia, Argentina, Sweden, Namibia, Vietnam, United States, Chile, Ecuador, China, Hong 78% Kong SAR, Japan, Ukraine, Senegal, Denmark, South Africa, Pakistan, Canada, Myanmar, Spain Source: Own calculations based on FishStat. Note: n.a. = not applicable.  All irrigated and rain-fed crop areas must add up to the total  Quantities of meal and oil production must be consistent crop harvested area reflected in FAO data. Where these areas with the crush/reduction demand from the feedstock are disaggregated to subnational spatial definitions, these commodities (either fish or oil-bearing crops) such that oil meal must all add up to the national FAO totals. QProdn RR oil meal crush QDemd , where the unit conversion of feed-  In the case of fish, all aquaculture and capture production stock to oil or meal is given by the parameter RR oil , meal . (inland/freshwater and marine) must add up to the total fish  The feed relationships must achieve a balance between production quantities indicated in FAO data. the animals (including fish) fed, and the quantities of  Area times yield must equal the FAO production levels. feed demand that are reflected in the data such that  All types of demand (food, feed, crush/reduction, bio- feed QDemd ∑ anim anim FCR animQProdn , where FCR anim is the amount of fuels, and other) must add up to FAO total demand feed that is required per unit of animal production. Q Total Demd =Q food Demd Q feed Demd Q crush Demd Q biofuel Demd +Qother Demd m .  Within each country, the relationship C. Details of Establishing the Consistent Base-Year Picture Qprodn QDemd + QNetExport ΔS must hold, where Maximum Entropy Estimation Qprodn QDemd QNetExport are the quantities of production, demand, and net export, respectively, and represents the An optimization approach is adopted in the program to obtain the amount being added to stocks. consistent base-year picture. The process involves estimation of  Across all countries, for a given commodity, the relationship two sets of parameters, reduction ratios (RRs) and feed conversion ∑ QNetExport reg reg = 0 must hold for all 115 model regions, such ratios (FCRs), as well as imputing associated variables (production that trade is balanced globally. and use of feed). This is done by fixing some variables at the levels F I S H TO 2030: P R O S P E C T S F O R F I S H E R I E S A N D A Q UA C U LT U R E C H A P T E R 2 — P R E PA R I N G I M PA C T M O D E L F O R F I S H TO 2 030 35 indicated by the data and penalizing deviations from the data for country level, so that the market equilibrium for the base year can the variables that are allowed to deviate. The objective criterion be reproduced in a way that is consistent within the structure of the is to minimize the “distance” between the “target values” and the market model. solution values of the parameters (RRs and FCRs) that satisfy the constraints. In this case, the constraints are the adding-up con- A least-squares type of fitting approach can be used in problems ditions that are employed in the IMPACT model (see technical like this in order to minimize distance while satisfying certain con- appendix B to this chapter). To put into a simple mathematical straints. However, there is often the issue of how to satisfy several expression, the essential problem that we are trying to solve is different targets simultaneously, without imposing undue weight the following: on any particular objective criterion over another. The problem becomes particularly vexing when confronted with a relative scar- C ∑ i ∑ k RRik − RR ik + ∑ i ∑ j ∑ k FCRijk 2000 minRR ,FCR FCR ijk 2000 city of reliable data, which may result in a far fewer number of data points than the number of unknowns that need to be determined. subject to the adding-up conditions, including This situation leads to the type of “ill-posed” problems, where the prodn feedik 2000 = RRik ∑ (Q j redctn jk 2000 + Q waste) jk 2000 and degrees of freedom are non-positive, and the problems cannot be solved by linear algebraic inversion of a data matrix with respect to ∑ k 2000 = ∑ j ∑ k Q jk 2000 prodn feedik aquaprodn FCRijk 2000 , a vector of variables. where In order to resolve this issue, cross-entropy-based techniques can  RRik denotes RR for feed item i in country k, RR ik its target be used, which can derive unknown distributions from fairly lim- value ited data and includes one’s own “prior” beliefs on the underlying  FCRijkt FCR for feed item i, for species j, in country k, and in nature of the distribution where possible (Kullback 1959, Kullback year t, FCR ijkt its target value prodn and Leibler 1951). Cross-entropy methods have been successfully  feedikt the total feed produced of item i in country k in used in many types of statistical analyses in the physical and social year t sciences and also have been used in IFPRI’s work. Examples include  Q redctn jkt the reduction demand for species j in country k in year t balancing of the social accounting matrix (SAM) of a computable waste  Q jkt the demand for fish processing waste from species general equilibrium model (Robinson, Cattaneo, and El-Said 2000) j to be used in fishmeal/fish oil production in country k in and calculation of the distribution of irrigated and rain-fed crops year t based on global data from a variety of (sometimes inconsistent) aquaprodn  Q jkt the aquaculture production of species j in country datasets (You and Wood 2004). The overall principle used to carry k in year t. out the estimation here is that of maximum entropy (Shannon 1948a, 1948b). It uses an optimization-based approach to measuring Note that RRs do not vary across fish species or over time (see next). fit based on a metric of “distance,” or deviation, which is grounded Note also that data for Q waste jk 2000 are unavailable and this variable is in the formalism of entropy-based econometrics and statistical imputed internally in the process (see next). methods. The optimization program allows a difference between the parame- ter values that satisfy the constraints and the target values, but it puts Assumptions used in the maximum entropy estimation of the two a penalty on this deviation according to a measure of distance that sets of parameters are detailed in the following two subsections. In is conveyed by the functions RRik − RR ik and FCRijkt FCR ijkt . particular, the RRs are discussed in the context of imputation of the The program also imposes penalties on deviation between the val- use of fish processing waste in fishmeal and fish oil production. The ues of certain variables and corresponding data in the process of FCR estimation discussion also includes the procedure of estimating enforcing certain necessary balances (such as that of trade) at the growth rates of FCRs. A G R I C U LT U R E A N D R U R A L D E V E LO P M E N T D I S C U S S I O N PA P E R 36 C H A P T E R 2 — P R E PA R I N G I M PA C T M O D E L F O R F I S H TO 2030 Imputing the Use of Fish Processing Waste in Fishmeal  Amount of whole fish used in fishmeal production indicated and Fish Oil Production by FAO FIPS FBS data, The amount of fish processing waste used in the production of fish-  Country-specific reduction ratios, and meal and fish oil produced is imputed for 2000 using the following  Amount of fishmeal production indicated by the FAO two sets of parameters. data, it was decided that additional countries must have used fish processing waste. These additional countries are  Reduction ratio: amount of fishmeal and fish oil produced Argentina, Australia, Belgium-Luxembourg, Brazil, the British (in product weight) per unit of fish used (in live weight Isles (including Ireland), Côte d’Ivoire, France, Germany, Italy, equivalent) Spain/Portugal, Uruguay, and Vietnam. Thus, in the imputed  Waste ratio: amount of fish processing waste generated from amount fish processing waste use originate from all of these a unit of whole fish (in live weight equivalent). countries. In this study, reduction ratios are defined for each country, rather The actual imputation of the volume of processing waste used in than each species. By this, we implicitly assume that the quantity fishmeal and fish oil production is conducted in conjunction with and quality of fishmeal and fish oil produced per unit of fish are the maximum entropy estimation of reduction ratios. As the “prior,” the same across species. FAO FIPS FBS specifies which species are or the target value RR ik in the estimation process, published figures reduced (as whole fish) to become fishmeal and fish oil. They are for Peru (0.23 for fishmeal and 0.06 for fish oil) are used (Jackson shrimp, crustaceans, mollusks, freshwater and diadromous, demer- 2010). Table 2.8 shows the regional values of the estimated reduc- sals, pelagics, and other marine. Waste ratios are specified for each tion ratios. species (see below). The weight of processing waste generated depends on the species, The imputation is further based on the following assumptions. processing stage, and the technology used. Table 2.9 compiles the First, that processing waste is not traded and that the amount of estimates of waste ratios from various sources. The estimates are processing waste available in a country is based only on the cap- based on the processing stage defined as “dressed head-off,” where 22 ture and aquaculture production that occurs within the country. Second, that every unit of whole fish can generate an amount of fish TABLE 2.8: Estimated Reduction Ratios processing weight according to the waste ratios. Third, that each FISHMEAL FISH OIL unit of fish processing waste yields the same quantity of fishmeal Global average 0.23 0.05 ECA 0.18 0.06 and fish oil as a unit of whole fish according to the reduction ratios NAM 0.19 0.07 and that there is no quality difference. LAC 0.23 0.05 Since the IFFO estimates that fish processing waste currently con- EAP 0.30 0.06 tributes to 25 percent of the global fishmeal production (Shepherd CHN 0.28 0.01 JAP 0.26 0.04 2012), we try to make the imputed values as close as possible to SEA 0.29 0.02 this number. FAO data suggest there are 10 countries that cur- SAR 0.15 0.04 rently use fish processing waste in fishmeal production: Canada, IND 0.03 0.02 Chile, Denmark, Iceland, Japan, Mexico, Norway, Russian Federation, MNA 0.14 0.05 Thailand, and the United States. However, in order to obtain consis- AFR 0.32 0.06 tency between the ROW 0.22 0.07 Source: Data based on Jackson 2010. Note: These are regional weighted averages based on reduction demand in 2000. ECA = Europe and Central Asia; NAM = North America; LAC = Latin America and Caribbean; CHN = China; JAP = Japan; EAP = other East Asia and 22 These assumptions are made for simplicity. In reality, both fish waste the Pacific; SEA = Southeast Asia; IND = India; SAR = other South Asia; trade and fish trade for processing and reexport of intermediate and MNA = Middle East and North Africa; AFR = Sub-Saharan Africa; ROW = rest final products are observed (FAO 2012). of the world. F I S H TO 2030: P R O S P E C T S F O R F I S H E R I E S A N D A Q UA C U LT U R E C H A P T E R 2 — P R E PA R I N G I M PA C T M O D E L F O R F I S H TO 2 030 37 TABLE 2.9: Waste Ratios Used in the IMPACT Model As the target value FCR ijkt in the maximum entropy estimation of COMMODITY WASTE SPECIES REPRESENTED AND METHODOLOGY NOTES FCRs, we heavily rely on Tacon and Metian (2008), who compiled Shrimp 45% Average of two shrimp species for “Raw Headless” processing stage (Crapo, Paust, and Babbitt 2004) FCR estimates for various species and various countries from the Salmon 26% Average of five different salmon species (Crapo, Paust, and literature. Using those values, we determined minimum, maximum, Babbitt 2004) and typical values of FCR for each species. When corresponding Tilapia 50% Clement and Lovell 1994; Garduño-Lugo and others 2003 species are not found in Tacon and Metian (2008), authors’ judg- Pangasius/catfish 40% Argue, Liu, and Dunham 2003; Clement and Lovell 1994; Li and others 2001; Silva and Dean 2001 ments are used. These minimum, maximum, and typical FCR values Tuna 25% Crapo, Paust, and Babbitt 2004 are used as prior in the estimation. MDemersal 29% Average of flounder, sole, turbot, halibut, cod, and hake (Crapo, Paust, and Babbitt 2004) In order to estimate growth rates of FCRs, the maximum entropy Mullet 26% Average of wild and farmed trout (Crapo, Paust, and Babbitt estimation is repeated for the year 2009. The two sets of values are 2004) used to derive FCR growth rates for fishmeal and fish oil for each OFresh 26% Use the same value as commodity “mullet” (Crapo, Paust, and Babbitt 2004) species. In many cases, however, the direct calculation of growth rates generated unreasonable values and adjustments had to be the head, fins, skin, and viscera, among other parts, are removed made. No FCR growth rates for soybean meal are estimated this from the fish. time. In implementation of the growth rates in the IMPACT model, in order to prevent unrealistically low or high levels of FCRs, lower Estimating FCRs and upper bounds of FCR values are imposed. The minimum and In the maximum entropy program, FCRs for the year 2000 are es- maximum values used as prior information in estimation are used timated in each country for the species that are considered to use as the bounds. fishmeal, fish oil, and soybean meal as input. These species are Finally, using the FCR estimates, the levels of feed use are imputed shrimp, crustaceans, salmon, tilapia, Pangasius/catfish, carp, OCarp, such that: EelStg, MDemersal, mullet, CobSwf, OPelagic, and OMarine. That is, the model does not account for feeding in aquaculture of mollusks, use feedikt = ∑ j Q aquaprodn jkt FCR ijkt , tuna, or OFresh. While mollusks and OFresh are typically not directly use where feedikt denotes the total feed used of feed item i in country fed, tuna is fed with feed fish rather than processed meal. In order k in year t and FCR ijkt the estimated values of FCR for feed item i, for to account for the cost of feeding in determining tuna supply, fish- species j, in country k, and in year t. meal and fish oil prices are included in the supply function for tuna. Livestock FCRs (poultry and hogs) are also estimated in the same program. A G R I C U LT U R E A N D R U R A L D E V E LO P M E N T D I S C U S S I O N PA P E R C H A P T E R 3 — I M PA C T P R O J E C T I O N S TO 2 0 3 0 U N D E R T H E B A S E L I N E S P E C I F I C AT I O N 39 Chapter 3: IMPACT PROJECTIONS TO 2030 UNDER THE BASELINE SPECIFICATION In this chapter, we systematically present the IMPACT model output FIGURE 3.1: Global Fish Production: Data and Projections, under the baseline specifications. The baseline scenario reflects the 1984–2030 trends of the global fish markets that are currently observed in the Total (data) Capture (model) 200 Total (model) Aquaculture (data) aggregate statistics. On the consumption demand side, we incor- Capture (data) Aquaculture (model) 180 porate the trends in human population and income growth. On the 160 production side, incorporated in the model are trends in the cap- 140 ture fisheries harvest, aquaculture production, and feed use and ef- Million tons 120 ficiency in aquaculture. Specifications of some of these “exogenous” 100 trends will be modified in the next chapter. Thus, the results here 80 provide the benchmark of what the model projects given the un- 60 derlying drivers of change in global fish supply, demand, and trade. 40 20 3.1. PRODUCTION 0 19 4 19 7 90 19 3 96 20 9 20 2 20 5 20 8 20 1 20 4 20 7 20 0 20 3 20 6 29 8 0 0 0 1 1 1 2 2 2 8 9 9 19 19 19 Global Trend Sources: FishStat and IMPACT model projections. We begin our presentation of the baseline results with fish produc- tion at the global level. Figure 3.1 depicts the projected global fish supply to 2030 and how it is divided between capture and aquacul- ture production, together with their historical path. The projected FIGURE 3.2: Volume and Share of Capture and Aquaculture capture production remains fairly stable over the 2000–30 period, Production in Global Harvest as has been observed in the data. In contrast, the global aquacul- 2011 (Data) 2030 (Projection) ture projection maintains its steady rise from historical levels, reach- Capture Aquaculture Capture Aquaculture ing the point where it equals global capture production by 2030. Global fish supply is projected to rise to 187 million tons by 2030. These projections are consistent with projections by OECD-FAO to 2021 (OECD-FAO 2012). See technical appendix to this chapter for 63.6 90.4 93.6 93.2 further comparisons of our projections with those by OECD-FAO (2012). Total harvest Figure 3.2 shows the breakdown of global fish supply between cap- 154.0 million tons Total harvest ture and aquaculture production. While the share of capture fisher- 186.3 million tons ies is nearly 60 percent of global production in 2011, it is expected Sources: FishStat and IMPACT model projections. A G R I C U LT U R E A N D R U R A L D E V E LO P M E N T D I S C U S S I O N PA P E R 40 C H A P T E R 3 — I M PA C T P R O J E C T I O N S TO 2030 U N D E R T H E B A S E L I N E S P E C I F I C AT I O N to fall to exactly half by 2030, after growing only by 2.8 million tons. FIGURE 3.3: Average Annual Growth Rates of Capture and Aquaculture is expected to grow by 30 million tons over this same Aquaculture Production, 1960–2029 period. In terms of food fish production, the model predicts that Capture Aquaculture 12 aquaculture will contribute 62 percent of the global supply by 2030. 10 These results are consistent with the overview given in the 8 introductory chapter and reinforce the importance of aquacul- ture in augmenting global fish supply. However, the growth of 6 Percent aquaculture is expected to further decelerate. Figure 3.3 extends 4 figure 1.3 to include the projected annual growth rates for the projection periods of 2010–19 and 2020–29. For these two peri- 2 ods, the projected growth rate of aquaculture production is below 0 the level in the 1960s. Nearly zero growth is projected for capture 9 9 9 9 9 9 9 –6 –7 –8 –9 –0 –1 –2 –2 60 70 80 90 00 10 20 production. 19 19 19 19 20 20 20 Sources: FishStat and IMPACT model projections. Regional Distribution Geographically, fish production is concentrated in Asia, LAC, and during 2010–30, representing 6.8 percent of global production by ECA (table 3.1). In 2008, Asia (total of EAP, CHN, JAP, SEA, SAR, and 2030. SEA is expected to grow 37.5 percent, and it will likely repre- IND) represented 65 percent of global fish production, with CHN sent more than 15 percent of global production by 2030. China’s accounting for more than a third of global production. Global fish fish production is expected to grow 31.4 percent, accounting for production is expected to further concentrate in Asia toward 2030 an overwhelming 36.9 percent of the world’s fish production by (69 percent). IND has the largest projected growth, 60.4 percent, 2030. China represented the largest and one of the fastest-growing TABLE 3.1: Projected Total Fish Production by Region DATA (000 TONS) PROJECTION (000 TONS) SHARE IN GLOBAL TOTAL % CHANGE 2010 2030 2008 2010 2020 2030 (PROJECTION) (PROJECTION) 2010–30 Global total 142,285 151,129 172,035 186,842 100.0% 100.0% 23.6% ECA 14,564 14,954 15,369 15,796 9.9% 8.5% 5.6% NAM 6,064 6,226 6,319 6,472 4.1% 3.5% 3.9% LAC 17,427 19,743 20,957 21,829 13.1% 11.7% 10.6% EAP 3,724 3,698 3,832 3,956 2.4% 2.1% 7.0% CHN 49,224 52,482 62,546 68,950 34.7% 36.9% 31.4% JAP 4,912 5,169 4,911 4,702 3.4% 2.5% –9.0% SEA 20,009 21,156 25,526 29,092 14.0% 15.6% 37.5% SAR 6,815 7,548 9,210 9,975 5.0% 5.3% 32.1% IND 7,589 7,940 10,346 12,731 5.3% 6.8% 60.4% MNA 3,518 3,832 4,440 4,680 2.5% 2.5% 22.1% AFR 5,654 5,682 5,865 5,936 3.8% 3.2% 4.5% ROW 2,786 2,696 2,714 2,724 1.8% 1.5% 1.0% Sources: FishStat and IMPACT model projections. Note: ECA = Europe and Central Asia; NAM = North America; LAC = Latin America and Caribbean; CHN = China; JAP = Japan; EAP = other East Asia and the Pacific; SEA = Southeast Asia; IND = India; SAR = other South Asia; MNA = Middle East and North Africa; AFR = Sub-Saharan Africa; ROW = rest of the world. F I S H TO 2 030: P R O S P E C T S F O R F I S H E R I E S A N D A Q UA C U LT U R E C H A P T E R 3 — I M PA C T P R O J E C T I O N S TO 2 0 3 0 U N D E R T H E B A S E L I N E S P E C I F I C AT I O N 41 TABLE 3.2: Projected Aquaculture Production by Region DATA (000 TONS) PROJECTION (000 TONS) SHARE IN GLOBAL TOTAL % CHANGE 2008 2010 2020 2030 2010 (PROJECTION) 2030 (PROJECTION) 2010–30 Global total 52,843 57,814 78,625 93,612 100.0% 100.0% 61.9% ECA 2,492 2,734 3,270 3,761 4.7% 4.0% 37.5% NAM 655 631 728 883 1.1% 0.9% 40.0% LAC 1,805 1,642 2,770 3,608 2.8% 3.9% 119.7% EAP 751 795 936 1,066 1.4% 1.1% 34.0% CHN 33,289 36,562 46,790 53,264 63.2% 56.9% 45.7% JAP 763 765 861 985 1.3% 1.1% 28.7% SEA 6,433 7,171 11,384 14,848 12.4% 15.9% 107.1% SAR 1,860 2,185 3,493 4,163 3.8% 4.4% 90.5% IND 3,585 3,885 6,232 8,588 6.7% 9.2% 121.1% MNA 921 1,086 1,679 1,911 1.9% 2.0% 75.9% AFR 231 302 418 464 0.5% 0.5% 53.6% ROW 57 55 64 72 0.1% 0.1% 29.5% Sources: FishStat and IMPACT model projections. Note: ECA = Europe and Central Asia; NAM = North America; LAC = Latin America and Caribbean; CHN = China; JAP = Japan; EAP = other East Asia and the Pacific; SEA = Southeast Asia; IND = India; SAR = other South Asia; MNA = Middle East and North Africa; AFR = Sub-Saharan Africa; ROW = rest of the world. countries in the Fish to 2020 assessment. Although other regions are 17 percent higher growth response per generation, availability of also expected to increase fish supply, their relative contribution to balanced supplementary feed for different life stages for diversified the global supply will likely decline. Japan’s fish production is pro- cultivable species, and appropriate disease management measures jected to contract during this period. (Ayyappan 2012). Considering just aquaculture production, China’s share in global Table 3.3 shows the regional distribution of fish production from production is even larger (table 3.2). In 2008, China represented 63.2 capture fisheries. In contrast to aquaculture production, the dis- percent of global aquaculture production, and the projected share tribution of capture production is more evenly spread out across in 2030 will decline to 56.9 percent. While all regions are expected regions. China, LAC, SEA, and ECA each had more than 10 percent to expand their aquaculture production, the largest expansion is of the share of global capture harvest in 2008. The largest growth expected in SEA and IND. SEA is expected to represent 15.9 percent in harvest is expected for SAR, while Japan is expected to reduce of global aquaculture production in 2030, while IND would repre- capture production by 15 percent over the 2010–30 period. sent 9.2 percent. LAC and South Asia (excluding India) (SAR) are also projected to experience large aquaculture growth over the 2010–30 By Species period. Middle East and North Africa (MNA) and Sub-Saharan Africa Figure 3.4 depicts the projected fish supply by species (both cap- (AFR) also show substantial expected growth over this period, but ture fisheries and aquaculture), while figure 3.5 shows the pro- they begin from much lower production levels in 2010 compared jected dynamics of aquaculture. Further, table 3.4 summarizes the to other regions. Given the recent aquaculture research and devel- projected change in species share over time. The fastest growth opment efforts in these countries, the projection results are quite is expected for tilapia, carp, and Pangasius/catfish. Due to the plausible. For example, aquaculture is considered a “sunrise sector” rapid expansion of aquaculture, production of tilapia is projected in India. In recent years, India has made significant achievement in to more than double between 2008 and 2030. Some high-value aquaculture research and development, including development of species (shrimp, salmon, and EelStg) are expected to grow by improved rohu carp through selective breeding with a record of 50 to 60 percent over the period. Some low-value species (carp A G R I C U LT U R E A N D R U R A L D E V E LO P M E N T D I S C U S S I O N PA P E R 42 C H A P T E R 3 — I M PA C T P R O J E C T I O N S TO 2030 U N D E R T H E B A S E L I N E S P E C I F I C AT I O N TABLE 3.3: Projected Capture Fisheries Production by Region DATA (000 TONS) PROJECTION (000 TONS) SHARE IN GLOBAL TOTAL % CHANGE 2008 2010 2020 2030 2010 (PROJECTION) 2030 (PROJECTION) 2010–30 Global total 89,443 93,315 93,410 93,229 100.0% 100.0% –0.1% ECA 12,072 12,220 12,099 12,035 13.1% 12.9% –1.5% NAM 5,409 5,596 5,591 5,589 6.0% 6.0% –0.1% LAC 15,621 18,101 18,187 18,221 19.4% 19.5% 0.7% EAP 2,973 2,903 2,896 2,890 3.1% 3.1% –0.4% CHN 15,935 15,920 15,756 15,686 17.1% 16.8% –1.5% JAP 4,149 4,403 4,050 3,717 4.7% 4.0% –15.6% SEA 13,575 13,986 14,142 14,244 15.0% 15.3% 1.8% SAR 4,955 5,363 5,717 5,811 5.7% 6.2% 8.4% IND 4,004 4,055 4,114 4,143 4.3% 4.4% 2.2% MNA 2,597 2,746 2,761 2,769 2.9% 3.0% 0.8% AFR 5,422 5,380 5,447 5,472 5.8% 5.9% 1.7% ROW 2,729 2,641 2,649 2,652 2.8% 2.8% 0.4% Sources: FishStat and IMPACT model projections. Note: ECA = Europe and Central Asia; NAM = North America; LAC = Latin America and Caribbean; CHN = China; JAP = Japan; EAP = other East Asia and the Pacific; SEA = Southeast Asia; IND = India; SAR = other South Asia; MNA = Middle East and North Africa; AFR = Sub-Saharan Africa; ROW = rest of the world. FIGURE 3.4: Projected Global Fish Supply by Species FIGURE 3.5: Projected Global Aquaculture Fish Supply by Species 2008 Data 2020 Projection 2030 Projection Shrimp 2008 Data 2020 Projection 2030 Projection Crustaceans Shrimp Mollusks Crustaceans Salmon Mollusks Tuna Salmon Tilapia Tuna Pangasius /catfish Tilapia Carp Pangasius /catfish OCarp Carp EelStg OCarp OFresh EelStg MDemersal OFresh Mullet MDemersal CobSwf Mullet OPelagic CobSwf OMarine OPelagic OMarine − 0 00 00 00 00 00 00 00 – 5,000 10,000 15,000 20,000 25,000 00 ,0 ,0 ,0 ,0 ,0 ,0 ,0 5, 10 15 20 25 30 35 40 Thousand tons Thousand tons Sources: FishStat and IMPACT model projections. Sources: FishStat and IMPACT model projections. Note: Pangasius/catfish = Pangasius and other catfish; OCarp = silver, bighead, Note: Pangasius/catfish = Pangasius and other catfish; OCarp = silver, bighead, and grass carp; EelStg = aggregate of eels and sturgeon; OFresh = freshwater and and grass carp; EelStg = aggregate of eels and sturgeon; OFresh = freshwater and diadromous species (excluding tilapia, Pangasius/catfish, carp, OCarp, and EelStg); diadromous species (excluding tilapia, Pangasius/catfish, carp, OCarp, and EelStg); MDemersal = major demersal fish; CobSwf = aggregate of cobia and swordfish; MDemersal = major demersal fish; CobSwf = aggregate of cobia and swordfish; OPelagic = other pelagic species; OMarine = other marine fish. OPelagic = other pelagic species; OMarine = other marine fish. and OCarp) also will likely grow fast. On the other hand, only the introduction of genetically improved tilapia in Asia in the marginal growth in supply is expected for species with limited 1990s, many countries around the globe, including China, cur- aquaculture potential (for example, OPelagic, MDemersal, and rently have ongoing tilapia genetic improvement programs (ADB tuna). Given the ongoing research and development efforts on 2005; Dey 2000; Eknath and others 2007; Gjedrem, Robinson, various aquaculture species, these results are quite realistic. Since and Rye 2012). Genetic improvement programs have also been F I S H TO 2 030: P R O S P E C T S F O R F I S H E R I E S A N D A Q UA C U LT U R E C H A P T E R 3 — I M PA C T P R O J E C T I O N S TO 2 0 3 0 U N D E R T H E B A S E L I N E S P E C I F I C AT I O N 43 TABLE 3.4: Projected Species Shares in Aquaculture Production TABLE 3.5: Projected Top Three Fish Producing Regions by Species 2008 2010 2020 2030 DATA PROJECTION PROJECTION PROJECTION   2008 – DATA 2030 – PROJECTION Shrimp 7% 7% 8% 9%   1ST 2ND 3RD 1ST 2ND 3RD (SHARE) (SHARE) (SHARE) (SHARE) (SHARE) (SHARE) Crustaceans 2% 2% 2% 2% Shrimp  CHN SEA LAC CHN SEA LAC Mollusks 26% 28% 26% 24% 43% 27% 10% 39% 36% 10% Salmon 4% 4% 4% 4% Crustaceans  CHN NAM ECA CHN NAM SEA Tuna .. .. .. .. 57% 11% 7% 63% 9% 7% Tilapia 5% 5% 7% 7% Mollusks  CHN SEA LAC CHN SEA LAC Pangasius/catfish 5% 5% 5% 5% 63% 7% 7% 69% 7% 5% Carp 19% 20% 20% 21% Salmon  ECA LAC NAM ECA LAC NAM OCarp 19% 18% 16% 16% 52% 19% 15% 49% 29% 11% EelStg 1% .. .. 1% Tuna  SEA LAC EAP SEA LAC EAP OFresh 7% 6% 6% 7% 23% 13% 12% 25% 14% 12% MDemersal 2% 2% 2% 2% Tilapia  CHN SEA AFR SEA CHN MNA Mullet .. 1% 1% 1% 35% 27% 14% 37% 29% 15% CobSwf 0.1% 0.1% .. .. Pangasius/ SEA CHN AFR SEA CHN AFR OPelagic 0.3% 0.3% 0.2% 0.2% catfish  50% 22% 9% 55% 19% 8% OMarine 1% 1% 1% 1% Carp  CHN IND SAR CHN IND SEA Note: Pangasius/catfish = Pangasius and other catfish; OCarp = silver, bighead, and grass carp; EelStg = aggregate of eels and sturgeon; OFresh = freshwater and 49% 27% 10% 37% 36% 13% diadromous species (excluding tilapia, Pangasius/catfish, carp, OCarp, and EelStg); MDemersal = major demersal fish; CobSwf = aggregate of cobia and swordfish; OCarp  CHN IND SAR CHN SAR IND OPelagic = other pelagic species; OMarine = other marine fish; .. = negligible. 93% 3% 2% 92% 3% 3% EelStg  CHN JAP ECA CHN JAP ECA 82% 7% 5% 89% 5% 3% initiated for carp and shrimp (Dey and others 2010; Hung and OFresh  CHN SAR SEA CHN SEA SAR others 2013; Ninh and others 2013). Though about 97 percent 32% 21% 17% 35% 20% 20% MDemersal  ECA CHN NAM ECA CHN NAM of the world salmon production is currently based on improved 26% 20% 11% 26% 20% 11% stock (Gjedrem and Baranski 2009), the salmon industry has been Mullet  MNA CHN SEA MNA CHN SEA maintaining a strong research and development effort. There 45% 13% 13% 57% 11% 11% has been improvement in technical efficiency over time in the CobSwf  CHN ECA JAP CHN ECA LAC Norwegian salmon industry, mainly through restructuring the 32% 24% 9% 33% 26% 8% industry as well as making improvements in government regula- OPelagic  LAC ECA SEA LAC ECA SEA tions (Asche and Roll 2013). 34% 15% 14% 39% 14% 13% OMarine  SEA CHN SAR CHN SEA SAR Table 3.5 shows the top three regions in the production of each 32% 26% 20% 29% 28% 21% species in 2008 and 2030. Overall, the model does not predict a sub- Sources: FishStat and IMPACT model projections. stantial shift in the major players in the global fish markets. SEA is Note: Pangasius/catfish = Pangasius and other catfish; OCarp = silver, bighead, and grass carp; EelStg = aggregate of eels and sturgeon; OFresh = freshwater expected to take some of China’s share in the global shrimp supply, and diadromous species (excluding tilapia, Pangasius/catfish, carp, OCarp, and EelStg); MDemersal = major demersal fish; CobSwf = aggregate of cobia and while LAC is likely to grow to account for a third of global salmon swordfish; OPelagic = other pelagic species; OMarine = other marine fish; ECA = Europe and Central Asia; NAM = North America; LAC = Latin America and supply by 2030. The latter primarily represents recovery after the ISA Caribbean; CHN = China; JAP = Japan; EAP = other East Asia and the Pacific; SEA = Southeast Asia; IND = India; SAR = other South Asia; MNA = Middle East outbreak and subsequent growth in Chile. and North Africa; AFR = Sub-Saharan Africa; ROW = rest of the world. A G R I C U LT U R E A N D R U R A L D E V E LO P M E N T D I S C U S S I O N PA P E R 44 C H A P T E R 3 — I M PA C T P R O J E C T I O N S TO 2030 U N D E R T H E B A S E L I N E S P E C I F I C AT I O N 3.2. CONSUMPTION 18.2 kilograms in 2030. The trend in per capita consumption, howev- In all of the simulations presented in this study, the drivers of change er, is diverse across regions. In general, per capita fish consumption on the demand side are specified according to the income and is expected to grow fast in the regions with the highest projected population growth trends as found in table 3.6. According to the income growth (CHN, IND, SEA). However, the highest growth in fish World Bank (2012), between 2010 and 2030, China’s gross domes- consumption is expected in SAR, where per capita fish consump- tic product (GDP) per capita is expected to almost triple. Income tion is expected to grow at 1.8 percent per year over the 2010–30 levels in IND and SEA are expected to almost double. On the other period. In all of these regions, however, the growth in per capita fish hand, the UN (2011) projects the highest population growth in AFR. consumption is expected to slow relative to the 2000–06 period. Between 2010 and 2030, the population in AFR is projected to in- Japan, traditionally the world’s largest consumer of seafood, is the crease by 57.6 percent, or at the annual rate of 2.3 percent. only region where per capita fish consumption declined over the Currently, about 80 percent of the fish produced globally is con- 2000–06 period (it declined from 67.7 kilograms to 59.2 kilograms). sumed by people as food. The model results suggest that this The model predicts a continued decline, but at a slower rate. proportion is not expected to change into 2030. Given that the A declining trend of fish consumption is also projected for EAP, LAC, production is expected to grow by 23.6 percent during the 2010–30 and AFR. period (table 3.1) and the world population is projected to grow at Per capita fish consumption is projected to decline in AFR. Starting 20.2 percent over the same period (table 3.6), the world will likely from a modest level of fish consumption in 2006—7.5 kilograms, manage to increase the fish consumption level, on average. which was the second lowest, after IND (5.0 kilograms)—per capita As seen in table 3.7, at the global level, annual per capita fish con- fish consumption in AFR is projected to decline to 5.6 kilograms by sumption is projected to increase from 17.2 kilograms in 2010 to 2030. TABLE 3.6: Income and Population Growth Assumptions GDP PER CAPITA POPULATION GDP/c % POPULATION % SHARE IN (US$) CHANGE (MILLIONS) CHANGE GLOBAL TOTAL 2010 2010 2010 2030 (DATA) 2010–30 (DATA) 2010–30 (DATA) (PROJECTION) Global total/average 6,941 17.4% 6,941 20.2% 100% 100% ECA 12,906 40.5% 891 3.3% 12.8% 11.0% NAM 36,764 25.2% 347 16.3% 5.0% 4.8% LAC 4,986 32.2% 586 18.5% 8.4% 8.3% EAP 13,724 48.6% 110 12.8% 1.6% 1.5% CHN 2,797 177.0% 1,355 3.4% 19.5% 16.8% JAP 40,092 22.4% 126 –5.4% 1.8% 1.4% SEA 1,875 88.4% 550 18.7% 7.9% 7.8% SAR 606 51.1% 460 28.8% 6.6% 7.1% IND 828 92.6% 1,241 23.7% 17.9% 18.4% MNA 3,380 29.0% 382 31.8% 5.5% 6.0% AFR 646 77.3% 874 57.6% 12.6% 16.5% ROW 7,103 79.0% 19 13.2% 0.3% 0.3% Sources: UN 2011; World Bank 2012. Note: ECA = Europe and Central Asia; NAM = North America; LAC = Latin America and Caribbean; CHN = China; JAP = Japan; EAP = other East Asia and the Pacific; SEA = Southeast Asia; IND = India; SAR = other South Asia; MNA = Middle East and North Africa; AFR = Sub-Saharan Africa; ROW = rest of the world. F I S H TO 2 030: P R O S P E C T S F O R F I S H E R I E S A N D A Q UA C U LT U R E C H A P T E R 3 — I M PA C T P R O J E C T I O N S TO 2 0 3 0 U N D E R T H E B A S E L I N E S P E C I F I C AT I O N 45 Projections on per capita fish consumption in table 3.7 combined population, global fish consumption is also heavily centered in with the population growth projections in table 3.6 determine the Asia. The Asian regions are also projected to have steady and rapid aggregate projected trends seen in table 3.8. As with the world consumption growth over the period, with IND and SAR expecting TABLE 3.7: Projected Per Capita Fish Consumption by Region DATA (KG/PERSON/YEAR) PROJECTION (KG/PERSON/YEAR) ANNUAL GROWTH RATE 2000 2006 2010 2020 2030 2000–06a 2010–30b Global average 15.7 16.8 17.2 18.0 18.2 1.1% 0.3% ECA 17.0 18.5 17.4 17.2 18.2 1.5% 0.2% NAM 21.8 24.3 22.9 24.5 26.4 1.8% 0.7% LAC 8.8 9.4 8.4 8.0 7.5 1.1% –0.6% EAP 32.1 36.5 27.1 26.1 23.8 2.2% –0.7% CHN 24.4 26.6 32.6 37.8 41.0 1.4% 1.2% JAP 67.7 59.2 64.7 63.7 62.2 –2.2% –0.2% SEA 24.6 27.9 25.8 28.3 29.6 2.1% 0.7% SAR 8.5 11.4 11.0 13.4 15.7 5.1% 1.8% IND 4.5 5.0 5.6 6.2 6.6 1.7% 0.8% MNA 8.3 10.2 9.3 9.4 9.4 3.5% 0.0% AFR 7.1 7.5 6.8 6.1 5.6 0.8% –1.0% ROW 18.4 20.1 9.4 9.6 9.6 1.5% 0.1% Sources: FAO FIPS FBS and IMPACT model projections. Note: ECA = Europe and Central Asia; NAM = North America; LAC = Latin America and Caribbean; CHN = China; JAP = Japan; EAP = other East Asia and the Pacific; SEA = Southeast Asia; IND = India; SAR = other South Asia; MNA = Middle East and North Africa; AFR = Sub-Saharan Africa; ROW = rest of the world. a Based on data. b Based on projections. TABLE 3.8: Projected Total Food Fish Consumption by Region DATA (000 TONS) PROJECTION (000 TONS) SHARE IN GLOBAL TOTAL % CHANGE 2006 2010 2020 2030 2010 (PROJECTION) 2030 (PROJECTION) 2010–30 Global total 111,697 119,480 138,124 151,771 100.0% 100.0% 27.0% ECA 16,290 15,488 15,720 16,735 13.0% 11.0% 8.1% NAM 8,151 7,966 9,223 10,674 6.7% 7.0% 34.0% LAC 5,246 4,900 5,165 5,200 4.1% 3.4% 6.1% EAP 3,866 2,975 3,068 2,943 2.5% 1.9% –1.1% CHN 35,291 44,094 52,867 57,361 36.9% 37.8% 30.1% JAP 7,485 8,180 7,926 7,447 6.8% 4.9% –9.0% SEA 14,623 14,175 17,160 19,327 11.9% 12.7% 36.3% SAR 4,940 5,063 7,140 9,331 4.2% 6.1% 84.3% IND 5,887 6,909 8,688 10,054 5.8% 6.6% 45.5% MNA 3,604 3,571 4,212 4,730 3.0% 3.1% 32.5% AFR 5,947 5,980 6,758 7,759 5.0% 5.1% 29.7% ROW 367 179 198 208 0.2% 0.1% 15.7% Sources: FAO FIPS FBS and IMPACT model projections. Note: ECA = Europe and Central Asia; NAM = North America; LAC = Latin America and Caribbean; CHN = China; JAP = Japan; EAP = other East Asia and the Pacific; SEA = Southeast Asia; IND = India; SAR = other South Asia; MNA = Middle East and North Africa; AFR = Sub-Saharan Africa; ROW = rest of the world. A G R I C U LT U R E A N D R U R A L D E V E LO P M E N T D I S C U S S I O N PA P E R 46 C H A P T E R 3 — I M PA C T P R O J E C T I O N S TO 2030 U N D E R T H E B A S E L I N E S P E C I F I C AT I O N the largest growth to 2030, though from lower initial levels. Adding are expected to increase substantially from the 2010 level to 2020 together all its regions (CHN, EAP, JAP, SEA, IND, and SAR), Asia is and then 2030 levels. This implies an increasing import dependency expected to represent 70 percent of global fish consumption by in AFR, and it might expose the region to greater variability in the 2030. MNA, which represented 3.2 percent of global fish consump- fish supply and vulnerability of their food security to shocks that oc- tion in 2006, is projected to grow by more than 30 percent during cur in the global markets. Given the pattern of fish consumption the 2010–30 period. and imports, the import to consumption ratio in AFR would rise from 14 percent in 2000 to nearly 34 percent in 2030. Even though per capita consumption is expected to decline in Sub- Saharan Africa, the total consumption for the region is expected to Looking across other regions, we project that strong net export grow by 30 percent over the projection horizon. This is much more trends of regions like SEA, LAC, CHN, and IND will be balanced out than the 4.5 percent growth in production projected over that by the strong net imports by other regions such as NAM, ECA, JAP, same period (table 3.1), and suggests that much of the increased AFR, and MNA. fish consumption would be supported by imports as seen in the To look further into the patterns of global fish trade, tables 3.10a next section. and 3.10b list the top three net exporter and net importer regions, respectively, of each traded species in 2006 (data) and 2030 (pro- 3.3. TRADE AND PRICES jection). While the net trade patterns remain the same for many Table 3.9 summarizes the model results on net exports for each re- cases, some new patterns are projected to emerge. LAC countries gion, where a positive number in the table indicates net exports and are projected to increase their share of shrimp net exports from a negative number indicates net imports (shaded cells in the table). The row for AFR suggests that the net imports of fish by this region TABLE 3.10a: Projected Top Three Net Fish Exporting Regions by Species 2006 – DATA 2030 – PROJECTION 1ST 2ND 3RD 1ST 2ND 3RD TABLE 3.9: Projected Net Exports of Fish by Region (SHARE) (SHARE) (SHARE) (SHARE) (SHARE) (SHARE) DATA (000 TONS) PROJECTION (000 TONS) % CHANGE Shrimp SEA CHN LAC SEA LAC CHN 2006 2010 2020 2030 2010–30 45% 17% 14% 55% 25% 7% Global total 12,258 12,677 14,652 17,756 40.1% Crustaceans CHN SEA LAC CHN SAR EAP trade volume 72% 17% 5% 89% 9% 2% ECA –4,166 –4,145 –3,994 –4,602 11.0% Mollusks CHN LAC SEA CHN LAC EAP NAM –2,405 –2,911 –4,121 –5,464 87.7% 58% 23% 15% 59% 29% 5% LAC 2,520 2,018 2,879 3,678 82.3% Salmon LAC ECA ROW LAC ECA ROW EAP –983 155 151 394 154.0% 85% 11% 4% 82% 15% 2% CHN 4,288 2,002 2,210 3,567 78.1% Tuna CHN SEA ROW ROW CHN EAP JAP –3,570 –4,239 –4,233 –3,953 –6.8% 35% 23% 21% 29% 27% 24% SEA 2,741 5,372 6,482 7,735 44.0% Freshwater and SEA CHN AFR SEA CHN IND SAR 362 2,097 1,614 150 –92.9% diadromous IND 596 623 1,220 2,232 258.1% 49% 35% 14% 79% 17% 4% MNA –560 –456 –675 –1,042 128.6% Demersals ROW LAC IND IND LAC SEA AFR –806 –927 –1,629 –2,633 184.2% 49% 31% 7% 26% 25% 23% ROW 1,518 410 95 –63 –115.3% Pelagics LAC NAM MNA SEA ECA EAP Sources: FAO FIPS FBS and IMPACT model projections. 34% 17% 17% 56% 16% 11% Note: ECA = Europe and Central Asia; NAM = North America; LAC = Latin America and Caribbean; CHN = China; JAP = Japan; EAP = other East Asia and the Pacific; Other marine CHN SEA SAR CHN IND ROW SEA = Southeast Asia; IND = India; SAR = other South Asia; MNA = Middle East 69% 20% 8% 70% 27% 3% and North Africa; AFR = Sub-Saharan Africa; ROW = rest of the world. F I S H TO 2 030: P R O S P E C T S F O R F I S H E R I E S A N D A Q UA C U LT U R E C H A P T E R 3 — I M PA C T P R O J E C T I O N S TO 2 0 3 0 U N D E R T H E B A S E L I N E S P E C I F I C AT I O N 47 TABLE 3.10b: Projected Top Three Net Fish Importing Regions FIGURE 3.6: Projected Change in Real Prices between 2010 by Species and 2030 by Commodities 2006 – DATA 2030 – PROJECTION 100% 1ST 2ND 3RD 1ST 2ND 3RD 90% (SHARE) (SHARE) (SHARE) (SHARE) (SHARE) (SHARE) 80% Shrimp NAM ECA JAP NAM ECA JAP 70% 46% 29% 16% 60% 21% 11% 60% 50% Crustaceans JAP NAM ECA JAP NAM ECA 40% 61% 20% 19% 45% 28% 17% 30% Mollusks ECA JAP NAM NAM ECA SEA 20% 43% 33% 18% 39% 30% 11% 10% 0% Salmon CHN JAP NAM CHN NAM JAP ta p s Sa s on ad una em us Pe ls s Fi ine Fi l l ea oi an k th agic rim sa s o lm sh ar m lu ce m T er 33% 30% 19% 55% 19% 18% Sh sh m ol l ro M er s ru D di Tuna ECA NAM JAP ECA NAM JAP C O d an 46% 24% 17% 42% 24% 17% er at hw Freshwater and ECA NAM JAP AFR ECA NAM es diadromous Fr 42% 41% 8% 50% 21% 13% Source: IMPACT model projections. Demersals CHN ECA EAP ECA CHN JAP 31% 31% 21% 43% 32% 15% Pelagics AFR SEA EAP ROW CHN NAM that the African need for fish imports increases as the population 45% 28% 14% 34% 24% 23% increases, but the price of traditional import fish will likely rise (due Other marine JAP ECA AFR JAP ECA LAC to rise in fishmeal price) such that imports will be substituted with 46% 19% 13% 45% 17% 11% freshwater fish, which are predicted to become relatively more Sources: FAO FIPS FBS and IMPACT model projections. Note: The results shown in tables 3.10a and 3.10b represent net exports and abundantly available for direct human consumption. net imports, respectively. Unlike many other agricultural commodities, in fish trade it is common that one country is both an exporter and importer of certain species. For example, the United States and Europe are both large exporters and Figure 3.6 shows the projected changes in the real prices during importers of salmon, but the fact is buried underneath the net trade results that North America is a large net importer and that Europe (and Central Asia) is the the 2010–30 period. The model projects that the prices of all fish second-largest net exporter, after Latin America. Latin America, on the other hand, is a larger exporter but not an importer of salmon; thus, the net trade and fish products will increase during the period. While modest results are easier to interpret for this region. ECA = Europe and Central Asia; NAM = North America; LAC = Latin America and Caribbean; CHN = China; JAP = increases are expected for most fish species, higher price increases Japan; EAP = other East Asia and the Pacific; SEA = Southeast Asia; IND = India; SAR = other South Asia; MNA = Middle East and North Africa; AFR = Sub-Saharan are expected for fish in the pelagics, other marine, and demersals Africa; ROW = rest of the world. categories. These are used as ingredients of fishmeal and fish oil, whose prices are expected to rise substantially more than those of fish for direct consumption. 14 percent in 2006 to 25 percent in 2030. China’s share in global salmon net imports is projected to increase from a third in 2006 to more than half by 2030. SEA is expected to increase their net export 3.4. FISHMEAL AND FISH OIL share in freshwater and diadromous fish from 49 percent in 2006 Production to 79 percent in 2030. On the other hand, much of freshwater and Globally, a little less than 20 percent of total fish produced is cur- diadromous fish was destined to ECA in 2006, while exactly half of it rently used for fishmeal and fish oil production, and the proportion will likely be imported by AFR by 2030. is expected to remain unchanged into 2030. As seen in table 3.11, Note that the importance of African imports for pelagics and other global production of fishmeal in 2030 is projected to be around 7.6 marine fish declines but the importance of fish in the freshwater and million tons. Looking across regions, as expected, LAC is the larg- diadromous category increases substantially. The model predicts est fishmeal-producing region, accounting for about 40 percent of A G R I C U LT U R E A N D R U R A L D E V E LO P M E N T D I S C U S S I O N PA P E R 48 C H A P T E R 3 — I M PA C T P R O J E C T I O N S TO 2030 U N D E R T H E B A S E L I N E S P E C I F I C AT I O N TABLE 3.11: Projected Total Fishmeal Production by Region DATA (000 TONS) PROJECTION (000 TONS) SHARE IN GLOBAL TOTAL % CHANGE 2008 2010 2020 2030 2010 (PROJECTION) 2030 (PROJECTION) 2010–30 Global total 5,820 7,044 7,401 7,582 100.0% 100.0% 7.6% ECA 703 1,000 1,005 1,008 14.2% 13.3% 0.7% NAM 262 372 375 376 5.3% 5.0% 1.1% LAC 2,305 3,033 3,064 3,080 43.1% 40.6% 1.5% EAP 50 82 97 105 1.2% 1.4% 27.9% CHN 1,319 815 903 941 11.6% 12.4% 15.4% JAP 204 421 421 421 6.0% 5.6% 0.0% SEA 556 615 719 779 8.7% 10.3% 26.7% SAR 78 57 66 71 0.8% 0.9% 25.3% IND .. 10 11 12 0.1% 0.2% 17.3% MNA 88 92 119 133 1.3% 1.8% 44.4% AFR 99 170 192 203 2.4% 2.7% 19.3% ROW 157 376 428 452 5.3% 6.0% 20.3% Sources: Compilation of data from FishStat, Oil World, and the IFFO and IMPACT model projections. Note: ECA = Europe and Central Asia; NAM = North America; LAC = Latin America and Caribbean; CHN = China; JAP = Japan; EAP = other East Asia and the Pacific; SEA = Southeast Asia; IND = India; SAR = other South Asia; MNA = Middle East and North Africa; AFR = Sub-Saharan Africa; ROW = rest of the world; .. = negligible. world’s fishmeal supply. In fact, in Latin America, reduction demand Utilization accounts for about three-quarters of their fish use. The projected Table 3.12 shows the projected distribution of fishmeal use across fishmeal production of Latin America in 2030 is slightly more than region. Note that this table pertains to total fishmeal use, including that of all of Asia combined (including JAP and EAP). SEA—the its use for livestock production. The use of fishmeal is concentrated fourth-largest producing region of fishmeal after LAC, ECA, and in China, which is estimated to represent more than 40 percent of CHN—is expected to grow rapidly in the global fishmeal market, global fishmeal use throughout the projection period. Since avail- with their production volume reaching more than 10 percent of the able data suggest that the use of fishmeal in the livestock sector in global total by 2030. China is negligible, the projected growth of fishmeal use in China is driven almost entirely by the growth of the aquaculture sector. Use of Fish Processing Waste The second-biggest user of fishmeal is the SEA region, and the ECA Throughout the projection period, the model indicates that about region is projected to be the third-largest user. Given the substantial 15 percent of the global fishmeal supply originates from fish proc- rise in the projected prices of fishmeal and fish oil, these products essing waste. Due to the issues surrounding data inconsistency are expected to be selectively allocated for the production of high- discussed in chapter 2, the model overestimates the production value commodities, both in aquaculture and livestock production. of fishmeal relative to the available data. Thus, the baseline model Efficiency of feed use that is embedded in the default specification does not reproduce the waste use figure indicated by the IFFO, also reduces the need for feed per unit of fish and livestock output. where an estimated 25 percent of global fishmeal is currently The last point is demonstrated in the next subsection. produced from fish processing waste (Shepherd 2012). The restric- tions on the set of countries that are allowed in the model to use Feed Efficiency Improvement fish processing waste will be relaxed in a scenario presented in the The projected growth of fishmeal production and utilization sharply next chapter. differ from the fast growth projected for aquaculture production, F I S H TO 2 030: P R O S P E C T S F O R F I S H E R I E S A N D A Q UA C U LT U R E C H A P T E R 3 — I M PA C T P R O J E C T I O N S TO 2 0 3 0 U N D E R T H E B A S E L I N E S P E C I F I C AT I O N 49 TABLE 3.12: Projected Fishmeal Use by Region PROJECTION (000 TONS) SHARE IN GLOBAL TOTAL % CHANGE 2010 2030 2010 2020 2030 (PROJECTION) (PROJECTION) 2010–30 Global total 7,045 7,402 7,583 100.0% 100.0% 7.6% ECA 1,009 1,075 1,195 14.3% 15.8% 18.5% NAM 79 68 72 1.1% 1.0% –8.6% LAC 214 163 136 3.0% 1.8% –36.3% EAP 39 20 15 0.6% 0.2% –62.6% CHN 3,262 3,379 3,390 46.3% 44.7% 3.9% JAP 434 505 595 6.2% 7.8% 36.9% SEA 1,148 1,244 1,264 16.3% 16.7% 10.1% SAR 232 311 298 3.3% 3.9% 28.3% IND 257 416 466 3.6% 6.1% 81.8% MNA 155 110 80 2.2% 1.1% –48.5% AFR 208 105 67 2.9% 0.9% –67.6% ROW 10 6 5 0.1% 0.1% –49.0% Source: IMPACT Model projections. Note: ECA = Europe and Central Asia; NAM = North America; LAC = Latin America and Caribbean; CHN = China; JAP = Japan; EAP = other East Asia and the Pacific; SEA = Southeast Asia; IND = India; SAR = other South Asia; MNA = Middle East and North Africa; AFR = Sub-Saharan Africa; ROW = rest of the world. especially of those species such as shrimp and salmon that have FIGURE 3.7: Projected Production and Fishmeal Use in Global a higher dependence on fishmeal for their production. These Fed Aquaculture Production/fishmeal use, thousand tons results originate from the assumption used in the model that the 70,000 Fishmeal use Fed aquaculture production importance of fishmeal and fish oil in aquaculture will decline as the 60,000 industry continues to develop alternative feeds from plant-based 50,000 sources and to improve efficiencies in feeding practices over time. 40,000 As was mentioned before, this is one of the key drivers of aquacul- 30,000 ture growth incorporated in the model. 20,000 Figures 3.7 and 3.8 illustrate the assumed efficiency improvement 10,000 in fishmeal use in global aquaculture. Figure 3.7 contrasts the rate – at which aquaculture production of fed species23 is projected to 20 0 20 2 20 4 06 08 10 12 14 16 18 20 20 2 24 26 28 30 0 0 0 2 20 20 20 20 20 20 20 20 20 20 20 20 grow and the growth rate of fishmeal use. The projected growth in Source: IMPACT model projections. fed aquaculture over the 2000–30 period, equivalent to an annual improvement in the efficiency of aquaculture fishmeal use results in average growth rate of 3.9 percent per year, is much faster than the a constant decline in the average FCR. projected growth in fishmeal use in aquaculture (an average annual growth rate of 1.7 percent). Figure 3.8 plots the average feed conver- Aquaculture vs. Livestock sion ratio (FCR) for global fed aquaculture—that is, how much fish is The pressure for aquaculture to improve efficiency of fishmeal use produced per unit of fishmeal used in fed aquaculture. The assumed also reflects the increasing competition for fishmeal on the global animal feed markets between aquaculture and livestock produc- 23 Aquaculture production of shrimp, crustaceans, salmon, tilapia, Pangas- ers. Given the substantial and sustained growth of aquaculture ius/catfish, carp, OCarp, EelStg, MDemersal, CobSwf, OPelagic, and OMa- rine are considered in the calculation. that is projected, the overall amount of fishmeal that goes toward A G R I C U LT U R E A N D R U R A L D E V E LO P M E N T D I S C U S S I O N PA P E R 50 C H A P T E R 3 — I M PA C T P R O J E C T I O N S TO 2030 U N D E R T H E B A S E L I N E S P E C I F I C AT I O N FIGURE 3.8: Projected Average Feed Conversion Ratio for TABLE 3.13: Projected Net Exports of Fishmeal by Region Fishmeal in Global Fed Aquaculture (000 tons) 0.25 DATA PROJECTION PROJECTION PROJECTION 2008 2010 2020 2030 0.20 Global total 2,111 3,518 3,768 3,882 0.15 ECA –660 –8 –70 –187 NAM 16 293 307 304 0.10 LAC 2,122 2,820 2,901 2,945 0.05 EAP –29 43 77 90 CHN –1,361 –2,446 –2,476 –2,449 0 JAP –320 –13 –84 –173 00 02 04 06 08 10 12 14 16 18 20 22 24 26 28 30 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 SEA –144 –533 –525 –485 Source: IMPACT model projections. SAR 2 –175 –244 –227 aquaculture will likely continue to grow. Higher feed prices will im- IND –1 –246 –404 –454 ply that only feed-efficient and high-valued aquaculture products MNA 37 –63 9 53 can be profitable with such inputs. As seen in figure 1.5, the use AFR –35 –38 88 136 ROW 154 366 422 447 of global fishmeal by aquaculture grew from nil in 1960 to 10 per- Sources: FAO FIPS FBS and IMPACT model projections. cent in 1980 and to 73 percent in 2010, and accordingly the share Note: ECA = Europe and Central Asia; NAM = North America; LAC = Latin America and Caribbean; CHN = China; JAP = Japan; EAP = other East Asia and the Pacific; of swine and poultry production has fallen sharply (Jackson and SEA = Southeast Asia; IND = India; SAR = other South Asia; MNA = Middle East and North Africa; AFR = Sub-Saharan Africa; ROW = rest of the world. Shepherd 2010; Shepherd 2012). Trade TECHNICAL APPENDIX Table 3.13 shows the projected trade patterns of fishmeal across A. Comparisons with OECD-FAO Analysis to 2020 regions that make possible the fishmeal use patterns seen in In this analysis, we undertake a comparison of the projections by table 3.12 and the production patterns in table 3.2. Again, trade the IMPACT model to those in the fish section of the 2012 OECD- is represented in terms of net exports, with negative numbers FAO analysis World Agricultural Outlook 2012–2021. The OECD-FAO representing net imports. From the table, we see that the largest fish analysis is conducted as part of the annual agricultural outlook exporter is Latin America, whereas the biggest importer is China. series carried out by the joint effort of OECD and FAO agricultural The volumes of Chinese imports and Latin American exports are market outlook teams using the combined AgLink-CoSiMo multi- similar. According to Globefish (2011), exports from Peru and Chile market, partial equilibrium model (Dowey 2007). accounted for 70 to 80 percent of Chinese fishmeal imports during The Fish and Seafood Model used for the OECD-FAO fish projections the 2008–09 period. The projection results suggest that the trade is not fully integrated into the larger AgLink-CoSiMo framework. patterns are likely to continue into 2030. Rather it consists of a stand-alone model covering the same 56 The model predicts that net imports of Southeast Asia will decrease countries (and country-groupings) of AgLink-CoSiMo. There are es- over that period, given that a faster growth is expected for fishmeal sentially three commodities in this model: (an aggregate of all) fish, production than its use in the region. Since most of the fishmeal use fishmeal, and fish oil. Each commodity has its corresponding world in Southeast Asia is accounted for by aquaculture (already reaching market and a market-clearing price. From the rather short model 99 percent by 2015, as seen in table 3.13), the dynamics of efficiency description given in the chapter on fish in the OECD-FAO outlook in fishmeal use explain these trends almost entirely in this region. (OECD-FAO 2012), it is not fully clear how the Fish and Seafood India is projected to increase imports over the projection period Model is linked to the larger AgLink-CoSiMo model. Presumably the and become the third-largest net importer of fishmeal in 2030, after oilseed meal prices from the larger model are “fed” into the Fish and China and Southeast Asia. Seafood Model in an exogenous way so that aquaculture demand F I S H TO 2 030: P R O S P E C T S F O R F I S H E R I E S A N D A Q UA C U LT U R E C H A P T E R 3 — I M PA C T P R O J E C T I O N S TO 2 0 3 0 U N D E R T H E B A S E L I N E S P E C I F I C AT I O N 51 for feed can adjust accordingly. Furthermore, the livestock demand FIGURE 3.9: Comparison of IMPACT and OECD-FAO for fishmeal may also be imposed on the Fish and Seafood Model Projections for Global Fish Supply in an exogenous way, while the fishmeal prices that the Seafood 200 OECD-FAO IMPACT Model generates endogenously may be used in the larger AgLink- CoSiMo model as well as the feed requirement of aquaculture. The OECD-FAO model combines fish of all species into one aggre- 150 gate category, while they do maintain the distinction between cap- ture and aquaculture production. Aquaculture supply in both the Million tons IMPACT and OECD-FAO models takes into account the availability of 100 fishmeal and its price, and the two models seem to do so in a similar manner. Given that the OECD-FAO model also uses the UN projec- tions of human population, its total food fish demand projections are likely derived as per capita demand times the population. The 50 OECD-FAO baseline projections use a set of GDP growth drivers that are derived from an internal economic outlook process (based with- in OECD). These are not identical to, but clearly parallel to, the World – Bank GDP projections used in this study. The existing documenta- 00 04 08 12 16 20 24 28 20 20 20 20 20 20 20 20 tion for the OECD-FAO Fish and Seafood Model does not go into Sources: OECD-FAO 2012 and IMPACT model projections. specifics on supply-side drivers or their underlying assumptions. On the supply side, because of fishing quotas, only 12 percent of world capture is assumed to react to price in the model, while 99 percent of world aquaculture reacts to price. Farmed species requiring con- FIGURE 3.10: Comparison of IMPACT and OECD-FAO centrated feeds also react to feed prices composed by fishmeal, fish Projections for Global Aquaculture Production oil, and cereals in different proportion, depending of the species. 100 OECD-FAO IMPACT After having “anchored” their starting assumptions based on the data for 2010 and the provisional data for 2011, the OECD-FAO 80 model initiates their projections in year 2012 and continues up to year 2021. The available report (OECD-FAO 2012) does not show country-level details of the projections. Therefore, we have aggre- 60 Million tons gated our results over countries as well as over species. Despite these aggregations, we are able to gain some insight from the model result comparisons to 2021. 40 Figure 3.9 compares the projections of the two models for the total fish supply at the global level. The figure shows that IMPACT projec- 20 tions from the year 2000 meet up perfectly with OECD-FAO projec- tions at their starting point of 2012 and proceed in tandem to their ending point of 2021. – 0 4 08 12 16 20 24 28 0 0 Figure 3.10 compares the two sets of projections for global aqua- 20 20 20 20 20 20 20 20 culture production. Again, there is a close match between the two Sources: OECD-FAO 2012 and IMPACT model projections. A G R I C U LT U R E A N D R U R A L D E V E LO P M E N T D I S C U S S I O N PA P E R 52 C H A P T E R 3 — I M PA C T P R O J E C T I O N S TO 2030 U N D E R T H E B A S E L I N E S P E C I F I C AT I O N FIGURE 3.11: Comparison of IMPACT and FIGURE 3.12: Comparison of IMPACT and OECD-FAO OECD-FAO Projections for Global Projections for Global Fishmeal Supply Food Fish Consumption 8 OECD-FAO IMPACT 160 OECD-FAO IMPACT 7 140 6 120 5 Million tons 100 4 Million tons 80 3 60 2 40 1 20 – 00 04 08 12 16 20 24 28 – 20 20 20 20 20 20 20 20 00 04 08 12 16 20 24 28 20 20 20 20 20 20 20 20 Sources: OECD-FAO 2012 and IMPACT model projections. Sources: OECD-FAO 2012 and IMPACT model projections. series in 2012, where the IMPACT projections meet with the OECD- Figure 3.12 compares the projected fishmeal supply by the two FAO projections, through 2021. models. The figure shows a fairly consistent gap between the two sets of projections. Again, this is likely due to the consistent over- These comparisons suggest that, even though the two models prediction of fishmeal production by IMPACT relative to the data differ substantially in the aggregation levels of fish species and to reconcile across data on fish reduction demand and fishmeal countries, the overall tendency of the two models in predicting production and trade. Nonetheless, the projected trends are similar how the changes in demand drivers affect the expansion of supply in both series: both models project a modest but steady increase is similar and consistent. The two models appear to be congruent during the 2012–21 period. in the how the FAO data are used on capture and aquaculture production. Since the projections of aquaculture production by the two models However, when we compare projections of food fish consumption, are nearly the same, the gap found in the two projection series of we begin to see differences in the model projections. As seen in fishmeal supply implies that fishmeal requirement per unit of aqua- figure 3.11, the projections of food fish consumption by the IMPACT culture production in the OECD-FAO model must be lower. This is model are consistently lower than those of the OECD-FAO model. confirmed in the rough estimates of feed conversion ratio implied This appears to be related to the consistent underprediction of food by the projections. In figure 3.13, the implied FCRs—derived as the fish consumption by IMPACT relative to the data for the calibration ratio of total fishmeal supply to total aquaculture production—from period discussed in section 2.5. The underprediction by IMPACT is the two models show a similar trajectory over the 2012–21 period. because the projections are allowed to deviate from the data to rec- However, the ratio for the IMPACT model is consistently higher than oncile the inconsistency originating from fishmeal-related data. In that for the OECD-FAO model. Thus, the two models likely differ in contrast, the OECD-FAO model results appear to adhere to the FAO the way to arrive at the amount of feed necessary to support a given data on fish utilization (that is, food fish consumption and meals amount of aquaculture production. It should be noted that the feed input). This is also seen in the projection comparison for fishmeal requirement parameters in IMPACT are derived from the literature production next. and specified for each of the 15 categories of fish. It is unlikely that F I S H TO 2 030: P R O S P E C T S F O R F I S H E R I E S A N D A Q UA C U LT U R E C H A P T E R 3 — I M PA C T P R O J E C T I O N S TO 2 0 3 0 U N D E R T H E B A S E L I N E S P E C I F I C AT I O N 53 FIGURE 3.13: Comparison of IMPACT and OECD-FAO a similar approach is taken for the single aggregate fish category in Projections for Implied FCR for Fishmeal the OECD-FAO model. 0.25 While we are unable to undertake a more comprehensive compari- OECD-FAO IMPACT son between the two studies, this brief analysis illustrates some im- portant points of similarity and difference in how the future of fish 0.20 supply and demand are projected. It confirms that the two studies make similar use of data and methodology and that basic supply and food demand projections are in line. However, there likely exist 0.15 Million tons differences in the way the fishmeal and fish oil supply and demand are modeled. Some differences could also arise in the way the fish side of IMPACT is linked to the rest of the crop and livestock mar- 0.10 kets, compared with how this is done in the OECD-FAO framework. They describe a much “looser” link between the two parts of the model in their technical description study (OECD-FAO 2012), but it 0.05 is not possible to fully judge the extent and implications of these differences without a much more detailed investigation. While a full comparison of the two models is beyond the scope of this study, – the brief analysis presented here could be a useful starting point for 00 04 08 12 20 24 28 20 20 20 20 20 20 20 discussion for comparison and further model improvement for both Sources: OECD-FAO 2012 and IMPACT model projections. research teams. A G R I C U LT U R E A N D R U R A L D E V E LO P M E N T D I S C U S S I O N PA P E R C H A P T E R 4 — I M PA C T P R O J E C T I O N S TO 2 0 3 0 U N D E R S E L E C T E D S C E N A R I O S 55 Chapter 4: IMPACT PROJECTIONS TO 2030 UNDER SELECTED SCENARIOS In this chapter, we explore a set of illustrative scenarios, so that 4.1. SCENARIO 1: FASTER AQUACULTURE GROWTH we can (1) better understand the sensitivities of the model re- As the first exercise, we implement a scenario in which aquaculture sults to changes in some key parameters and (2) gain insights production will grow at a faster pace for all species in all countries into potential impacts on the global fish markets of changes in and regions. We have constructed this scenario in the same spirit the drivers of future fish supply, demand, and trade. In the spirit as the aquaculture scenarios in the Fish to 2020 study (see Delgado of the scenarios carried out in the Fish to 2020 study, we explore and others 2003, table 4.1), in which the exogenous growth rates some similar cases where aquaculture is able to grow faster than of aquaculture production were increased and decreased by 50 under the baseline scenario (scenario 1) and cases with differ- percent of the baseline values. Here, we increase the aquaculture ent assumptions on the future productivity growth of capture growth rates by 50 percent from 2011 through 2030. A scenario of fisheries. In particular, we explore both positive and negative reduced growth rates is discussed later in this chapter in the context scenarios of future capture fisheries, with the former potentially of aquaculture disease outbreak. being achieved through effective tenure reforms (scenario 5) and the latter associated with global climate change (scenario Table 4.1 compares the results of the scenario with the baseline 6). We also implement some scenarios that were not considered results on aquaculture production. At the global level, the total in the Fish to 2020 study. In particular, we investigate how allow- production at the end of the projection period would increase from ing expanded use of fish processing waste in fishmeal and fish 93.6 million tons under the baseline scenario to more than 101 mil- oil production might affect the market of these fish-based prod- lion tons under the current scenario, representing an 8.1 percent in- ucts (scenario 2) and the implications of a large-scale disease crease. At the regional level, the projected aquaculture production outbreak in aquaculture for the markets of affected species and levels in 2030 are higher in this scenario compared to the baseline other commodities (scenario 3). In addition to these scenarios scenario in most regions. Some regions, in particular LAC and MNA, that are based on supply-side shocks, we consider an alternative would benefit proportionately more from this scenario. In contrast, demand-side scenario, where consumers in China expand their North America and Japan would lose from this scenario. That is, demand for high-value fish products more aggressively than in even though we have increased the exogenous growth rates by 50 the baseline case (scenario 4). percent in all regions, it does not necessarily translate into the same growth rate increase across regions in the final results. Each scenario is constructed by changing some specific set of pa- rameters. By examining the results from these illustrative cases, we This is, in part, due to the interactions with the demand side, es- are able to gain a deeper appreciation for how some key drivers of pecially with the market for fishmeal, an important aquaculture supply and demand growth can change the market outcomes in production input. Faster growth of aquaculture production would the medium-term outlook to 2030 and help identify where policy or entail more fishmeal use by some (mostly carnivorous) fish species technology interventions can be most useful. groups, which is expected to drive fishmeal price upward. The latter, A G R I C U LT U R E A N D R U R A L D E V E LO P M E N T D I S C U S S I O N PA P E R 56 C H A P T E R 4 — I M PA C T P R O J E C T I O N S TO 2030 U N D E R S E L E C T E D S C E N A R I O S TABLE 4.1: Projected Effects of Faster Aquaculture Growth on Aquaculture Supply by Region FASTER GROWTH SCENARIO 1 BASELINE (000 TONS) (SCENARIO 1) (000 TONS) RELATIVE TO BASELINE 2008 2030 2030 2030 (DATA) (PROJECTION) (PROJECTION) (PROJECTION) Global total 52,843 93,612 101,220 8.1% ECA 2,492 3,761 3,796 0.9% NAM 655 883 840 –4.9% LAC 1,805 3,608 4,455 23.5% EAP 751 1,066 1,114 4.5% CHN 33,289 53,264 56,153 5.4% JAP 763 985 948 –3.8% SEA 6,433 14,848 16,882 13.7% SAR 1,860 4,163 4,850 16.5% IND 3,585 8,588 9,179 6.9% MNA 921 1,911 2,400 25.6% AFR 231 464 525 13.1% ROW 57 72 76 6.6% Sources: FishStat and IMPACT model projections. Note: ECA = Europe and Central Asia; NAM = North America; LAC = Latin America and Caribbean; CHN = China; JAP = Japan; EAP = other East Asia and the Pacific; SEA = Southeast Asia; IND = India; SAR = other South Asia; MNA = Middle East and North Africa; AFR = Sub-Saharan Africa; ROW = rest of the world. in turn, would slow down aquaculture expansion for some species. commodity prices and the fishmeal market and their feedback to On the other hand, as aquaculture growth accelerates, there would the aquaculture fish supply. be larger dampening effects on fish prices, which also in turn would Table 4.2 shows the final outcome after such feedback for produc- work to slow down the fish supply. The manner in which the price tion and world prices of each species. While the model predicts that would rise or drop differs for each commodity. Further, each region global aquaculture production in 2030 under this scenario would has a different commodity mix, some with more fishmeal-intensive increase by 8.1 percent relative to the baseline specification, the aquaculture and others with fish species whose market demands increase in total fishmeal use would be limited to 2 percent, up from are more sensitive to price changes. Therefore, the final results, 7,582 thousand tons under the baseline scenario to 7,744 thousand obtained as the equilibrium outcome in the global fish markets, tons. In contrast, its projected price in 2030 is higher by 13 percent represent intricate balancing of supply and demand, responding to than under the baseline. Since the supply of fishmeal ingredients signals transmitted by world prices. from capture fisheries is more or less fixed, the supply response of In the Fish to 2020 study, this scenario was aimed to explore what the fishmeal to price increases is limited. The supply of fishmeal thus cross-price response would be in capture production if aquaculture would be reallocated, through the market price mechanism, away were to grow at an accelerated pace. However, the scenario does from livestock and lower-value fish toward higher-value aquaculture not affect the capture production in this study because capture species. In the process, supply of some high-value fish products, supply is not price responsive—it is determined solely by specified such as mollusks and salmon, is increased so much that their prices exogenous rates of growth in the current model. Therefore, we do are projected to fall noticeably relative to the baseline case. In fact, not measure the same effects as in Fish to 2020. In this study, what except for fishmeal and pelagics, prices in all categories would fall we really measure is the effects of faster aquaculture growth on relative to the baseline scenario. Fish in the OPelagic category are F I S H TO 2 030: P R O S P E C T S F O R F I S H E R I E S A N D A Q UA C U LT U R E C H A P T E R 4 — I M PA C T P R O J E C T I O N S TO 2 0 3 0 U N D E R S E L E C T E D S C E N A R I O S 57 TABLE 4.2: Projected Effects of Faster Aquaculture Growth on Aquaculture Supply and Commodity Prices PRODUCTION AQUACULTURE PRODUCTION IN SCENARIO 1 RELATIVE TO CONSUMPTION CATEGORY 2030 (000 TONS) BASELINE CATEGORY FASTER GROWTH BASELINE (SCENARIO 1) PRODUCTION PRICE Shrimp 8,061 8,868 10% –0.5% Shrimp Crustaceans 2,174 2,369 9% –0.9% Crustaceans Mollusks 22,689 25,359 12% –1.7% Mollusks Salmon 3,613 4,015 11% –1.9% Salmon Tuna 13 14 6% –0.2% Tuna Tilapia 6,446 8,343 29% –0.7% Freshwater and diadromous Pangasius/catfish 5,040 5,079 1% Carp 19,301 19,999 4% OCarp 15,190 15,369 1% EelStg 480 489 2% OFresh 6,473 6,523 1% MDemersal 2,105 2,229 6% –0.2% Demersals Mullet 524 604 15% CobSwf 41 43 4% 1.3% Pelagics OPelagic 199 198 –0.2% OMarine 1,261 1,720 36% –0.7% Other marine Fishmeal 7,582 7,744 2% 13.0% Fishmeal Source: IMPACT model projections. Note: Pangasius/catfish = Pangasius and other catfish; OCarp = silver, bighead, and grass carp; EelStg = aggregate of eels and sturgeon; OFresh = freshwater and diadromous species (excluding tilapia, Pangasius/ catfish, carp, OCarp, and EelStg); MDemersal = major demersal fish; CobSwf = aggregate of cobia and swordfish; OPelagic = other pelagic species; OMarine = other marine fish. the main ingredient of fishmeal, and greater competition for this generated limited insight since it did not completely endogenize category between direct human consumption and use in fishmeal the important links between food fish markets and fishmeal and fish production would contribute to the price increase. oil markets. In the current model, the use of fish for human con- sumption and for conversion into feed is determined endogenously Together, these effects explain why the increase in projected aqua- through supply-demand balance regulated through world prices. culture growth under this scenario is not as uniformly large across Further, the model now allows the use of fish processing waste in regions and across species as one would have expected purely from the production of fishmeal and fish oil, as explained in chapter 2. the scenario design. This exercise also illustrates why the links with The IFFO estimates that currently about 25 percent of the world’s fishmeal and fish oil markets are so important in understanding the fishmeal is generated from fish processing waste (Shepherd 2012). place of aquaculture products in the world food economy. The proportion is expected to rise, given the growth of aquaculture of large fish species and associated development of fish process- 4.2. SCENARIO 2: EXPANDED USE OF FISH ing industry, together with the trend of rising fishmeal and fish oil PROCESSING WASTE IN FISHMEAL AND FISH prices. OIL PRODUCTION The Fish to 2020 study examined a scenario of improved efficiency In the current scenario, we remove the restriction in the number in fishmeal and fish oil use in aquaculture. However, the model of countries that are able to use fish processing waste from the A G R I C U LT U R E A N D R U R A L D E V E LO P M E N T D I S C U S S I O N PA P E R 58 C H A P T E R 4 — I M PA C T P R O J E C T I O N S TO 2030 U N D E R S E L E C T E D S C E N A R I O S 20 countries/groups of countries in the baseline scenario,24 which TABLE 4.3: Projected Amount of Fish Processing Waste Used was determined following the data and in the data reconciliation in Fishmeal Production by Region (000 tons) process (see technical appendix C to chapter 2). The scenario now PROCESSING WASTE BASELINE (SCENARIO 2) allows any country that produces fishmeal to use fish processing 2010 2030 2030 waste starting in 2011. The assumptions on fish species whose (PROJECTION) (PROJECTION) (PROJECTION) waste can be used for fishmeal and fish oil production as well as Global total 5,304 5,656 10,206 ECA 2,193 2,194 2,381 the volume of waste per unit of live fish are shown in table 2.9. It NAM 1,141 1,165 1,057 is assumed that the amount of fish processing waste each country LAC 849 1,012 738 can use is restricted to the amount of waste produced, whether EAP 2 3 173 capture or aquaculture, in the country in a given year. This repre- CHN n.a. n.a. 2,856 sents a limitation of the model because, in reality, fish are traded for JAP 521 521 521 processing purposes and final processed products are also widely SEA 597 760 1,315 traded. In particular, China and Thailand increasingly import raw SAR n.a. n.a. 318 material for reexport of processed products (FAO 2012). However, IND n.a. n.a. 419 since the current model does not have separate supply functions for MNA n.a. n.a. 91 processed seafood, it does not keep track of the countries in which AFR 1 1 98 ROW n.a. n.a. 240 fish processing takes place. For this reason, the model is subject to Source: IMPACT model projections. overprediction of waste use in some countries and underprediction Note: ECA = Europe and Central Asia; NAM = North America; LAC = Latin America and Caribbean; CHN = China; JAP = Japan; EAP = other East Asia in others. Nonetheless, the extent of this cannot be confirmed, as no and the Pacific; SEA = Southeast Asia; IND = India; SAR = other South Asia; MNA = Middle East and North Africa; AFR = Sub-Saharan Africa; ROW = rest of data exist on this issue to our knowledge. The levels of fish waste use the world; n.a. = not applicable. are endogenously determined based on world prices of fishmeal, while we do not assign any market-clearing mechanism or price to come from those countries that are newly allowed to use waste, the supply or use of fish processing waste. The specific procedure is notably China. described in chapter 2. As a result of the additional volume of fish processing waste made Allowing in the model all countries to freely use fish processing available for fishmeal production, the model projects a substantial waste would effectively increase the supply of feedstock that is increase in fishmeal supply in 2030: from 7,582 thousand tons in available for reduction into fishmeal and fish oil, much of which baseline to 8,473 thousand tons in this scenario, or a 12 percent in- is likely to be used given the high fishmeal price simulated in the crease. The expanded use of processing waste would slightly reduce baseline scenario. Table 4.3 contrasts the results under the baseline the pressure on capture fisheries of supplying fishmeal ingredients. and the current scenarios in terms of usage of fish processing waste. The whole fish used for fishmeal production would reduce from Globally, the projected use of processing waste in 2030 increases 28,367 thousand tons under the baseline scenario to 27,646 thou- from 5.7 million tons under the baseline case to more than 10 mil- sand tons under the scenario. The proportion of fishmeal produced lion tons under the scenario. Looking across regions, while relatively based on processing waste also increases from 15 percent to 26 few regions use processing waste in the baseline case, waste use is percent (figure 4.1). represented in all regions under the scenario. In fact, most of the gain in the use of fish processing waste under the scenario would The increase in fishmeal supply would result in a reduction in its price. For the 11.8 percent increase in supply, the correspond- 24 The 20 countries/groups of countries are Argentina, Australia, Belgium- ing price reduction is projected to be 14.1 percent (table 4.4). The Luxembourg, Brazil, British Isles (including Ireland), Canada, Chile, Côte reduced fishmeal price in turn would encourage aquaculture pro- d’Ivoire, France, Germany, Italy, Japan, Mexico, Russian Federation, Scan- dinavia, Spain/Portugal, Thailand, United States, Uruguay, and Vietnam. duction. At the global level, aquaculture production in 2030 would F I S H TO 2 030: P R O S P E C T S F O R F I S H E R I E S A N D A Q UA C U LT U R E C H A P T E R 4 — I M PA C T P R O J E C T I O N S TO 2 0 3 0 U N D E R S E L E C T E D S C E N A R I O S 59 FIGURE 4.1: Projected Increase in Fishmeal Production due to increase 1.9 percent relative to the baseline case, from 93.6 million Usage of Whole Fish and Fish Processing Waste tons to 95.4 million tons. Looking across species, a reduced price of in 2030 fishmeal would benefit the tuna, salmon, and crustacean aquacul- 9,000 Processing waste Whole fish ture industry as well as species in the freshwater and diadromous 8,000 category. All of these changes would lead to reduced fish prices. 7,000 6,000 It is worth noting that expansion and improvements of processing Thousand tons 5,000 facilities for fishmeal and fish oil production could have unintended 4,000 effects on wild fisheries. While the intended benefit is increased use 3,000 of catch and processing waste that are currently unused, expanded 2,000 processing capacity and markets could also result in greater reduc- 1,000 tion demand for fish, including those that otherwise would be used – for direct human consumption and potentially encouraging har- Baseline Processing waste vest of all kinds of fish, including some protected and endangered (Scenario 2) Source: IMPACT model projections. species. TABLE 4.4: Projected Effects of Expanded Use of Fish Processing Waste in Fishmeal Production on Aquaculture Supply and Commodity Prices PRODUCTION CONSUMPTION CATEGORY AQUACULTURE PRODUCTION IN 2030 (000 TONS) SCENARIO 2 RELATIVE TO BASELINE CATEGORY WASTE USE BASELINE (SCENARIO 2) PRODUCTION PRICE Global total 93,612 95,389 1.9% n.a. Global Total Shrimp 8,061 8,111 0.6% –0.1% Shrimp Crustaceans 2,174 2,227 2.4% –0.3% Crustaceans Mollusks 22,689 22,657 –0.1% –0.1% Mollusks Salmon 3,613 3,731 3.2% –0.7% Salmon Tuna 13 14 4.2% –0.2% Tuna Tilapia 6,446 6,587 2.2% –0.4% Freshwater and diadromous Pangasius/catfish 5,040 5,149 2.2% Carp 19,301 19,807 2.6% OCarp 15,190 15,800 4.0% EelStg 480 496 3.4% OFresh 6,473 6,676 3.1% MDemersal 2,105 2,102 –0.1% –0.4% Demersals Mullet 524 522 –0.3% CobSwf 41 41 0.4% –1.8% Pelagics OPelagic 199 201 1.1% OMarine 1,261 1,268 0.6% –0.5% Other marine Fishmeal 7,582 8,473 11.8% –14.1% Fishmeal Source: IMPACT model projections. Note: n.a. = not applicable; Pangasius/catfish = Pangasius and other catfish; OCarp = silver, bighead, and grass carp; EelStg = aggregate of eels and sturgeon; OFresh = freshwater and diadromous species (excluding tilapia, Pangasius/catfish, carp, OCarp, and EelStg); MDemersal = major demersal fish; CobSwf = aggregate of cobia and swordfish; OPelagic = other pelagic species; OMarine = other marine fish. A G R I C U LT U R E A N D R U R A L D E V E LO P M E N T D I S C U S S I O N PA P E R 60 C H A P T E R 4 — I M PA C T P R O J E C T I O N S TO 2030 U N D E R S E L E C T E D S C E N A R I O S 4.3. SCENARIO 3: A MAJOR DISEASE OUTBREAK IN FIGURE 4.2: Global Shrimp Supply under Baseline and SHRIMP AQUACULTURE IN ASIA Disease Scenarios As aquaculture continues to rapidly expand, the risk of catastrophic 14 Baseline Scenario disease outbreaks has become a major concern (Arthur and oth- 12 ers 2002). In open and semiopen aquaculture production systems, 10 especially, an incidence of contagious disease is difficult to contain Million tons 8 within a production unit and is likely to spread throughout the sys- tem through waters and even beyond the system through other 6 vectors and marketing activities. Furthermore, impacts of major, 4 catastrophic disease outbreaks are felt globally, given much of sea- 2 food are internationally traded. According to FAO (2012), 38 percent 0 of fish produced was exported in 2010. 00 03 06 09 12 15 18 21 24 27 30 20 20 20 20 20 20 20 20 20 20 20 In this scenario, we illustrate the impacts on the global market of a Source: IMPACT model projections. large-scale disease outbreak in a high-value aquaculture sector, in particular, shrimp aquaculture in Asia. We simulate a sudden decline in the aquaculture production of shrimp in the affected countries Since capture production of all species, including shrimp, is main- by 35 percent in 2015, relative to the baseline projection value in tained at the baseline levels through the specified exogenous rates 2015. We then allow the affected shrimp aquaculture to recover to of growth, the only possible direct response to the shock is in de- the 2015 level under the baseline by the year 2020. After that, the mand for shrimp25 and other commodities through changes in rela- production is allowed to continue along the same growth trajec- tive prices as well as aquaculture supply. Since shrimp aquaculture tory as was the case under the baseline. Essentially, we set back the is a heavy consumer of fishmeal, we also investigate the potential growth of Asian shrimp aquaculture by five years in this scenario and impacts of this sudden decline and temporary slowdown in Asian investigate how the shrimp production in other regions would re- shrimp aquaculture on feed demand and price and its feedback to spond and how the prices and regional trade might be affected. We other markets. have chosen this region for the disease scenario because of the high The shock simulated in this scenario and its impacts on the global concentration of shrimp production in Asia, especially in China and shrimp supply (from both capture and aquaculture origins) are il- Southeast Asia. The affected Asian countries are Bangladesh, China, lustrated in figure 4.2. The impact of 35 percent reduction in Asian India, Indonesia, Malaysia, Myanmar, the Philippines, Singapore, Sri shrimp aquaculture is somewhat attenuated in the global market of Lanka, Thailand, and Vietnam. all shrimp: the global shrimp supply would be reduced by 15 per- In contrast to scenario 1, in which we increase the growth rates of cent in the first year of the shock (disease outbreak). A fast recovery aquaculture production, we examine in this scenario the implica- is assumed during the subsequent five years so that, in the absence tions of a reduction in the growth rates due to a disease shock to of market interactions, the supply would recover to the baseline production. A scenario with reduced aquaculture growth rates was level in 2020. However, the projected recovery falls short of this also implemented in the earlier Fish to 2020 study. However, their level, presumably due to the price-dampening effects of the fast model contained a highly aggregated set of commodities for both recovery phase. After that the global production would continue aquaculture and capture production, and it could not be used for to grow, but it would never reach the baseline trajectory by 2030. specific scenarios in which a particular species was affected or the industry was affected by a shock of particular nature. Since the new model contains more disaggregated commodities, we can be much 25 While consumer perception and demand for affected commodities may be influenced by such large-scale disease outbreaks (Hansen and Ono- more flexible and targeted in designing scenarios. zaka 2011), such effects are not considered in this study. F I S H TO 2 030: P R O S P E C T S F O R F I S H E R I E S A N D A Q UA C U LT U R E C H A P T E R 4 — I M PA C T P R O J E C T I O N S TO 2 0 3 0 U N D E R S E L E C T E D S C E N A R I O S 61 TABLE 4.5: Projected Impact of Disease Outbreak on Shrimp Aquaculture Production (000 tons) 2015 2020 2030 DISEASE DISEASE DISEASE BASELINE (SCENARIO 3) BASELINE (SCENARIO 3) BASELINE (SCENARIO 3) AFFECTED REGIONS CHN 2,287.2 –691.3 2,571.5 –154.3 2,970.3 –73.8 IND 209.0 –59.5 228.7 –2.2 263.7 –0.3 SAR 160.7 –45.5 177.2 –0.4 207.2 0.5 SEA 2,085.7 –560.2 2,611.4 –280.8 3,593.1 –208.8 Subtotal 4,742.7 –1,356.4 5,588.8 –437.7 7,034.2 –282.3 UNAFFECTED REGIONS NAM 1.9 0.3 2.1 0.1 2.6 0.1 ECA 0.3 0.0 0.5 0.0 0.9 0.0 LAC 604.9 69.4 719.5 21.9 955.9 15.7 JAP 1.6 0.3 1.8 0.1 2.2 0.1 EAP 6.2 0.9 6.8 0.3 8.2 0.2 MNA 31.5 3.6 35.3 1.1 40.5 0.7 AFR 9.7 0.9 10.6 0.3 11.9 0.2 ROW 3.8 0.4 4.2 0.1 4.8 0.1 Subtotal 660.0 76.0 780.8 23.8 1,027.0 16.9 Global total 5,402.6 –1,280.5 6,369.6 –413.9 8,061.3 –265.5 Source: IMPACT model projections. Note: ECA = Europe and Central Asia; NAM = North America; LAC = Latin America and Caribbean; CHN = China; JAP = Japan; EAP = other East Asia and the Pacific; SEA = Southeast Asia; IND = India; SAR = other South Asia; MNA = Middle East and North Africa; AFR = Sub-Saharan Africa; ROW = rest of the world. * Subtotals and global totals may not match due to rounding error. Table 4.5 and figure 4.3 illustrate how the impacts of the disease FIGURE 4.3: Impact of Disease Outbreak on Shrimp outbreak on shrimp aquaculture are distributed across regions Aquaculture Production and over time.26 In the regions affected by the hypothetical dis- 2015 2020 2030 Percent change relative to baseline scenario 20% ease outbreak, the loss to shrimp aquaculture production is more 10% than 1.36 million tons in 2015, with nearly half of the loss coming N R A D H SA SE IN C from the drop in Chinese production alone. The loss in Southeast 0% P C P R A AM A W Asian production in 2015 is more than half a million tons. All other JA EA EC LA N AF O M N R –10% regions unaffected by the hypothetical shrimp disease outbreak are expected to respond to the decline in Asian shrimp aquacul- –20% ture by expanding their production initially in 2015. As seen in figure 4.3, each unaffected region would increase its shrimp aqua- –30% culture production by about 10 percent or more in 2015. However, –40% Aggregate regions Source: IMPACT model projections. 26 Note that, in figure 4.3, production reduction in affected regions is less than 35 percent. This is due to the fact that the 35 percent decline is im- posed though the exogenous growth rates of aquaculture, rather than production itself. Following the sharp reduction in shrimp supply driven with the vast majority of shrimp aquaculture occurring in Asia by the reduced exogenous growth rate values, the market would re- (projected to be 88 percent under the baseline scenario in 2015), spond by raising the world shrimp price, which would lead to increased shrimp supply in all parts of the world, including Asia. the ability of the unaffected regions to offset the Asian losses is A G R I C U LT U R E A N D R U R A L D E V E LO P M E N T D I S C U S S I O N PA P E R 62 C H A P T E R 4 — I M PA C T P R O J E C T I O N S TO 2030 U N D E R S E L E C T E D S C E N A R I O S limited. Collectively the unaffected regions are expected to in- disease outbreak into 2030, although the magnitude of the impacts crease shrimp aquaculture production only by 76 thousand tons in would be gradually reduced. The aquaculture production of shrimp 2015, to which Latin America alone contributes 69 thousand tons. in China and Southeast Asia in 2030 would still be lower than under As a result, in the year Asia is hypothetically hit by a major shrimp the baseline projection: by 2.5 percent and 5.8 percent, respectively. disease outbreak, the global reduction in shrimp supply would still On the other hand, in other affected regions (IND and SAR), produc- amount to 1.28 million tons. tion would recover closer to the baseline projections by 2020. The rate of recovery, specified as the exogenously given growth rate As the affected Asian countries recover from the shock of the hypo- of shrimp aquaculture, is specified uniformly across affected coun- thetical disease outbreak, the contributions of unaffected regions in tries so that the production is expected to recover, more or less, to filling the supply gap would be reduced. However, in all unaffected the projected levels under the baseline scenario by 2020. However, regions, positive impacts would still be felt into 2030. Shrimp aqua- the impact of the simulated setback in the expansion of aquacul- culture production in these regions would be higher than under the ture would vary by region. The opportunity cost of lost growth baseline scenario by 0.7 percent to 2.4 percent in 2030 (figure 4.3). during the disease period (2015–20) is relatively high for countries with high expected growth rates of aquaculture during the period, Table 4.6 shows the resulting shifts in the shrimp trade patterns. such as Thailand and China. From figure 4.3, by 2020 shrimp aqua- Note the figures in the table include shrimp of both capture and culture production in Southeast Asia would be about 10 percent aquaculture origins. The Asian regions affected by the hypotheti- below what would be without the outbreak, as represented by the cal disease outbreak are all net exporters of shrimp at the onset of baseline scenario. China’s production would be 6 percent below the disease outbreak. In 2015, all affected regions, except for India, the baseline projection by 2020. The two major players in global would reduce their shrimp exports. India, with its substantial sup- shrimp aquaculture would continue to feel the shock of the 2015 ply from capture shrimp fishery, is projected to increase exports, TABLE 4.6: Comparison of Projected Net Exports of Shrimp by Region with and without Disease Outbreak (000 tons) 2015 2020 2030 DISEASE DISEASE DISEASE BASELINE (SCENARIO 3) BASELINE (SCENARIO 3) BASELINE (SCENARIO 3) AFFECTED REGIONS  CHN 574 391 457 471 244 279 IND 97 115 115 136 151 164 SAR 122 91 137 141 162 165 SEA 1,156 830 1,401 1,201 1,848 1,697 UNAFFECTED REGIONS  NAM –1,328 –1,172 –1,560 –1,510 –2,043 –2,009 ECA –671 –553 –688 –652 –702 –682 LAC 533 648 636 672 853 877 JAP –398 –343 –386 –370 –365 –357 EAP –182 –147 –203 –192 –225 –219 MNA –7 10 –6 –1 –4 –1 AFR –28 –8 –36 –29 –51 –47 ROW 133 137 132 134 130 131 Source: IMPACT model projections. Note: ECA = Europe and Central Asia; NAM = North America; LAC = Latin America and Caribbean; CHN = China; JAP = Japan; EAP = other East Asia and the Pacific; SEA = Southeast Asia; IND = India; SAR = other South Asia; MNA = Middle East and North Africa; AFR = Sub-Saharan Africa; ROW = rest of the world. F I S H TO 2 030: P R O S P E C T S F O R F I S H E R I E S A N D A Q UA C U LT U R E C H A P T E R 4 — I M PA C T P R O J E C T I O N S TO 2 0 3 0 U N D E R S E L E C T E D S C E N A R I O S 63 capitalizing on the higher shrimp price while reducing domestic consequences of a large-scale seafood supply shock would be rel- consumption. evant for some other aquaculture commodities. Most of the unaffected regions are net importers of shrimp, ex- Finally, this particular example of a supply shock—a major out- cept for LAC and ROW. All of the net importers would be forced break of shrimp disease in Asian aquaculture—also underscores to reduce their imports in 2015 given the shortage in the global the importance of disease management in any major aquaculture shrimp supply. The reduction in imports varies across the regions, sector. As production intensifies and as fish population increases from 12 to 19 percent in NAM, JAP, ECA, and EAP to 73 percent in within a production system, both risks of disease outbreak and AFR. Higher-income regions, with their stronger income elasticities, the consequences of outbreak intensify in aquaculture (Arthur would likely manage to secure shrimp consumption better than and others 2002). Given the projection of continued expansion lower-income countries. The winners in this scenario would be LAC of aquaculture, disease shocks of the magnitude simulated in this and MNA. Traditionally a net exporter of shrimp, LAC would manage scenario can and are likely to occur. With seafood being one of the to increase their export thanks to the gain in aquaculture produc- most internationally traded food commodities, efforts to prevent tion in 2015. MNA would turn from a net importer to a net exporter major catastrophic disease outbreaks in aquaculture and, in such during the shock in the global shrimp market, even though they an event, to minimize negative impacts on the seafood market would again become a net importer by 2020. represent a global public good. The contrast between baseline and scenario trade values would be- come much less sharp in 2020 and then in 2030, as the production 4.4. SCENARIO 4: ACCELERATED SHIFT OF CONSUMER PREFERENCES IN CHINA trajectory in affected countries would more or less have already re- covered by then, as seen in figure 4.3. However, prolonged impacts In this scenario, we explore the implications of demand-side would remain especially for Southeast Asia, where the reduced changes on global fish markets. In particular, we investigate po- exports relative to the baseline case would be 200 thousand tons in tential impacts of shifts in consumer preferences for food fish in 2020 and 150 thousand tons in 2030, or a 14 percent and 8 percent China, the single most important country in the global seafood reduction, respectively. market. The earlier Fish to 2020 study also included a China-focused scenario, but it was motivated by the uncertainty over the veracity This scenario helps illustrate the global implications of a localized of the available data for China. They implemented the scenario to shock to fish supply. We have deliberately chosen a supply shock of examine the implications of having slower-than-reported produc- a moderate size, but in a region that controls the vast majority of the tion and consumption growth in the country. On the contrary, global supply of the affected species. As expected, the model pre- the focus in this analysis is on the implications of socioeconomic dicts adjustments in the aquaculture supply in unaffected regions. change—in particular, accelerated preference shifts toward high- However, if a supply shock hits a major production region, such value fish—in the mega fish consumption market. “swing” capacity to fill in the gap would be limited and the world Consumer preference shifts toward high-value fish are already being market would be hit hard. observed in China, driven mainly by demographic change, urbaniza- In an event of a sudden downturn in seafood supply, consumers tion, higher rates of education, and greater overall levels of income around the globe would have to adjust to the resulting higher in Chinese society (Fish Site 2012, Godfrey 2011, Redfern Associates shrimp price by cutting back on consumption. We have seen in 2010, World Bank 2013c). These trends are incorporated in the de- this simulation exercise that Sub-Saharan Africa would have to cut fault specification in terms of elasticities, particularly income elas- back 70 percent of their shrimp net imports in the initial year of the ticities, for different food fish commodities. This scenario simulates shock. While shrimp is a high-value commodity that is unlikely to faster food fish demand growth for high-value fish commodities in form a staple of food-insecure, vulnerable populations, food security China by changing these parameters. More specifically, we allow A G R I C U LT U R E A N D R U R A L D E V E LO P M E N T D I S C U S S I O N PA P E R 64 C H A P T E R 4 — I M PA C T P R O J E C T I O N S TO 2030 U N D E R S E L E C T E D S C E N A R I O S the per capita consumption of the higher-value products—namely TABLE 4.7: Projected Changes in the Food Fish Consumption in shrimp, crustaceans, salmon, and tuna—to increase three times China due to Accelerated Preference Shift higher compared to the baseline case in 2030. For medium-value SCENARIO 4 RELATIVE commodities, namely mollusks, we allow the per capita demand to 2030 (000 TONS) TO BASELINE double the baseline 2030 level. Demand growth for all other com- CHINA BASELINE (SCENARIO 4) modities is left unchanged at their baseline levels. TARGETED SPECIES Shrimp 4,183 13,021 211% In order to reflect these increases in food fish demand, the income Crustaceans 1,504 4,583 205% elasticities of demand for these commodities in China are set at a Salmon 971 2,876 196% level that realizes the target increases in per capita food fish de- Tuna 165 493 199% mand in 2030. Effectively, a higher income elasticity implies that Mollusks 17,695 36,201 105% people desire to consume proportionately more of a good for a NONTARGETED SPECIES given increase in their incomes. In this exercise we maintain the Freshwater and diadromous 25,833 26,616 3% Demersals 5,456 5,231 –4% levels of exogenous drivers of demand—that is, population and Pelagics 145 126 –13% income growth—at the baseline levels, so that we capture the Other marine 1,409 1,346 –4% pure effects of preference change. In other words, in this scenario, Total fish consumption 57,361 90,494 58% we simulate what would happen if consumer tastes reacted more Source: IMPACT model projections. strongly to income growth, rather than trying to simulate faster rates of socioeconomic growth itself. Adjustments are also made The relevant question in this scenario is: how would this massive to price elasticities of demand in order to maintain internal con- increase in Chinese consumption be made possible? Would the ac- sistency in the scaling of the food demand system and to achieve celerated growth in demand stimulate world aquaculture produc- the target increases in per capita food fish demand in 2030 even tion? Would trade patterns shift in such a way that it would impact in the face of higher commodity prices. To avoid complications food insecure regions, such as Sub-Saharan Africa? associated with modifying the demand system midway through the projections, the changes to these elasticities are implemented To shed light on these questions, we first examine the projected from the start of the model simulations in 2000. Thus, the entire aquaculture production in 2030 (table 4.8). Under this scenario, trajectory of consumption growth to 2030 is affected under this global aquaculture supply would increase by 22.6 million tons, 10.5 scenario. million tons less than the incremental demand by Chinese consum- ers. Aquaculture production would increase in all regions relative to Table 4.7 compares the model output for China’s total food fish the baseline scenario. However, the magnitude of increase would consumption under the baseline and current scenarios. By design, differ across regions. With relatively low aquaculture production consumption of the targeted species (shrimp, crustaceans, salmon, under the baseline, JAP and NAM are expected to increase their tuna, and mollusks) is substantially larger under the scenario in aquaculture production by the largest percentage (88 percent for 2030. We also note that there are changes (albeit much smaller) in JAP, due to expansion of mollusk production, and 77 percent in the consumption patterns of other nontargeted commodities, ow- NAM, due to expansion in production of shrimp, crustaceans, and ing to the fact that there is cross-price demand response between mollusks). EAP has the next highest growth (64 percent), followed fish categories. In particular, while consumption of freshwater and by ECA (43 percent), LAC (43 percent), and SEA (32 percent). China diadromous category would increase by 3 percent, consumption itself would increase production by 21 percent. Having no signifi- of other categories (pelagics, demersals, and other marine) would cant production base for high-value commodities, the other regions decline. Overall, fish consumption in China would increase by 58 (SAR, MNA, and AFR) are expected to grow only up to 10 percent percent, or by 33 million tons, under this scenario. relative to the baseline case. Note that the model may exaggerate F I S H TO 2 030: P R O S P E C T S F O R F I S H E R I E S A N D A Q UA C U LT U R E C H A P T E R 4 — I M PA C T P R O J E C T I O N S TO 2 0 3 0 U N D E R S E L E C T E D S C E N A R I O S 65 TABLE 4.8: Projected Aquaculture Production in 2030 under FIGURE 4.4: Projected Net Exports of Fish by Region under Accelerated Consumer Preference Shift in China Accelerated Consumer Preference Shift in China SCENARIO 4 Baseline China scenario 20,000 RELATIVE 2030 (000 TONS) TO BASELINE 15,000 CHINA BASELINE (SCENARIO 4) 10,000 Global total 93,612 116,205 24% Thousand tons 5,000 ECA 3,761 5,253 40% W AM A A R P N O EC AF JA M N R NAM 883 1,561 77% 0 D C P A R LAC 3,608 5,168 43% IN EA SE LA SA –5,000 EAP 1,066 1,747 64% CHN 53,264 64,502 21% –10,000 JAP 985 1,851 88% –15,000 SEA 14,848 19,626 32% SAR 4,163 4,500 8% –20,000 N H C IND 8,588 9,446 10% –25,000 Aggregate regions MNA 1,911 1,978 4% Source: IMPACT model projections. AFR 464 496 7% Note: ECA = Europe and Central Asia; NAM = North America; LAC = Latin America and Caribbean; CHN = China; JAP = Japan; EAP = other East Asia and the Pacific; ROW 72 76 7% SEA = Southeast Asia; IND = India; SAR = other South Asia; MNA = Middle East and North Africa; AFR = Sub-Saharan Africa; ROW = rest of the world. Source: IMPACT model projections. Note: ECA = Europe and Central Asia; NAM = North America; LAC = Latin America and Caribbean; CHN = China; JAP = Japan; EAP = other East Asia and the Pacific; SEA = Southeast Asia; IND = India; SAR = other South Asia; MNA = Middle East and North Africa; AFR = Sub-Saharan Africa; ROW = rest of default and the current scenarios. The figure shows that China’s net the world. trade position would shift dramatically in 2030 relative to the base- the favorable results generated for regions with an existing produc- line case: turning from a position of net exporter to that of sizable tion base of high-value commodities and the unfavorable results net importer of fish. In response to this, the exports by regions such for other regions. Since supply responses in the model are based as SEA and LAC would increase to supply this massive import de- on growth rates, a sector with positive initial value can grow or de- mand by China. Other regions, such as ECA, JAP, NAM, MNA, and AFR, cline during the projection period, while production in a sector with would decrease their net imports. Of these, reduced net imports by no initial value remains zero throughout the simulation. In reality, JAP and NAM are attributed to increased exports of high-value prod- new sectors can pop up in any country when business opportuni- ucts made possible by substantial production expansion (table 4.8). ties arise—a scenario that is ruled out by the model. Nonetheless, In contrast, in MNA and AFR, aquaculture would fail to ride the wave the results suggest the favorable position of countries that have of the demand boom and achieve only limited expansion (table 4.8), an existing production base in capturing the opportunities to ex- and reduction in the net imports could have food security implica- pand and diversify their production when a boom in a high-value, tions. The net imports in 2030 would be lower by 30 percent in MNA export-oriented sector, such as hypothesized in this scenario, would and by 14 percent in AFR relative to the baseline case. present itself. In fact, in all regions but China total food fish consumption in 2030 The drastic changes in global aquaculture production induced by would be lower under this scenario relative to the default case shifts in China’s consumer preferences would certainly impact global (figure 4.5). Food security concerns are already severe in AFR, where trade patterns across all regions and across all categories of fish. per capita food fish consumption is expected to decline during the Much of the changes are concentrated in the five specific species projection period. A further reduction in consumption by 5 percent, that were targeted in this scenario. Figure 4.4 illustrates the compari- triggered by expansion of consumption elsewhere, could aggravate son of projected total net regional exports of fish in 2030 under the the food fish supply gap in this region. A G R I C U LT U R E A N D R U R A L D E V E LO P M E N T D I S C U S S I O N PA P E R 66 C H A P T E R 4 — I M PA C T P R O J E C T I O N S TO 2030 U N D E R S E L E C T E D S C E N A R I O S FIGURE 4.5: Projected Change in Total Food Fish improvements and proper tenure reforms to reduce fishing effort, Consumption in 2030 by Region under letting the aquatic ecosystems and stocks recover, reducing the Accelerated Consumer Preference Shift in China open-access nature of fisheries, and sustainably managing their 70% productivity. Global fisheries restoration efforts will likely be accel- 60% 50% erated under several new key global initiatives, such as the Global 40% Partnership for Oceans. Percent change 30% 20% In scenario 5, we explore the implications of productivity recovery 10% W AM A A R A P R C P D N O EC EA SE SA AF LA JA IN M N R 0% in global capture fisheries on global fish markets. This scenario N H –10% contrasts to the “ecological collapse” scenario simulated in the Fish C –20% –30% to 2020 study, which aimed at reflecting the environmental conse- –40% quence of not taking on these actions and continuing to overhar- Aggregate regions vest and compromise aquatic ecosystems to the point where they Source: IMPACT model projections. Note: ECA = Europe and Central Asia; NAM = North America; LAC = Latin America undergo biological collapse. To simulate this scenario of improved and Caribbean; CHN = China; JAP = Japan; EAP = other East Asia and the Pacific; SEA = Southeast Asia; IND = India; SAR = other South Asia; MNA = Middle East and capture productivity, we make changes to the capture fisheries North Africa; AFR = Sub-Saharan Africa; ROW = rest of the world. growth parameters. More specifically, we augment the exogenous annual growth rates of capture production by 0.6 percentage points Finally, similarly to the case in scenario 1, where aquaculture pro- so that the total harvest would gradually increase starting from 2011 duction is assumed to expand faster than in the default scenario, and reach the global MSY estimated in The Sunken Billions in 2030. fast demand-driven aquaculture expansion in the current scenario The increase in the capture production growth rates is applied to would also place severe pressure on the global fishmeal market. all capture fisheries in the model, including inland fisheries. Note Overall, the model predicts that fishmeal supply would be higher that the 0.6 percentage points are added to the existing exogenous by 4 percent and the fishmeal price higher by 29 percent in 2030 growth rates. As a result, those capture fisheries modeled as declin- relative to the baseline case (results not shown in table). As a result, ing in the baseline specification would decline more slowly, recov- global aquaculture (and livestock) production would experience a ering or growing fisheries would grow faster, and stagnant fisheries secondary effect of a tighter fishmeal market as seen in section 4.1 would grow exactly at the annual rate of 0.6 percent under this sce- and fishmeal use would be further reallocated across aquaculture nario. The MSY value in The Sunken Billions is scaled for consistency species. with FAO harvest data employed in this study. 4.5. SCENARIO 5: IMPROVEMENT OF CAPTURE Figure 4.6 depicts the evolution of projected total fish supply on a FISHERIES PRODUCTIVITY global level under this scenario (represented by bars) together with The next two scenarios pertain to the productivity of capture fish- the baseline values (lines). The improvement in capture fisheries eries. In a recent World Bank study, The Sunken Billions (Arnason, productivity allows the global capture production level to reach Kelleher, and Willmann 2009), it was estimated that successfully more than 105 million tons by 2030, which represents a 13 percent restored and managed world fisheries would sustainably provide increase over the level under the baseline case. Under this scenario, 10 percent more yield annually relative to the 2004 harvest level. aquaculture still grows at an impressive rate over the projection pe- Restoring and improving the productivity of stressed capture fish- riod, but it does not quite reach the baseline 2030 level due to lower eries will be possible in many cases if correct actions are taken by market prices resulting from the additional supply from capture country governments, marine resource managers, and the fishing fisheries. Furthermore, under this scenario global capture fisheries fleets and communities. These actions would include management would supply 15 million tons more than aquaculture would in 2030, F I S H TO 2 030: P R O S P E C T S F O R F I S H E R I E S A N D A Q UA C U LT U R E C H A P T E R 4 — I M PA C T P R O J E C T I O N S TO 2 0 3 0 U N D E R S E L E C T E D S C E N A R I O S 67 FIGURE 4.6: Global Fish Supply under Improved Productivity, FIGURE 4.7: Projected Changes in Capture Production in 2000–30 2030 under Capture Productivity Improvement 200 Scenario Scenario capture Scenario aquaculture 180 Baseline capture Baseline total 40% 160 35% 140 30% Percent change Million tons 120 25% 100 20% 80 15% 60 10% 40 5% 20 0% 0 ECA NAM LAC EAP CHN JAP SEA SAR IND MNA AFR ROW 2000 2005 2010 2015 2020 2025 2030 Aggregrate regions Sources: FishStat and IMPACT model projections. Source: IMPACT model projections. Note: ECA = Europe and Central Asia; NAM = North America; LAC = Latin America and Caribbean; CHN = China; JAP = Japan; EAP = other East Asia and the Pacific; whereas capture and aquaculture production would contribute es- SEA = Southeast Asia; IND = India; SAR = other South Asia; MNA = Middle East and North Africa; AFR = Sub-Saharan Africa; ROW = rest of the world. sentially an equal amount to the global supply in 2030 under the baseline case. FIGURE 4.8: Projected Changes in Fish Consumption in 2030 under Capture Productivity Improvement Although this scenario uniformly adds 0.6 percent to the exog- Scenario enous growth rates of all capture fisheries in the model, the default 16% growth rates vary, with some fisheries growing and others declining 14% or stagnant. As a result, there would be differences in the rate at 12% which each region gains in terms of capture production. Figure 4.7 Percent change 10% attempts to illustrate the point. The figure confirms projected in- 8% creases in the production levels of capture production in all regions. 6% Japan would benefit the most under this scenario, achieving 34 per- 4% cent increase in capture production in 2030 relative to the baseline 2% scenario. Other regions would enjoy 11–15 percent increase in cap- 0% ECA NAM LAC EAP CHN JAP SEA SAR IND MNA AFR ROW ture production. SAR would benefit the least, achieving 4 percent –2% Aggregrate regions Source: IMPACT model projections. increase in their capture production in 2030 relative to the baseline Note: ECA = Europe and Central Asia; NAM = North America; LAC = Latin America and case. The last result is attributable to the relatively rapid growth that Caribbean; CHN = China; JAP = Japan; EAP = other East Asia and the Pacific; SEA = Southeast Asia; IND = India; SAR = other South Asia; MNA = Middle East and the capture fisheries of this region would be enjoying already under North Africa; AFR = Sub-Saharan Africa; ROW = rest of the world. the baseline scenario (0.4 percent per year between 2010 and 2030). except North America27 (figure 4.8). Again, there would be regional Thus an additional 0.6 percent in the growth rate under this scenario variations in the consumption gains from this scenario. The degree to would not bring about as dramatic an increase in the capture pro- which fish consumption increases would depend on how much of duction level. This contrasts sharply with the results for Japan, where the increased harvest would be consumed domestically rather than the production under the baseline scenario is expected to decline exported. Some countries gain more than 10 percent in domestic at the annual rate of 9 percent. 27 From the results, it is likely that the increased domestic capture produc- Increase in capture harvest would also imply increased food fish tion in North America would substitute imports rather than increasing consumption than under the baseline scenario across all regions overall domestic consumption. A G R I C U LT U R E A N D R U R A L D E V E LO P M E N T D I S C U S S I O N PA P E R 68 C H A P T E R 4 — I M PA C T P R O J E C T I O N S TO 2030 U N D E R S E L E C T E D S C E N A R I O S fish consumption. In particular, Sub-Saharan Africa would achieve parts of the world, we simulate the global market effects based consumption increase of about 13 percent, as much of the addi- on impacts of climate change on capture fisheries estimated else- tional harvest would be retained for consumption within the region, where. As an illustrative example of climate change impacts, we use possibly after some intraregional trade. In terms of per capita fish the predictions of catch potential provided by Cheung and others consumption, the region would achieve 6.4 kilograms, as opposed (2010). Incorporating the effects of various climate change symp- to 5.6 kilograms under the baseline case in 2030. toms in the oceans, Cheung and others (2010) provide maximum catch potential (maximum sustainable yield, MSY) in exclusive eco- These results demonstrate that a recovery of global capture fisher- nomic zone (EEZ) in 2055 under two scenarios: ies can have varied but potentially substantial impacts on regional seafood sectors and on food security. There are known difficulties a. Global atmospheric carbon dioxide content is kept con- stant at the level of year 2000. that are inherent in the attempt to implement comprehensive and b. Global atmospheric carbon dioxide content rises according coherent reform of capture fisheries. These reforms often require re- to an IPCC-based scenario out to the year 2100. gional cooperation as well as determination and political will in each nation. By illustrating the considerable gains that can be enjoyed at Scenario (a) may be interpreted as a case where mitigation mea- the regional level, these results are consistent and in support of the sures would be in place so that effectively “no” additional climate benefit of regional cooperation in fisheries reform. change would occur beyond the level in the year 2000. On the other hand, scenario (b) may be a case where, in the absence of radical mitigation measures, the “normal” progression of environmental ef- 4.6. SCENARIO 6: IMPACTS OF CLIMATE CHANGE fects would accumulate over time, including rising ocean tempera- ON THE PRODUCTIVITY OF CAPTURE FISHERIES ture and ocean acidification. The recent World Bank report Turn Down the Heat: Why a 4°C Warmer World Must Be Avoided (World Bank 2013b) describes the possible According to Cheung and others (2010), some countries would gain future state of the world in which the global temperature climbs 4 and others would lose in terms of catch potential due to climate degrees Celsius above preindustrial levels. The report states: “Even change–induced changes in the oceans. In particular, they provide, with the current mitigation commitments and pledges fully imple- for selected countries, how much the catch potential is projected to mented, there is roughly a 20 percent likelihood of exceeding 4°C change between 2005 and 2055 under the two climate change sce- by 2100. If they are not met, a warming of 4°C could occur as early narios. According to their results, in general, potential catch would as the 2060s.” Coastal communities will be among the first to be im- increase in high-latitude regions while catch would tend to drop in pacted by changes in the oceans due to climate change. Changes the tropics. These results would hold generally under both scenarios in the oceans will include a rise in the sea level, which directly af- but would be more prominent under scenario (b). fects habitability of the coastline, and rising water temperature and Based on these two sets of climate change scenarios, we simulate ocean acidification, which affect productivity of local fisheries and climate change impacts on global fish markets. There are three spe- the health of marine life and ecosystems. The impact of climate cific assumptions of the scenario. First, the percent-change values change on capture fisheries is a topic that has been closely studied for catch potential provided for the selected countries in Cheung by a number of groups in public and private sectors, academia, and and others (2010) are extrapolated to other countries represented in civil society organizations. Here we wish to contribute to the grow- the IMPACT model based on geographical proximity. Second, while ing, but still limited, body of knowledge by carrying out some illus- their projection period is from 2005 to 2055, our projection period trative simulations of impacts of climate change on capture fisheries is to 2030. Thus, we truncate their projections by cutting their per- and global fish markets. cent changes in half. Third, we modify the exogenous growth rates Rather than using an ecosystem model that can directly provide of capture fisheries for the 2006–30 period based on the extrapo- estimates of productivity changes for capture fisheries in various lated and truncated percent-change values. We apply the modified F I S H TO 2 030: P R O S P E C T S F O R F I S H E R I E S A N D A Q UA C U LT U R E C H A P T E R 4 — I M PA C T P R O J E C T I O N S TO 2 0 3 0 U N D E R S E L E C T E D S C E N A R I O S 69 TABLE 4.9: Projected Capture Production in 2030 under TABLE 4.10: Projected Fish Supply in 2030 under Baseline and Baseline and Climate Change Scenarios Climate Change Scenarios CC-a RELATIVE CC-b RELATIVE CC-a RELATIVE CC-b RELATIVE 2030 (000 TONS) TO BASELINE TO CC-a 2030 (MILLION TONS) TO BASELINE TO CC-a BASELINE CC-a CC-b BASELINE CC-a CC-b Global 93,229 90,217 90,200 –3% –0.02% Capture 93.23 90.22 90.20 –3% –0.02% total Aquaculture 93.61 94.70 94.79 1% 0.1% ECA 12,035 12,876 13,771 7% 7% Global total 186.84 184.92 184.99 –1% 0.04% NAM 5,589 5,467 5,433 –2% –1% Source: IMPACT model projections. LAC 18,221 17,285 17,240 –5% –0.3% Note: CC-a = climate change with mitigation; CC-b = climate change without drastic mitigation. EAP 2,890 3,044 2,938 5% –3% CHN 15,686 15,824 15,582 1% –2% lower projection for 2030 capture production than in our baseline JAP 3,717 4,426 4,338 19% –2% scenario. At the global level, however, CC-a scenario’s prediction SEA 14,244 12,710 12,228 –11% –4% for capture production in 2030 is only 3 percent lower than our SAR 5,811 3,994 4,038 –31% 1% baseline scenario. With this in mind, we shall compare the results of IND 4,143 3,821 3,826 –8% 0.1% two sets of climate change scenarios. Comparing between results MNA 2,769 2,555 2,524 –8% –1% under scenarios CC-a and CC-b (the simulation based on Cheung AFR 5,472 5,330 5,394 –3% 1% and others’ (2010) scenario (b)), the model predicts that the overall ROW 2,652 2,884 2,888 9% 0.1% Source: IMPACT model projections. effect of climate change on global capture production is negligible Note: CC-a = climate change with mitigation; CC-b = climate change without (a 0.02 percent decrease). Across regions, however, the projected drastic mitigation; ECA = Europe and Central Asia; NAM = North America; LAC = Latin America and Caribbean; CHN = China; JAP = Japan; EAP = other East impact of climate change on production varies. Cheung and others Asia and the Pacific; SEA = Southeast Asia; IND = India; SAR = other South Asia; MNA = Middle East and North Africa; AFR = Sub-Saharan Africa; ROW = rest of (2010) suggest that the catch potential of high-latitude European the world. nations would increase under scenario b), which is reflected in the exogenous capture growth rates to all species and to all countries results under scenario CC-b. Capture production in ECA region that have marine capture fisheries in the model. No change is made would be 7 percent greater under CC-b than under CC-a. SAR to the capture growth rates for the 2000–05 period. and AFR would also gain capture harvest by 1 percent. The largest negative impacts would be observed in SEA (–4 percent) and EAP Though various studies have shown the likely negative effect of (–3 percent). climate change on fish production and consumption in small island countries like Pacific island countries and territories (see, for ex- While the shock is introduced to capture fisheries in this scenario, ample, Bell, Johnson, and Hobday 2011), our model structure does the ultimate interest to society is the impact of climate change on not allow us to implement this climate change scenario for small the total fish supply. Table 4.10 shows the projected fish supply from island countries. As mentioned earlier, small island countries have capture and aquaculture origins and their total. Comparing the been aggregated in the ROW category. We also did not explicitly baseline to CC-a scenarios, global aquaculture production would differentiated between various types of capture fisheries, such as increase by only 1 percent in 2030, picking up only 39 percent of oceanic fisheries, costal fisheries, and inland fisheries. the loss in capture fisheries. As a result, the total fish supply would be lower under CC-a than under the baseline scenario. Again, the Table 4.9 lists the projected regional capture production in 2030 change between CC-a and CC-b is small for both aquaculture and under the baseline and the two climate change scenarios. First, global total fish supply. the simulation based on Cheung and others’ (2010) scenario (a) with mitigation (CC-a, hereafter) generates a gloomier picture of Finally, table 4.11 shows the projections of regional food fish consump- global capture fisheries than our baseline simulation. In regions tion. In CC-a, the impacts on regional capture fisheries are attenuated other than ECA, EAP, CHN, JAP, and ROW, CC-a yields 2–31 percent in regional consumption patterns due to aquaculture expansion and A G R I C U LT U R E A N D R U R A L D E V E LO P M E N T D I S C U S S I O N PA P E R 70 C H A P T E R 4 — I M PA C T P R O J E C T I O N S TO 2030 U N D E R S E L E C T E D S C E N A R I O S TABLE 4.11: Projected Food Fish Consumption in 2030 under The apparently insignificant impacts of climate change on global Baseline and Climate Change Scenarios fish markets projected in this section are due in part to the shorter CC-a RELATIVE CC-b RELATIVE time horizon considered in this report. While the projected changes 2030 (000 TONS) TO BASELINE TO CC-a in global marine capture fisheries by Cheung and others (2010) BASELINE CC-a CC-b Global 151,771 149,851 149,915 –1% 0.04% continue beyond 2030 and into 2055, we have truncated such total changes by half. Climate change is an ongoing process whose ECA 16,735 16,445 16,490 –2% 0.3% impacts would materialize decades, even centuries, later. Readers NAM 10,674 10,670 10,680 –0.04% 0.1% should be reminded that the results presented in this report rep- LAC 5,200 5,089 5,087 –2% –0.04% resent medium-term projections into 2030 and do not represent EAP 2,943 2,894 2,896 –2% 0.1% long-term impacts of climate change. Nonetheless, already by 2030, CHN 57,361 56,977 57,029 –1% 0.1% JAP 7,447 7,422 7,410 –0.3% –0.2% climate change will likely affect global fish markets in the form of SEA 19,327 18,935 18,917 –2% –0.1% distributional changes in the global marine fish harvest and result- SAR 9,331 9,070 9,040 –3% –0.3% ing trade patterns. IND 10,054 9,966 9,972 –1% 0.1% MNA 4,730 4,620 4,630 –2% 0.2% Besides these six, a number of other relevant and practical scenarios AFR 7,759 7,562 7,566 –3% 0.1% can also be implemented. For example, a scenario where small pe- ROW 208 200 198 –4% –1% lagic fisheries collapse to affect reduction industry (fishmeal and Source: IMPACT model projections. fish oil production) can be implemented to analyze the impact on Note: CC-a = climate change with mitigation; CC-b = climate change without drastic mitigation; ECA = Europe and Central Asia; NAM = North America; LAC = aquaculture production that is dependent on fishmeal and fish oil Latin America and Caribbean; CHN = China; JAP = Japan; EAP = other East Asia and the Pacific; SEA = Southeast Asia; IND = India; SAR = other South Asia; MNA = as input. Climate change also affects aquaculture production (see, Middle East and North Africa; AFR = Sub-Saharan Africa; ROW = rest of the world. for example, FAO 2009), and impacts of climate change on both international trade. The changes in the oceans due to the climate capture fisheries and aquaculture and their interactions on global change scenarios based on Cheung and others (2010) would have no fish markets can be analyzed using the model. These and other sce- major impacts on projected food fish consumption into 2030. narios may be studied in another volume. F I S H TO 2 030: P R O S P E C T S F O R F I S H E R I E S A N D A Q UA C U LT U R E CHAPTER 5 — DISCUSSION 71 Chapter 5: DISCUSSION 5.1. MAIN FINDINGS FROM THE ANALYSIS fish-based feeds is projected to grow at a steady pace, although Given the discussion of the baseline and scenario results to 2030, nowhere near as quickly as the overall trajectory of aquaculture we now synthesize some of the key messages that emerge from production that unfolds to 2030. Much faster growth in aquaculture our analysis of the global and regional supply, demand, and trade production than in fishmeal supply toward 2030 reflects our assump- of fish. First, the remarkably dynamic character of the aquaculture tion that there will continue to be steady improvements in the feed sector stands out in the results and underscores the tremendous and feeding efficiency within the aquaculture sector. While such contribution that will likely be made by the Asia region, in particu- technological improvements are exogenously imposed in the model, lar, in meeting the growing world seafood demand in the next 20 rather than endogenously determined, this trend is reflected in the years. The growth of global aquaculture production is projected to historical data. Dissemination of best management practices will likely continue at a strong pace, until it matches the production of cap- continue from more advanced regions—for example, Scandinavia ture fisheries by the year 2030. China will likely still be at the top of for salmon aquaculture—throughout the industry. This will likely the producer list of major fish species, but other Asian countries/ come about as a result of competition for quality, the pressures of regions (particularly SEA, IND, and SAR) will likely become stronger sustainability certifications, and the purely economic drive to lower contributors of future aquaculture growth. Species such as shrimp, costs per unit production of output as much as possible. The fishmeal salmon, tilapia, Pangasius, and carp are seen to grow fairly rapidly price that is projected to steadily rise over the projection horizon to over the projection horizon, with projected annual average growth 2030 represents a continuing imperative and driver for technological rates well in excess of 2 percent a year over the 2010–30 period. This change and efficiency gains. This is also reflected, in our results, in represents a strong contrast with the stagnant nature of capture, terms of the increasing use of fish processing waste for reduction into whose growth in the model is kept at exogenous rates that reflect fishmeal and fish oil. Use of such “free” feedstock will likely increase, its historical patterns of production. especially as fishery operations become more consolidated and ver- tically integrated within the industry. All in all, it is likely that techno- Even though our classification of fish categories is more aggregated logical change surrounding feed production and feeding practices on the consumption side, we are able to obtain a much clearer will enable aquaculture to become more efficient and sustainable in picture of trends in demand across species compared to the ear- nature as we move into the medium- to long-term horizon. lier Fish to 2020 analysis. The model projects strong growth in the global consumption of shrimp, although the largest overall share of In our 2030 projections, the prices of all fish products continue on consumption is in the freshwater and diadromous category, which a slightly increasing trajectory into the future, which is consistent encompasses such fast-growing categories as tilapia, Pangasius, and with what we observe in other global food commodity markets and carp. projections. This reflects the “tightness” of market conditions that we expect to prevail across a number of food categories, given the Other pelagic will likely continue to be the most important cat- steady demand growth for food and feed products that is expected egory for use in producing fishmeal and fish oil. The supply of these to continue on a global level toward 2030 and beyond. This in turn A G R I C U LT U R E A N D R U R A L D E V E LO P M E N T D I S C U S S I O N PA P E R 72 CHAPTER 5 — DISCUSSION is driven by steady growth in emerging economies like China, India, across regions. The effects of restoring global capture fisheries will and the faster-growing countries within the Latin America and Sub- be positive in all aspects—both in production and consumption Saharan Africa regions. across all regions. This contrasts with the climate change scenario, where some regions gain in productivity due to more favorable As was the case in the earlier Fish to 2020 study, a good deal of biophysical conditions than other regions. This result underscores production and consumption growth and volume is expected to the importance of understanding and properly measuring the continue to be centered in China. The projections also highlight a region-specific environmental changes, including ocean tem- number of Southeast Asian countries increasing their aquaculture perature and acidity levels that would occur under various climate supply in order to meet the growing food fish demand of regions change projections, and their implications on regional fisheries. The such as China as well as their own internal consumption needs. The fact that there remains a large degree of uncertainty and incon- model predicts that the fishmeal to fuel the future growth of Asian sistency across various climate model projections in terms of the aquaculture will largely be imported from Latin America, which will degree and direction of these effects underscores the importance likely continue to produce a surplus of feed for both fish and live- of resolving these issues before meaningful interpretations can stock production. be given to their simulated effects on capture fisheries and, more broadly, on global fish markets. This also shows the growing need to One of the illustrative scenarios that we explore in this study points manage fisheries in a precautionary fashion, given all the unknowns toward important sources of feed for aquaculture: fish processing surrounding climate change. waste that can be reduced, along with whole fish, into fishmeal and fish oil. The projected sizable increase in fishmeal production A scenario on the demand side illustrates the effects of accelerated and reduction in world prices suggests this as a promising source food preference changes among consumers in China. Substantially of change within the industry. Increased use of processing waste increased demand for high- and medium-value fish species within will likely help relieve the pressure on marine stocks in the future China would shift the global fish trade. The trade position of China as well. This kind of development could be one of the factors that for these products would shift from either small net exporter or leads toward the “faster growth” scenario simulated for aquaculture large net importer to a solid net exporter by 2030. As a result fish worldwide, in which a greater pressure on fishmeal markets (in the consumption would be reduced in every other region by 1 to 32 absence of any other technological change) and higher prices for percent. fishmeal, fish oil, and their ingredient fish species are projected. Each of these scenarios helps to illustrate important dimensions The disease scenario for shrimp aquaculture in Asia shows, by de- of the world fish economy that might be substantially affected by sign, a very dramatic initial decline in shrimp production, but the technical, environmental, or socioeconomic change, and how sensi- shocks even out toward 2030 as the industry recovers. The shift in tive the model results are to those shifts. They also help point to trade patterns would be the main mechanism through which the areas that should receive closer attention in future work—such as Asian production loss is made up, by additional production from sourcing of alternative feedstock for fishmeal—and which countries other regions, especially Latin America. However, given the size of and regions could be significant game changers in the industry as it the aggregate shock to the market, Latin America would not be able evolves over the next 20 years. to fill the initial global supply gap. Given the projected tightness of shrimp markets due to the continuing growth in global demand, such large-scale disease outbreaks will likely cause substantial im- 5.2. DISCUSSION AND POSSIBLE FUTURE DIRECTIONS pacts on the global market. The series of comparative analyses presented in section 2.5 and The scenarios focusing on capture fisheries show potentially in the technical appendix to chapter 3 confirm that the IMPACT dramatic changes in production and distribution of production model generates projections that are consistent with the available F I S H TO 2 030: P R O S P E C T S F O R F I S H E R I E S A N D A Q UA C U LT U R E CHAPTER 5 — DISCUSSION 73 data during the 2000s and with the model projections provided be incorporated into the model by fine-tuning the growth rate pa- by OECD-FAO (2012). This gives us a degree of confidence in the rameters. Alternatively, capacity constraints may be explicitly speci- model’s ability to reproduce the observed dynamics of the global fied within the model. fish markets. However, the model shows some difficulty in generat- ing consistent price projections, despite its success in reproducing Capture Growth Rates their supply, demand, and trade volumes. In particular, for shrimp, Similarly, the representation of the dynamics of capture fisheries crustaceans, and other freshwater and diadromous categories, the can be improved. In this study, the harvest of capture fisheries is model does not reproduce the sharply falling price trend over the treated as entirely exogenous, and their trends are determined 2000–08 period. However, we are confident that the model has using the aggregate statistics. Use of fishery-specific information captured reasonably well the features of each of these markets at about capacity and trends will likely improve the model predictions the regional and global level. As is often the case with a large and and expand the scope of analysis. While the state of world fisheries detailed model, this model contains a necessary degree of simpli- are periodically assessed and reported by the FAO (see, for example, fication so that the model is consistently clear in the construction FAO 2012), many fisheries, especially those in developing countries, and framework, tractable in its computational properties, and easily are not currently assessed. However, tools are available to infer modifiable to introduce exploratory scenarios. Although there may the state of unassessed fisheries. For example, Costello and others be some features specific to certain segments of the fish markets (2012) provide a regression-based predictive model of the state of that have not been fully incorporated in our representation of fish unassessed fisheries. While efforts at the University of Washington supply and demand relationships, we feel that this has not compro- aim at quantifying the variability in management systems around mised the overall fidelity of the model to its original goal of repre- the world to evaluate which particular attributes lead to more senting the basic drivers of change and capturing the dynamics in successful outcomes for fish populations and fisheries (see, for ex- the global fish markets to 2030. ample, Melnychuk, Banobi, and Hilborn 2013), recently developed Fishery Performance Indicators (FPI) can be used for rapid assess- Nonetheless, there are areas where improvements can be made on ment of fisheries that are not formally assessed (Chu, Anderson, and the model representation, especially when additional and better Anderson 2012). Interacting the IMPACT model with an ecological data become available. simulation model may be an alternative way to characterize the dynamics of capture fisheries. Aquaculture Growth Rates One set of parameters that can be improved is exogenous growth Trends in Consumption Demand rates of aquaculture production. Except for those related to feed Consumption trend is another area where country- or region- and price-driven supply responses, the dynamics of aquaculture specific parameters could be improved. In this study, the dynam- are approximated by their historical trends observed in aggregate ics of demand for food by individual consumers is driven solely by statistics. Given the sheer number of country-fish species combina- growth in per capita income, and resulting per capita consumption tions represented in the IMPACT model, case-by-case investigation demand is scaled up for each country or region by the total popula- of aquaculture expansion potential is not conducted. For example, tion, whose trend is exogenously given. All other changes in con- each country faces species-specific and overall capacity constraints sumption in the model are purely price driven, which is regulated for aquaculture that are driven by geographical characteristics, such by price and income elasticities of demand that are fixed through- as soils, topography, climate, and water availability (Pillay and Kutty out the projection horizon. However, factors such as affluence 2005; Boyd, Li, and Brummett 2012). Without sufficiently controlling and urbanization as well as concerns for health, environment, and the culture environment (for example, putting fish in greenhouses), other ethical and social values are considered to affect consumer aquaculture of certain species is simply infeasible in certain coun- preferences (FAO 2012) and, hence, own- and cross-price elastici- tries. If country-level information is compiled, that information can ties and income elasticities of demand. Where available, country- or A G R I C U LT U R E A N D R U R A L D E V E LO P M E N T D I S C U S S I O N PA P E R 74 CHAPTER 5 — DISCUSSION region-specific updates in elasticity estimates could be incorpo- climate change, or a combination of ecological factors. While rated to reflect observed trends and shifts in consumer preferences. Fish to 2020 included such a scenario, it was not under- pinned by this kind of detailed biophysical modeling. Linking with Ecosystem Model Consumption and Trade Data As we have pointed out in earlier discussions, there could have At the time of model preparation, production data from the FishStat been greater detail brought to the analysis of capture fisheries if an database were available through 2009, whereas consumption and ecosystems-based framework had been used to replace our simple trade data from FAO FIPS FBS were available only up to 2007. Lack treatment of the sector in the IMPACT model. In earlier conceptions of more recent data for the consumption/trade side has posed a of the Fish to 2030 study, it was envisioned that the IMPACT model limitation in the calibration exercises for these series. could be linked to a marine ecosystems model such as those within the Ecopath with Ecosim (EwE) family of models (Christiansen More importantly, the use of FAO FIPS FBS data determines the level and Walters 2004a, 2004b). The EcoOcean model (Alder and oth- of disaggregation of fish species, which limits the scope of analysis ers 2007), for example, is built on the EwE modeling platform and in this study. For example, the level of species aggregation dictated provides global-level projections of capture production potential by the FAO FIPS FBS data precludes the analyses of most dynamic across all of the FAO fishing regions. This model could potentially be fish markets, such as for tilapia and Pangasius, mainly due to limita- linked to IMPACT. Since the EcoOcean model takes into account the tion of trade raw data availability for these species. Furthermore, the quantity and value of marine fishery landings, the effort required, FAO FIPS FBS data are geared toward understanding the supply- and the ability of the ecosystem to regenerate itself, it would have consumption balances of food commodities; for that reason, trade provided a much more dynamic feedback to the market-driven de- is expressed in terms of “net export.” However, fish is traded heavily mand for capture species in IMPACT—especially as it relates to the and extensively both in the form of fish and processed seafood. demand for whole fish that are reduced for fishmeal for the aqua- Allowing the model to represent fish trade for processing purposes culture sector. Since it was designed to project ecosystem impacts would require substantial changes in the code, and obtaining con- of longer-term environmental change, such as climate change, it sistent series of trade and consumption for fish and seafood would could have been used to extend the projections to a longer horizon require a more complicated data preparation. But it may be a pos- (for example, to 2050). sible direction to go in to improve the model representation. Linking with an ecosystem model could allow us to address a much Overall Data Quality wider range of policy-relevant questions by combining questions We have tried as much as possible to maintain a close match of trade and agricultural policy with questions of environmental between the model projections and the existing data for the management and technology adoption in fisheries. There would calibration period. However, we have to recognize that complete be considerable value-added created by developing a robust link congruence was not possible as the development of the fish part between a market-based, global food supply and demand model, of IMPACT relies on three different datasets that are not mutually such as IMPACT, and ecological process models for fisheries, such as consistent. As a result, much time has been spent identifying the EcoOcean. Such a link enables the study to sources of discrepancy and reconciling them to reconstruct a plau- a) examine more closely the effects of climate change–in- sible and consistent picture of the global fish markets. For example, duced changes in aquatic ecosystems and ocean condi- there are many cases in which positive fishmeal production is tions, the resulting impact on the productivity of marine observed (based on IFFO data) in countries where there is no re- ecosystems, and how that affects global fish market corded “reduction” of whole fish (in FAO data), and the converse. To dynamics; and a lesser degree, discrepancies are also identified between the more b) gain a better sense of how global supply and demand for fish products would be impacted by the ecological collapse of detailed production data from the FAO’s FishStat database and the certain fisheries due to various causes, including overfishing, more aggregated consumption and trade data from the FAO FIPS F I S H TO 2 030: P R O S P E C T S F O R F I S H E R I E S A N D A Q UA C U LT U R E CHAPTER 5 — DISCUSSION 75 FBS database. It is desirable that these two important datasets are Undoubtedly, this would be a desirable goal for anyone working updated in tandem so that they can be more readily used together. on fish supply and demand studies within the FAO and within the Unfortunately, at present, it is not technically possible because data- wider research community. Investment in consistent fisheries data sets are updated sequentially starting with production series, then seems especially sensible given the parallel modeling work that the trade series, and then consumption series. Furthermore, as many FAO undertakes in collaboration with the OECD (OECD-FAO 2012; countries do not report trade data that are as highly disaggregated see technical appendix to chapter 3). In many ways, our effort in this by species as production data, trade and consumption series are study to reconcile across the datasets is, by itself, a valuable contri- necessarily more aggregated. The FAO has initiated an effort to en- bution toward that goal and will hopefully form a basis for further courage member countries to report disaggregated fish trade data. work within the FAO and its partner organizations. A G R I C U LT U R E A N D R U R A L D E V E LO P M E N T D I S C U S S I O N PA P E R REFERENCES 77 REFERENCES ADB (Asian Development Bank). 2005. An Impact Evaluation Study of Bell, J. D., J. E. Johnson, and A. J. Hobday, eds. 2011. Vulnerability of Tropical the Development of Genetically Improved Farmed Tilapia and Their Pacific Fisheries and Aquaculture to Climate Change. Secretariat of the Dissemination in Selected Countries. Manila, Philippines: ADB. Pacific Community, Noumea, New Caledonia. Adenäuer, M. 2008. “CAPRI versus AGLINK-COSIMO: Two Partial Blanco-Fonseca, M. 2010. “Literature Review of Methodologies to Equilibrium Models—Two Baseline Approaches.” Paper presented Generate Baselines for Agriculture and Land Use.” Deliverable 4.1 at the 12th Congress of the European Association of Agricultural of FP7 Project “Common Agricultural Policy Regional Impact: The Economists, Ghent, Belgium, August 26–29. http://ageconsearch. Rural Development Dimension,” Joint Research Center, Institute for umn.edu/bitstream/44120/2/268.pdf (accessed August 12, 2013). Prospective Technological Studies (JRC-IPTS), Seville, Spain. http:// www.ilr.uni-bonn.de/agpo/rsrch/capri-rd/docs/d4.1.pdf (accessed July Alder, J., S. Guénette, J. Beblow, W. Cheung, and V. Christensen. 2007. 14, 2013). Ecosystem-Based Global Fishing Policy Scenarios. Fisheries Centre Research Reports, vol. 15, no. 7. University of British Columbia, Bondad-Reantaso, M. G., R. P. Subasinghe, J. R. Arthur, K. Ogawa, S. Vancouver: Fisheries Centre. Chinabut, R. Adlard, Z. Tan, and M. Shariff. 2005. “Disease and Health Management in Asian Aquaculture.” Veterinary Parasitology 132: Anderson, J. L. 2012. “Drivers of Change in Fisheries and Aquaculture: Is 249–72. There a Role of the World Bank?” Presentation made at the 6th World Fisheries Congress, Edinburgh, U.K., May 7–11. Boyd, C., L. Li, and R. E. Brummett. 2012. “Relationship of Freshwater Anderson, J. L. 2003. The International Seafood Trade. Cambridge, U.K.: Aquaculture Production to Renewable Freshwater Resources.” Journal Woodhead Publishing Ltd. of Applied Aquaculture 24 (2): 99–106. Argue, B. J., Z. Liu, and R. A. Dunham. 2003. “Dress-Out and Fillet Yields Britz, W., ed. 2005. “CAPRI Modelling System Documentation: Common of Channel Catfish, Ictalurus punctatus, Blue Catfish, Ictalurus furcatus, Agricultural Policy Regional Impact Analysis.” University of Bonn. and Their F1, F2 and Backcross Hybrids.” Aquaculture 228 (1–4): 81–90. http://www.ilr.uni-bonn.de/agpo/rsrch/capri/capri-documentation .pdf (accessed August 12, 2013). Armington, P. S. 1969. “A Theory of Demand for Products Distinguished by Place of Production.” IMF Staff Papers 16 (1): 159–78. Brown, C. J., E. A. Fulton, A. J. Hobday, R. J. Matear, H. P. Possingham, C. Bulman, V. Christensen and others. 2010. “Effects of Climate-Driven Arnason, R., K. Kelleher, and R. Willmann. 2009. The Sunken Billions: The Primary Production Change on Marine Food Webs: Implications Economic Justification for Fisheries Reform. Washington, DC: World for Fisheries and Conservation.” Global Change Biology 16 (4): Bank. 1194–212. Arthur, J. R., M. J. Phillips, R. P. Subasinghe, M. B. Reantaso, and I. H. Brummett, R. E. 2003. “Aquaculture and Society in the New Millennium.” MacRae, eds. 2002. Primary Aquatic Animal Health Care in Rural, Small- World Aquaculture 34 (1): 51–59. Scale, Aquaculture Development. Fisheries Technical Paper 406. Rome: Food and Agriculture Organization of the UN. Brummett, R. E. 2007. “Indigenous Species for African Aquaculture Development.” In Ecological and Genetic Implications of Aquaculture Asche, F. 2011. “Green Growth in Fisheries and Aquaculture Production and Activities, edited by T. Bert, 229–45. Dordrecht, Netherlands: Trade.” http://www.oecd.org/tad/sustainable-agriculture/48258799 Springer. .pdf. Brummett, R. E. 2013. “Growing Aquaculture in Sustainable Ecosystems.” Asche, F., and K. H. Roll. 2013. “Determinants of Inefficiency in Norwegian Agricultural and Environmental Services Department Notes (June), Salmon Aquaculture.” Aquaculture Economics and Management 17 (3): World Bank, Washington, DC. http://www-wds.worldbank. org/external 300–321. /default/WDSContentServer/WDSP/IB/2013/06/26/000442464 Ayyappan, S. 2012. “Indian Fisheries on a Fast Tract.” The Economic Times _20130626121301/Rendered/PDF/788230BRI0AES00without0the0a (New Delhi, India), September 28. bstract.pdf. Barnes, M. E., M. L. Brown, K. A. Rosentrater, and J. R. Sewell. 2012. “An Initial Cheung, W. W. L., V. W. Y. Lam, J. L. Sarmiento, K. Kearney, R. Watson, D. Investigation Replacing Fish Meal with a Commercial Fermented Zeller, and D. Pauly. 2010. “Large-Scale Redistribution of Maximum Soybean Meal Product in the Diets of Juvenile Rainbow Trout. Open Fisheries Catch Potential in the Global Ocean under Climate Change.” Journal of Animal Sciences 2 (4): 234–43. doi:10.4236/ojas.2012.24033. Global Change Biology 16 (1): 24–35. A G R I C U LT U R E A N D R U R A L D E V E LO P M E N T D I S C U S S I O N PA P E R 78 REFERENCES Chim, L., and T. Pickering. 2012. “Feed for Aquaculture.” Presentation at FAO. 2013. Food Outlook: Biannual Report on Global Food Markets (June). the SPC/IFREMER seminar “Fish Waste Utilization,” Noumea, New Rome: FAO. http://www.fao.org/docrep/018/al999e/al999e.pdf. Caledonia, June 11. Fish Site. 2012. “Demand for High-end US Seafood in South China Christensen, V., and C. J. Walters. 2004a. “Ecopath with Ecosim: Methods, Remains.” News article from the Fish Site, December 19. http://www Capabilities and Limitations.” Ecological Modelling 172 (2–4): 109–39. .thefishsite.com/fishnews/19045/demand-for-highend-us-seafood- in-south-china-remains (accessed March 4, 2013). Christensen, V., and C. J. Walters. 2004b. “Trade-Offs in Ecosystem-Scale Optimization of Fisheries Management Policies.” Bulletin of Marine Garduño-Lugo, M., I. Granados-Alvarez, M. A. Olvera-Novoa, and G. Muñoz- Science 74 (3): 549–62. Córdova. 2003. “Comparison of Growth, Fillet Yield and Proximate Composition between Stirling Nile Tilapia (Wild Type) (Oreochromis Chu, J., J. L. Anderson, and C. M. Anderson. 2012. Evaluation of New Fishery niloticus, Linnaeus) and Red Hybrid Tilapia (Florida red tilapia×Stirling Performance Indicators (FPIs): A Case Study of the Blue Swimming Crab red O. niloticus) Males.” Aquaculture Research 34 (12): 1023–28. Fisheries in Indonesia and Philippines. Washington, DC: World Bank. Gjedrem, T., and M. Baranski. 2009. Selective Breeding in Aquaculture: An Clement, S., and R. T. Lovell. 1994. “Comparison of Processing Yield and Introduction. Dordrecht, Netherlands: Springer. Nutrient Composition of Cultured Nile Tilapia (Oreochromis niloticus) and Channel Catfish (Ictalurus punctatus).” Aquaculture 119 (2–3): Gjedrem, T., N. Robinson, and M. Rye. 2012. “The Importance of Selective 299–310. Breeding in Aquaculture to Meet Future Demands for Animal Protein: A Review.” Aquaculture 350–353 (June): 117–29. Costello, C., D. Ovando, R. Hilborn, S. D. Gaines, O. Deschenes, and S. E. Lester. 2012. “Status and Solutions for the World’s Unassessed Globefish. 2011. “Commodity Update: Fishmeal and Fish Oil.” FAO, Rome. Fisheries.” Science 338 (6106): 517–20. Godfrey, M. 2011. “Change Is Afoot in China.” News article in Seafood Crapo, C., B. Paust, and J. Babbitt. 2004. Recoveries and Yields from Pacific News Supply and Trade, May 31. http://www.seafoodsource.com Fish and Shellfish. Marine Advisory Bulletin 37. University of Alaska /newsarticledetail.aspx?id=10414 (accessed March 4, 2013). Fairbanks: Alaska Sea Grant College Program. Gutiérrez, N. L., R. Hilborn, and O. Defeo. 2011. “Leadership, Social Capital Delgado, C. L., N. Wada, M. W. Rosegrant, S. Meijer, and M. Ahmed. and Incentives Promote Successful Fisheries.” Nature 470: 386–89. 2003. Fish to 2020: Supply and Demand in Changing Global Markets. WorldFish Center Technical Report 62. Washington, DC: International Hansen, H., and Y. Onozaka. 2011. “When Diseases Hit Aquaculture: An Food Policy Research Institute. Experimental Study of Spillover Effects from Negative Publicity.” Marine Resource Economics 26 (4): 281–92. Dey, M. M. 2000. “The Impact of Genetically Improved Farmed Nile Tilapia in Asia.” Aquaculture Economics and Management 4 (1–2): 109–26. Hardy, R. W. 1996. “Alternate Protein Sources for Salmon and Trout Diets.” Animal Feed Science and Technology 59 (1–3): 71–80. Dey, M. M., R. Briones, and M. Ahmed. 2005. “Projecting Supply, Demand and Trade for Specific Fish Types in Asia.”  Aquaculture Economics and Hardy, R. W. 2003. “Use of Soybean Meals in the Diets of Salmon and Trout.” Management 9 (1–2): 113–40. Technical factsheet written in conjunction with United Soybean Board and American Soybean Association. http://www.soymeal.org Dey, M. M., R. M. Briones, Y. T. Garcia, A. Nissapa, U. P. Rodriguez, R. K. /FactSheets/SalmonidTechReview.pdf (accessed February 28, 2013). Talukder, A. Senaratne and others. 2008. Strategies and Options for Increasing and Sustaining Fisheries and Aquaculture Production Havlík, P., U. A. Schneider, E. Schmid, H. Böttcher, S. Fritz, R. Skalsky, K. Aoki to Benefit Poorer Households in Asia. WorldFish Center Studies and and others. 2011. “Global Land-Use Implications of First and Second Reviews 1823. Penang, Malaysia: WorldFish Center. Generation Biofuel Targets.” Energy Policy 39 (10): 5690–702. Dey, M. M., P. Kumar, F. J. Paraguas, O. L. Chen, M. A. Khan, and N. Srichan- Hertel, T. W., ed. 1997. Global Trade Analysis: Modeling and Applications. tuk. 2010. “Performance and Nature of Genetically Improved Carp Cambridge: Cambridge University Press. Strains in Asian Countries.” Aquaculture Economics and Management Hung, D., N. H. Nguyen, R. W. Ponzoni, D. A. Hurwood, and P. B. Mather. 14 (1): 3–19. 2013. “Quantitative Genetic Parameter Estimates for Body and Carcass Dowey, D. 2007. “Documentation of the AGLINK-COSIMO Model.” Traits in a Cultured Stock of Giant Freshwater Prawn (Macrobrachium Working Party on Agricultural Policy and Markets, AGR/CA/ rosenbergii) Selected for Harvest Weight in Vietnam.” Aquaculture 404: APM(2006)16/FINAL, Organisation for Economic Co-operation and 122–29. Development, Paris. IFFO (International Fishmeal and Fish Oil Organisation). 2011. IFFO Eknath, A. E., H. B. Bentsen, R. W. Ponzoni, M. Rye, N. H. Nguyen, J. Statistical Yearbook 2011. London: IFFO. Thodesen, and B. Gjerde. 2007. “Genetic Improvement of Farmed Jackson, A. 2010. “IFFO Responsible Supply Certification Program.” Tilapias: Composition and Genetic Parameters of a Synthetic Presentation given at the “Global Outlook for Aquaculture Leadership Base Population of Oreochromis niloticus for Selective Breeding. Conference,” Kuala Lumpur, Malaysia, October 17–20. http://www Aquaculture 273: 1–14. .gaal liance.org/update/GOAL10/Jackson.pdf (accessed February 25, FAO (Food and Agriculture Organization of the UN). 2009. “Climate 2012). Change Implications for Fisheries and Aquaculture.” FAO Fisheries Jackson, A., and J. Shepherd. 2010. “Connections between Farmed and and Aquaculture Technical Paper 530, FAO, Rome. Wild Fish: Fishmeal and Fish Oil as Feed Ingredients in Sustainable FAO. 2012. The State of the World Fisheries and Aquaculture 2012. Rome: Aquaculture.” In Advancing the Aquaculture Agenda: Workshop FAO. Proceedings, by OECD, 331–43. F I S H TO 2 030: P R O S P E C T S F O R F I S H E R I E S A N D A Q UA C U LT U R E REFERENCES 79 Jansson, T., and M. Kempen. 2006. “Introduction to Multi-Commodity Redfern Associates. 2010. “The China Seafood Market: Sector Overview Models: The CAPRI Market Model.” Presentation given at the CAPRI 2010.” Joint market study of Redfern Associates with New Zealand Trade Training Session, Warsaw, Poland, June 26–30. www.capri-model.org and Enterprise. http://www.nzte.govt.nz/explore-export-markets /trainingPPTS/market.ppt (accessed July 14, 2013). /market-research-by-industry/Food-and-beverage/Documents /Seafood-market-in-China-May-2010.pdf (accessed March 4, 2013). Kullback, S. 1959. Information Theory and Statistics. New York: John Wiley and Sons. Robinson, S., A. Cattaneo, and M. El-Said. 2000. “Updating and Estimating a Social Accounting Matrix Using Cross Entropy Methods.” Trade and Kullback, S., and R. A. Leibler. 1951. “On Information and Sufficiency.” Macroeconomics Division Discussion Paper 58, International Food Annals of Mathematical Statistics 22: 79–86. Policy Research Institute, Washington, DC. Leaño, E. M., and C. V. Mohan. 2012. “Early Mortality Syndrome Threatens Rosegrant M. W., M. S. Paisner, S. Meijer, and J. Witcover. 2001. Global Asia’s Shrimp Farms.” Global Aquaculture Advocate (July–August): Food Projections to 2020: Emerging Trends and Alternative Futures. 38–39. Washington, DC: International Food Policy Research Institute. Leung, T. L. F., and A. E. Bates. 2013. “More Rapid and Severe Disease Rosegrant, M. W., M. Agcaoili-Sombilla, and N. D. Perez. 1995. “Global Outbreaks for Aquaculture at the Tropics: Implications for Food Food Projections to 2020: Implications for Investment.” 2020 Security.” Journal of Applied Ecology 50 (1): 215–22. Discussion Paper 5, International Food Policy Research Institute, Washington, DC. Li, M. H., B. B. Manning, E. H. Robinson, B. G. Bosworth, and W. R. Wolters. 2001. “Comparison of Growth, Processing Yield, and Body Rosegrant, M. W., and the IMPACT Development Team. 2012. “International Composition of USDA103 and Mississippi ‘Normal’ Strains of Channel Model for Policy Analysis of Agricultural Commodities and Trade Catfish Fed Diets Containing Three Concentrations of Protein.” Journal (IMPACT): Model Description.” Washington, DC: International Food of the World Aquaculture Society 32 (4): 402–08. Policy Research Institute. http://www.ifpri.org/sites/default/files /publications/impactwater2012.pdf (accessed February 25, 2012). Melnychuk, M. C., J. A. Banobi, and R. Hilborn. 2013. “Effects of Management Tactics on Meeting Conservation Objectives for Western North Shannon, C. E. 1948a. “A Mathematical Theory of Communication.” Bell American Groundfish Fisheries.” PLOS ONE 8 (2): e56684. System Technical Journal 27 (July): 379–423. Naylor, R. L., R. W. Hardy, D. P. Bureau, A. Chiu, M. Elliott, A. P. Farrell, I. Shannon, C. E. 1948b. “A Mathematical Theory of Communication.” Bell Forster and others. 2009. “Feeding Aquaculture in an Era of Finite System Technical Journal 27 (October): 623–56. Resources.” Proceedings of the National Academy of Sciences 106 (36): Shepherd, J. 2012. “Aquaculture: Are the Criticisms Justified? Feeding Fish 15103–10. to Fish.” World Agriculture 3 (2): 11–18. Ninh, N. H., R. W. Ponzoni, N. H. Nguyen, J. A. Woolliams, J. B. Taggart, Silva, J. L., and S. Dean. 2001. “Processed Catfish, Product Forms, B. J. McAndrew, and D. J. Penman. 2013. “A Comparison of Communal Packaging, Yields and Product Mix.” SRAC Publication 184. https:// and Separate Rearing of Families in Selective Breeding of Common srac.tamu.edu/index.cfm/event/getFactSheet/whichfactsheet/26/. Carp (Cyprinus carpio): Responses to Selection.” Aquaculture 408–409: 152–59. Somwaru, A., and S. Dirkse. 2012. Dynamic PEATSim Model: Documen- ting Its Use in Analyzing Global Commodity Markets. Technical OECD (Organisation for Economic Co-operation and Development). Bulletin 1933. Washington, DC: Economic Research Service, U.S. 2010. Advancing the Aquaculture Agenda: Workshop Proceedings. Paris: Department of Agriculture. http://www.ers.usda.gov/publications OECD. /tb-technical-bulletin/tb1933.aspx (accessed August 12, 2013). OECD-FAO. 2012. World Agricultural Outlook 2012–2021. Rome: FAO; Paris: Tacon, A. G. J., and M. Metian. 2008. “Global Overview on the Use of Fish OECD. Meal and Fish Oil in Industrially Compounded Aquafeeds: Trends and Future Prospects. Aquaculture 285 (1–4): 146–58. Oken, E., M. L. Østerdal, M. W. Gillman, V. K. Knudsen, T. I. Halldorsson, M. Strom, D. C. Bellinger, M. Hadders-Algra, K. F. Michaelsen, and S. F. Tacon, A. G. J., M. R. Hasan, and M. Metian. 2011. “Demand and Supply Olsen. 2008. “Associations of Maternal Fish Intake During Pregnancy of Feed Ingredients for Farmed Fish and Crustaceans: Trends and and Breastfeeding Duration with Attainment of Developmental Prospects.” FAO Fisheries and Aquaculture Technical Paper 564, FAO, Milestones in Early Childhood: A Study from the Danish National Rome. Birth Cohort.” American Journal of Clinical Nutrition 88 (3): 789–96. Tveterås, S., F. Asche, M. F. Bellemare, M. D. Smith, A. G. Guttormsen, A. Plassmann, F. 2004. “The Advantage of Avoiding the Armington Lem, K. Lien, and S. Vannuccini. 2012. “Fish Is Food: The FAO’s Fish Price Assumption in Multi-Region Models.” Working Paper 206, Depart- Index.” PLOS ONE 7 (5): e36731. doi:10.1371/journal.pone.0036731. ment of Economics, State University of New York, Binghamton, New York. http://www2.binghamton.edu/economics/research UN (United Nations). 2011. World Population Prospects: The 2010 Revi- /working-papers/pdfs/wp02/WP0206.pdf (accessed July 14, 2013). sion. United Nations Department of Economics and Social Affairs, United Nations, New York. http://esa.un.org/wpp/Documentation Pillay, T. V. R., and M. N. Kutty. 2010. Aquaculture: Principles and Practices, /publications.htm (accessed December 17, 2012). 2nd ed. Oxford, U.K.: Blackwell. USDA (U.S. Department of Agriculture). 2012. “Maternal Intake of Seafood Rana, K. J., S. Siriwardena, and M. R. Hasan. 2009. “Impact of Rising Feed Omega-3 Fatty Acids and Infant Health: A Review of the Evidence.” Ingredient Prices on Aquafeeds and Aquaculture Production. FAO Nutrition Insight 46. Center for Nutrition Policy and Promotion, Fisheries and Aquaculture Technical Paper 541, FAO, Rome. USDA, Washington, DC. A G R I C U LT U R E A N D R U R A L D E V E LO P M E N T D I S C U S S I O N PA P E R 80 REFERENCES van der Mensbrugghe, D. 2005. “LINKAGE Technical Reference World Bank. 2013b. Turn Down the Heat: Why a 4°C Warmer World Must Be Document: Version 6.0,” Mimeo, World Bank, Washington, DC. Avoided. Washington, DC: World Bank. http://siteresources.worldbank.org/INTPROSPECTS/Resources World Bank. 2013c. China 2030: Building a Modern, Harmonious and /334934-1100792545130/LinkageTechNote.pdf (accessed August 20, Creative High-Income Society. Washington, DC: World Bank. http:// 2013). documents.worldbank.org/curated/en/2012/02/15873233/china- Ward, T., and B. Phillips, eds. 2008. Seafood Ecolabelling: Principles and 2030-building-modern-harmonious-creative-high-income-society Practices. Chichester, U.K.: Wiley-Balckwell. (accessed March 3, 2013). World Bank. 2012. Global Economic Prospects: Managing Growth in a You, L., and S. Wood. 2004. “Assessing the Spatial Distribution of Crop Volatile World. Washington, DC: World Bank. http://siteresources. Production Using a Generalized Cross-Entropy Method.” Environ- worldbank.org/INTPROSPECTS/Resources/334934-1322593305595 ment and Production Technology Division Discussion Paper 126, /8287139-1339427993716/GEPJune2012_Full_Report.pdf (accessed International Food Policy Research Institute, Washington, DC. December 17, 2012). World Bank. 2013a. World Bank Group Agriculture Action Plan 2013–15. Washington, DC: World Bank. F I S H TO 2 030: P R O S P E C T S F O R F I S H E R I E S A N D A Q UA C U LT U R E Agriculture and Environmental Services (AES) 1818 H Street, NW Washington, D.C. 20433 USA Telephone: 202-477-1000 Internet: www.worldbank.org/ard