AGRICULTURE AND FOOD DISCUSSION PAPER MODELING THE IMPACTS OF AGRICULTURAL SUPPORT POLICIES ON EMISSIONS FROM AGRICULTURE David Laborde Abdullah Mamun Will Martin Valeria Piñeiro Rob Vos Agriculture and Food Discussion Paper Modeling the Impacts of Agricultural Support Policies on Emissions from Agriculture David Laborde Abdullah Mamun Will Martin Valeria Piñeiro Rob Vos August 2020 AUTHORS: David Laborde (d.laborde@cgiar.org) is a Senior Research Fellow, Abdullah Mamun (a.mamun@cgiar.org) is a Research Analyst in the Markets, Will Martin (w.martin@cgiar.org) is a Senior Research Fellow in the Markets, Valeria Piñeiro (v.pineiro@cgiar.org) is a Senior Research Coordinator in the Markets, and Rob Vos (r.vos@cgiar.org) is the Director in the Markets, Trade, and Institutions Division of the International Food Policy Research Institute, Washington, DC © 2020 The World Bank 1818 H Street NW, Washington, DC 20433 Telephone: 202-473-1000; Internet: www.worldbank.org Some rights reserved 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. Nothing herein shall constitute or be considered to be a limitation upon or waiver of the privileges and immunities of The World Bank, all of which are specifically reserved. Rights and Permissions This work is available under the Creative Commons Attribution 3.0 IGO license (CC BY 3.0 IGO) http://creativecommons.org/licenses/by/3.0/ igo. Under the Creative Commons Attribution license, you are free to copy, distribute, transmit, and adapt this work, including for commercial purposes, under the following conditions: Translations—If you create a translation of this work, please add the following disclaimer along with the attribution: This translation was not created by The World Bank and should not be considered an official World Bank translation. The World Bank shall not be liable for any content or error in this translation. Adaptations—If you create an adaptation of this work, please add the following disclaimer along with the attribution: This is an adaptation of an original work by The World Bank. Views and opinions expressed in the adaptation are the sole responsibility of the author or authors of the adaptation and are not endorsed by The World Bank. Third-party content—The World Bank does not necessarily own each component of the content contained within the work. The World Bank therefore does not warrant that the use of any third-party-owned individual component or part contained in the work will not infringe on the rights of those third parties. The risk of claims resulting from such infringement rests solely with you. If you wish to re-use a component of the work, it is your responsibility to determine whether permission is needed for that re-use and to obtain permission from the copyright owner. Examples of components can include, but are not limited to, tables, figures, or images. All queries on rights and licenses should be addressed to World Bank Publications, The World Bank Group, 1818 H Street NW, Washington, DC 20433, USA; fax: 202-522-2625; e-mail: pubrights@worldbank.org. Cover photo: © Tran Thi Hoa / World Bank Picture on page 3: Simone D. McCourtie / World Bank Picture on page 9: Maria Fleischmann / World Bank Picture on page 30: Arne Hoel / World Bank Picture on page 36: Markus Kostner / World Bank CONTENTS Abstract v Acknowledgements v Section One: Introduction 1 Section Two: The Emissions Database 5 Section Three: Agricultural Incentives 11 Coupled Subsidies 12 Import Measures 14 Export Measures 16 Coupled Subsidies vs Market Price Support 17 Section Four: Analyzing the Impacts of Agricultural Support 19 Impacts on Emissions of Coupled Subsidies 20 Impacts on Emissions of Border Distortions 22 Joint Impact of Border Measures and Coupled Subsidies 27 Removing all Production Subsidies and Market Price Support but maintaining Direct Income Support to Farmers 30 Productivity Improvements 33 Section Five: Conclusions 37 References 39 TABLES Table 1: Emissions by type of Greenhouse Gas, 2010 6 Table 2: Emissions* from Agriculture by Source, 2013-15 (shares in percent) 7 Table 3: Agricultural Emissions* by Commodity, 2014 (shares in percent) 8 Table 4: Emission Intensities for Key Products and Regions (kg CO2 eq./kg Of product) 8 Table 5: Shares of Agricultural Emissions by Commodity and Source, 2015 (% of total) 10 Table 6: Annual coupled subsidies and GSSE paid by governments, average for 2014-16 (US$ billion) 12 Table 7: Agricultural Import Protection in Covered Countries, % 15 Table 8: Export taxes and/or subsidies, % 16 Table 9: Coupled Subsidies vs Market Price Support, 2014-16 (%) 17 Table 10: Percentage Changes in Output following Removal of Coupled Subsidies, % 20 Table 11: Impacts on Emissions of Abolishing Coupled Subsidies, kt of CO2 equivalent 21 Table 12: Output Changes from Abolition of all Border Measures, % 23 Table 13: Changes in Real National Income and Farm Income if all Border Measures on Agricultural Commodities would be removed, % 25 Table 14: Impacts on Emissions of Abolishing Border Distortions, kt of CO2 equivalent 26 iii Table 15: Output Changes from Abolition of all Border Measures & Coupled Subsidies, % 28 Table 16: Impacts on Emissions of Abolishing Coupled Subsidies & Border Distortions, kt of CO2 equivalent 29 Table 17: Output impacts of removing all support with uniform transfers to maintain farm incomes, % 31 Table 18: Impacts on Emissions of Removing all Support with Compensation, kt of CO2 equivalent 32 Table 19: Reductions in Emissions from Agriculture from Productivity Shocks (% change in CO2 eq) 34 Table 20: Percentage Changes in Output by Sector and Region, 30% MFP increase 35 FIGURES Figure 1: Emissions from Agriculture and Land Use relative to Other Emissions, 2010 6 iv Modeling the Impacts of Agricultural Support Policies on Emissions from Agriculture ABSTRACT To understand the impacts of support programs on global those that reduce emission intensities by increasing overall emissions, this paper considers the impacts of domestic productivity because overall productivity growth creates a subsidies, price distortions at the border, and investments rebound effect by reducing product prices and expanding in emission-reducing technologies on global greenhouse output. A key challenge is designing policy reforms that gas (GHG) emissions from agriculture. In a step towards effectively reduce emissions without jeopardizing other a full evaluation of the impacts, it uses a counterfactual key goals such as improving nutrition and reducing pov- global model scenario showing how much emissions from erty. While the scenario analysis in this paper does not agricultural production would change if agricultural sup- propose any particular policy reform, it does provide an port were abolished worldwide. The analysis indicates important building block towards a full understanding the that, without subsidies paid directly to farmers, output of impacts of repurposed agricultural support measures on some emission-intensive activities and agricultural emis- mitigation of greenhouse gas emissions and adaptation to sions would be smaller. Without agricultural trade protec- climate change. That full analysis is being undertaken in tion, however, emissions would be higher. This is partly subsequent work, which will also take account of land-use because protection reduces global demand more than it change and alternative forms of agricultural policy sup- increases global agricultural supply, and partly because port to align objectives of food security, farmers’ income some countries that currently tax agriculture have high security, production efficiency and resilience, and environ- emission intensities. Policies that directly reduce emission mental protection. intensities yield much larger reductions in emissions than ACKNOWLEDGEMENTS This technical paper is part of the input for a World Bank study on “Environmental Impacts of Agricultural Support: Aligning Food Security and Climate Protection Objectives.” It was prepared as an output of the CGIAR Research Program on Policies, Institutions and Markets (PIM). Funding for this research was provided by The World Bank and PIM. The authors are grateful to Madhur Gautam for overall project supervision, advice and support; to Raffaello Cervigni, Richard Damania, Mike Toman, Stephen Ling, Dina Umali-Deininger, and Sergiy Zorya of the World Bank for detailed review comments; and to participants at workshops and conferences at an IFPRI policy seminar (21 April 2020); NBER (30 April 2020); and the GTAP Conference (17 June 2020) for helpful comments on earlier drafts of this paper. v SECTION ONE INTRODUCTION Agricultural production is strongly affected by climate change but is, at the same time, a major contributor to climate change—with agriculture and land use accounting for around a quarter of total global emissions of greenhouse gases (Tubiello 2019). World agriculture is also strongly affected by support policies that affect the composition and location of output and the approaches used for production. One key question is what impact these interventions have on greenhouse gas (GHG) emissions from agriculture by providing incentives to lower or higher emission-intensive crops or livestock pro- duction. A related question is how these support measures might be reformed so they could not only support farm incomes and productivity growth, but also provide incen- tives for reducing greenhouse gas emissions. The farm sector in the 51 countries included in the OECD’s annual policy monitoring program received an average of US$440 billion per year in the form of market price support and direct subsidies between 2016 and 2018, equal to about 15 percent of gross farm receipts (OECD 2019). These countries also spent US$105 billion per year on what OECD terms General Services Support, that is policies designed to create enabling conditions for agriculture, such as agricultural innovation systems, sanitary and phytosanitary standards and rural infrastructure. The US$440 billion in market price support includes direct subsidies that are “coupled” to output and create incen- tives for producers to expand output (US$176 bn); and “decoupled” subsidies that seek to avoid altering production incentives (US$64 bn). Recent work by Mamun, Martin and Tokgoz (2019) has examined these types of agri- cultural support and their relationship to GHG emissions from agriculture. This study highlighted a few key facts: • the extreme concentration of agricultural emissions by commodity - with beef, dairy and rice accounting for over 80 percent of agricultural GHG emissions; • a substantial fraction of emissions comes from land use and land use change; • the production of these emission-intensive goods is often heavily supported using market-price-support (MPS) measures, thereby creating strong incen- tives to increase output but reduce consumption of agricultural commodities in the protecting country. 1 Further analysis is needed, however, to be confident about differences in the emission intensity1 of commodities. As the implications of reforming agricultural support meas- noted by Mamun, Martin and Tokgoz (2019), individual ures for overall GHG emissions. Some of the key factors agricultural commodities are likely to be more responsive that need to be considered include: (i) the average rate of to differentials in agricultural support rates than is over- support to agriculture, (ii) differences between types of support, all agricultural output to the average rate of agricultural (iii) differences in rates of support across commodities, and (iv) support. This is because increasing total agricultural impacts of support on the production methods and processes. output requires either changing the amount of agricul- tural land, or intensifying production on existing land The average rate of support to agriculture matters because through increased input use, such as fertilizer and labor. high rates of support are likely to attract resources into Of course, the output mix can be changed by switching agriculture, increase output and, at constant technology, to other crops or livestock production on existing land. to increase emissions from agricultural activity. The result- Mamun, Martin and Tokgoz (2019) concluded that, on ing increase in agricultural output is likely small, because average, the existing structure of agricultural support pro- total agricultural production can only be increased by vides little incentive to farmers to switch from high to low raising total agricultural land use or substituting other emission-intensive commodities. inputs for land. Increases in agricultural land use, in turn, are likely to induce very large one-off increases in emis- Differences in agricultural support across countries also matter sions as land is cleared for use in agriculture, particularly for global outcomes. Mamun, Martin and Tokgoz (2019) if land is cleared by burning forests. show that the emission intensities of production for the same commodities differ substantially between rich and The type of support matters because of its influence on overall poor countries and, also, within those groups Historically, incentives to both producers and consumers. Some distor- support to farmers has been higher in high-income coun- tions to agricultural incentives, such as coupled subsidies tries than in poor, creating incentives to expand produc- paid by governments to producers, provide incentives to tion in lower-emission countries. However, in recent years, producers without providing any incentives for consumers MPS rates in high-income countries have declined, while to reduce consumption. Other types of support, such as they have increased in developing countries (Laborde and market-price support (MPS) in the form of tariff or non- Mamun 2019). The key point for this study is that support tariff barriers, provide incentives for producers to increase in countries with high emission intensities increases global output, while encouraging consumers in protected mar- output in those countries and, other things being equal, kets to reduce consumption of the protected good. This increases global emissions per unit of global output. suggests that MPS, while more trade-distorting than coupled subsidies, may have a smaller impact on global Support intended to influence production practices and agricultural output and emissions. Provision of MPS may processes, such as subsidies on fertilizers, pesticides or even reduce global agricultural output and emissions. improved seeds, also matter. In practice, these mostly aim Consider, for example, protection provided in countries to stimulate agricultural production which may induce with low agricultural production. The import protection more emissions unless improved practices are more will raise consumer prices in that region and, hence, lower resource efficient. Higher input use, such as in the case of consumption and global output. Decoupled support aims fertilizer, may be an additional source of GHG emissions, to provide support to farmers without creating incentives though improved, climate-resilient seeds may prove more to change output levels and—to the extent that it is truly environment-friendly. Some support programs, like the decoupled—can be ignored for our purposes. reformed Common Agricultural Policy of the EU, condi- tion access to support to compliance with environment- Differences in rates of support across commodities may have friendly production and land conservation practices. important impacts on overall emissions given large 1 Defined as the quantity of GHG emitted per unit of output. 2 Modeling the Impacts of Agricultural Support Policies on Emissions from Agriculture OECD’s General Services Support Estimate (GSSE), al. 2013). A fifth group of studies examines options for includes measures, such as agricultural research, devel- creating a more sustainable food system (e.g., Searchinger opment, and training, that aim to raise agricultural pro- et al. 2019). A sixth group addresses the possibilities and ductivity. While these measures are not part of the widely impacts of reforming agricultural support measures (e.g., reported Producer Support Estimate (PSE), they can be Mamun, Martin and Tokgoz 2020; Searchinger et al. critical to productivity growth and, hence, to incomes 2020). of farmers and food prices for consumers. Productivity gains tend to reduce the emission intensities of agricul- This study is a first stage of analysis of the impacts of tural products, for instance, through changes in produc- reforming agricultural subsidies for adaptation to and mit- tion processes that also reduce emission intensities (see, for igation of climate change. In it, we use as wide a coverage example, Mernit 2018), or through more efficient use of as possible of agricultural policy measures and emission intermediate inputs. sources and focus primarily on the implications of cur- rent agricultural support policies for emission outcomes. This paper is designed to complement existing studies on In doing so, we undertake a careful quantitative analysis the measurement of GHG emissions from agriculture of the impacts of incentives on agricultural outputs and and the implications of reform. One key set of studies emissions using IFPRI’s global computable general equi- (see, for example, Tubiello 2019) assesses the extent of librium model, MIRAGRODEP. This model-based anal- emissions from agriculture. A second group of studies ysis provides an opportunity to consider all the influences (see, for example, Jensen et al. 2019, Fellman et al. 2019, outlined above—impacts on overall output, differences in and OECD 2019) examines the implications of poten- incentives across countries, differences in incentives across tial policies such as carbon and consumption taxes for commodities, and differences in the technology used for reducing GHG emissions from global agriculture. A third production. It also allows us to examine the extent and group of studies focusses reform of agricultural support potential implications of environmental conditionalities policies in the European Union, using the CAPRI model incorporated in producer support measures. We consider (see, for example, Himics et al. 2018). A fourth group of not just the total emissions per unit of output, but also the studies (e.g., Henderson and Lankoski 2019) examines the source of those emissions—whether they are, for instance, potential implications of policy reform in OECD coun- from enteric fermentation by ruminants or from fertilizer tries and/or in specific agricultural sub-sectors (Gerber et use. Introduction 3 The analysis also provides a framework for assessing the resilience and help mitigate climate change. A necessary potential impact of innovations that reduce the emission- condition for designing reforms to achieve these goals is intensity of production, such as by investing in tech- to understand first the relationships between existing sup- nologies that reduce emissions of GHGs from enteric port measures and emissions from agriculture. The next fermentation in ruminants (Boadi et al. 2004; Haque stage of the analysis will build on this knowledge to design 2018). This is relevant to any consideration of agricul- and assess policies for achieving the mentioned objectives tural support measures because these measures, as defined and possible trade-offs among them. by the OECD (2018), include investments in agricultural knowledge development and dissemination. The next section of this paper explains the process by which we developed the database of emissions that Because we were unable to incorporate the impacts of allowed us to track emissions by source and product. The land use change on emissions, and because the impacts of third section details the approach we used to represent policy reform on agricultural emissions at constant land the distortions to agricultural incentives examined in the use are of interest in themselves, we consider only changes study. The fourth section discusses the simulations used in emissions from agriculture. This provides a platform to assess the impacts of changes in agricultural incentives for subsequent analysis in which the impacts of land use on global emissions. The fifth section considers the links change on global emissions are incorporated. between changes in productivity and global emissions. The sixth section discusses the approach to policy given As said, this paper is the first stage of ongoing research policy makers’ multiple goals and the instruments consid- to gain insights in the potential for reforming and repur- ered in this paper. posing support measures to help reduce poverty, increase 4 Modeling the Impacts of Agricultural Support Policies on Emissions from Agriculture SECTION TWO THE EMISSIONS DATABASE The focus of this paper is on the impact of agricultural support on GHG emissions. It is important first to put these emissions in the broader context of emissions from all sources, such as energy, industry, land use and transport. To ensure comparability of the data, we draw on the FAOSTAT (2019) data, for which the most recent compa- rable on agriculture and non-agriculture are for 2010. A key feature of the Figure 1 is the importance of agriculture and land use change in total emissions, with over 20 percent of all GHG emissions. At 21 percent of the total, it is small relative to energy and industry, but much larger than the emissions associ- ated with transport and residential/commercial use, that receive enormous attention in popular discussions. While it was discussed in the original UN Framework Conven- tion on Climate Change (UN 1992), it was seriously under-represented in subsequent commitments, partly because of poor and inconsistent data on the extent of the prob- lem. With the ready availability of estimates through FAOSTAT (see Tubiello 2019), the basic information needed to analyze the problem and approaches to its resolu- tion is now available. Emissions from agriculture and land use change are much more comprehensively included in the Paris Agreement than in earlier accords (Vermeulen 2016). Because the modeling framework we use for this paper covers only ongoing emissions from agriculture, we focus only on the agriculture component of emissions. An important distinction between agriculture and other sources of GHG emissions is the much greater relative importance of the relatively short-lived methane (CH4), with a lifespan of around 12 years in the atmosphere, and nitrous oxide (N2O), with a lifespan of 120 years, emissions from agriculture (Houghton et al. 1996). As shown in Table 1, these two gases account for almost all the emissions from agriculture, while the emissions from most other sectors are dominated by carbon dioxide (CO2). The Global Warming Potential (GWP) conversion factors used in the database are those of IPCC (Houghton et al. 1996, p22) that compare the GWP of each gas after 100 years2. 2 Where timing issues are important, Edwards and Trancik (2014) propose alternative approaches. 5 FIGURE 1. EMISSIONS FROM AGRICULTURE AND LAND USE RELATIVE TO OTHER EMISSIONS, 2010 Energy & Industry Land use & change 58% 11% Agriculture 10% Residential & commercial 8% International Transport 2% Domestic Transport 11% Source: FAO. Note. Shares are percent of total emissions, excluding sequestration by forests. TABLE 1. EMISSIONS BY TYPE OF GREENHOUSE GAS, 2010 Methane Carbon F-Gases Nitrous Oxide Total Dioxide Agriculture 56.4 0.0 0.0 43.6 100.0 Land Use 5.2 93.3 0.0 1.5 100.0 Energy 11.8 87.5 0.0 0.8 100.0 Residential/ 7.2 90.4 0.0 2.4 100.0 Commercial Industry/ 25.2 49.5 16.2 9.1 100.0 Other/Waste International 0.2 99.0 0.0 0.8 100.0 Transport Domestic 0.0 99.7 0.0 0.3 100.0 Transport Sources total 15.1 77.0 1.7 6.2 100.0 Source: FAOSTAT. For the present analysis, a database is required that links relationship between outputs and inputs in response to decisions on agricultural outputs and inputs with their changes in prices or technological changes in input-out- associated emissions. For this, we build on the FAOSTAT put relationships. global database (Tubiello 2019; FAO 2019) that links GHG emissions to output of commodities and the pro- We make two important improvements to the FAOSTAT duction processes used in countries. Holding emissions database: (i) adding additional categories of emissions, per unit of output and input constant allows us to develop and (ii) allocating emissions by commodity across sources a fixed-coefficient model of emissions. Our modeling of emissions. The two most important categories of framework also provides the option for us to vary the emissions to be added are energy use in agriculture and 6 Modeling the Impacts of Agricultural Support Policies on Emissions from Agriculture emissions from pesticide use. We also move emissions from The FAOSTAT database also provides data on emis- burning of savannah grasslands to the land use category, sions by commodity, as presented in Table 3. Production rather than the agricultural emissions category, as much of ruminant meat is by far the most important source of of this burning is unrelated to agricultural production. emissions, accounting for 48 percent of total emissions. Milk production is the second most important source, with Table 2 shows the importance of incorporating different close to 17 percent of the total. This is broadly consistent sources of emissions into our analysis. A striking feature with the importance of enteric fermentation and manure of the table is the importance of enteric fermentation in Table 1. Rice is the third most important source of in both rich and poor countries. This is primarily from global emissions. This importance of this category is very ruminant animals such as cattle and sheep and contrib- different between OECD and non-OECD countries, with utes close to 34 percent of total emissions. Another strik- its share of emissions six times higher in Non-OECD ing feature is the importance of livestock manure, which countries than in the OECD. contributes 22.5 percent of total emissions. Direct energy use by the sector accounts for 13 percent of total energy A key parameter to understand the impact on climate use and 16 percent in OECD members. Volatilization change of agricultural support measures (and changes of synthetic fertilizers is another important contributor, to such policies) is the relative emission intensity of pro- accounting for 14 percent of total emissions in the OECD duction in different regions, measured by the kg of CO2 member countries and 10 percent in non-OECD mem- -equivalent produced per kg of the product. If support bers. Rice cultivation accounted for almost 10 percent of is reduced in an area where the emission intensity is par- total emissions, being much more important in developing ticularly high and replaced by production from another countries than in the OECD. Crop residues were the next region where the emission intensity is low, this change is most important source, accounting for roughly almost likely to contribute to a reduction in total emissions. The five percent of emissions, considering both burning and emission intensities for key products and regions are pre- decomposition of residues, making them similar in impor- sented in Table 4. tance to emissions resulting from direct energy use by the agricultural sector. TABLE 2. EMISSIONS* FROM AGRICULTURE BY SOURCE, 2013-15 (SHARES IN PERCENT) OECD Non-OECD World Burning of crop residues 0.5 0.5 0.5 Burning of savannah 3.3 4.3 4.1 Crop residues 4.5 3.2 3.5 Energy use** 16.4 11.8 12.9 Enteric fermentation 31.0 34.4 33.6 Manure management 9.2 4.6 5.7 Manure left on pasture 10.7 14.6 13.7 Manure applied to soils 4.4 2.7 3.1 Cultivation of organic soils 2.9 1.9 2.1 Pesticides 1.5 1.3 1.4 Rice Cultivation 1.7 10.7 8.6 Synthetic Fertilizers 14.0 10.0 10.9 Total 100.0 100.0 100.0 Source: FAOSTAT. Notes: * Emissions measured in CO2 equivalent. **Data for the latest available year, 2012. The Emissions Database 7 TABLE 3. AGRICULTURAL EMISSIONS* BY COMMODITY, 2014 (SHARES IN PERCENT) OECD Non-OECD World Rice 3.2 20.8 16.6 Other cereals 18.4 8.5 10.9 Milk 18.3 16.8 17.1 Ruminant meat 49.8 47.6 48.1 Pig meat 7.0 3.4 4.3 Poultry meat 2.1 1.7 1.8 Eggs 1.1 1.2 1.1 Total 100.0 100.0 100.0 Source: FAOSTAT. Note: * Emissions measured in CO2 equivalent. TABLE 4. EMISSION INTENSITIES FOR KEY PRODUCTS AND REGIONS (KG CO2 EQ./KG OF PRODUCT) Cereals Eggs Bovine Chicken Pig Milk Rice excl. rice meat meat Australia 0.3 0.4 20.2 0.2 2.5 0.7 0.7 Brazil 0.2 0.8 35.7 0.3 2.6 1.2 0.5 EU 0.2 0.7 15.4 0.3 1.6 0.6 3.0 India 0.3 0.5 108.3 0.5 5.0 1.1 0.7 USA 0.2 0.5 12.1 0.3 2.0 0.4 1.1 OECD 0.2 0.5 15.1 0.3 1.7 0.5 1.2 Non-OECD 0.2 0.8 32.8 0.7 1.4 1.3 0.9 World 0.2 0.7 25.4 0.6 1.5 0.9 0.9 Source: FAOSTAT. A key feature of Table 4 is the enormous differences in of 7.6 (FAO 2017, p123). The dispersion of intensities is emission intensities across regions and commodities. The much less than in the case of beef, for which it is quite sub- emission intensity for bovine meat is by far the largest for stantial. Intensities are generally lower in higher-income any food. And it varies from 12.1 in the United States countries. to 108.3 in India. There is clearly a link between income levels and intensity, with the intensity for beef being more A difficulty with the characterization of emissions pro- than twice as high in the non-OECD group than in the vided in Tables 2 and 3 is that it does not clarify the source OECD. However, there are clearly also important idio- of emissions in the production of each commodity. If, for syncratic influences on emission intensities—they are, for instance, we want to examine the impacts of a fertilizer instance, slightly lower in the EU and the United States subsidy, these data—or models based on them—do not than for the OECD as a group. As another example, both allow us to investigate the impacts. Nor do they allow us Australia and Brazil have emission intensities that are to investigate the impacts of changes in production tech- slightly higher than for the OECD and non-OECD aver- nologies that apply to one production activity rather than ages. The emission intensities for milk look deceptively others—such as use of feed additives designed to reduce low but would range from 3.3 to 9.7 if expressed in milk GHG emissions—suitable for daily-handled dairy cows powder equivalent using the standard conversion factor but not beef cattle fed on extensive pastures. To deal with 8 Modeling the Impacts of Agricultural Support Policies on Emissions from Agriculture this problem, we developed a database of emissions by type and by commodity. We aimed to produce a database of emissions for production, based on FAOSTAT, with the key assumptions used in the carbon footprint of agricul- ture. The idea is to reverse engineer aggregated emission data to identify individual coefficients that properly traces the drivers of emissions: scale, location, composition and technological choice. In doing so, we followed a bottom-up approach for computing emissions at different stages and carried out quality control at different stages by checking if the numbers add up to the aggregated numbers at the total agriculture level in the FAOSTAT database. Thus, GHG by source are calculated in three steps, that is, we recalculate first the emission coefficients (EC) for N2O and CH4 for each emission source; then we recompute emissions of N2O and CH4 for each activity and finally, we recompute the CO2-equivalent from N2O and CH4 emissions.3 The emission sources used follow FAOSTAT and include: burning crop residues and savanna; emis- sions from crop residues; manure management, manure applied to cropland manure left on pasture; rice cultiva- tion; cultivation of organic soils; enteric fermentation, energy and synthetic fertilizers. Emissions from energy consumption in agriculture are taken from the FAOSTAT for the year 2012, the latest available year for this series. The FAOSTAT emission database further provides created for the present analysis, we keep the base activ- implied emission factors for various activities by emission ity data (we call them index data) to get average emission source, such as area harvested in rice cultivation and the value per index type (land, animals, output, fertilizer and nitrogen content of manure. The database in some cases energy). provides the base activity data, such as areas of organic soil cultivation; and the number of head of livestock for In the case of enteric fermentation and manure manage- enteric fermentation and manure management. In other ment (during storage and treatment, in application to soils cases, such as biomass burned (dry matter) when burning and left on pasture) we disaggregated the livestock num- crop residues, only computed activity data are provided, bers for buffaloes, camels, goats and sheep in line with rather than the underlying data. In such cases, we import the value of their products– meat, milk and wool (sheep). base activity data from the FAOSTAT crop and livestock The resulting livestock numbers were then linked to emis- production database. For synthetic nitrogen fertilizer, the sions using data from the FAOSTAT emissions database. activity data (i.e., agriculture use in nutrients) is missing In the final step we produced emissions data by country, and thus we used the FAO emission numbers instead of emission source and sector. The broad structure of the recomputing them. For the final version of the database emission shares is presented in Table 5. 3 We use 310 as the default value for computing the CO2 equivalent of N2O and 21 as the CO2 equivalent of CH4. The Emissions Database 9 TABLE 5. SHARES OF AGRICULTURAL EMISSIONS BY COMMODITY AND SOURCE, 2015 (% OF TOTAL) Rice Other Milk Ruminant Pig Poultry Eggs Total cereals meat meat meat Burning crops 0.2 0.5 0.0 0.0 0.0 0.0 0.0 0.7 Crop residue 1.3 3.1 0.0 0.0 0.0 0.0 0.0 4.4 Enteric fermentation 0.0 0.0 11.0 30.5 0.6 0.0 0.0 42.1 Manure management 0.0 0.0 1.6 2.4 2.8 0.4 0.3 7.5 Manure left on pasture 0.0 0.0 3.6 13.3 0.0 0.7 0.4 18.0 Manure applied to soils 0.0 0.0 1.0 1.1 0.9 0.7 0.4 4.2 Pesticides 0.2 0.8 0.0 0.1 0.0 0.0 0.0 1.1 Rice cultivation 12.6 0.0 0.0 0.0 0.0 0.0 0.0 12.6 Synthetic fertilizers 2.4 6.5 0.0 0.7 0.0 0.0 0.0 9.6 Total 16.6 10.9 17.1 48.1 4.3 1.8 1.1 100.0 Source: Authors’ computation. Notes: Results are global averages. 10 Modeling the Impacts of Agricultural Support Policies on Emissions from Agriculture SECTION THREE AGRICULTURAL INCENTIVES Farmers’ decisions on production levels and methods of production are influenced by three broad policy interventions: 1. Coupled and decoupled subsidies 2. Import measures 3. Export measures In this section of the study, we discuss the approaches that we used to measure the extent of these interventions. We first consider coupled subsidies. These consist of payments by the government that change the net returns farmers receive for their cur- rent outputs or pay for their inputs and, hence, create incentives for farmers to change their output levels. We look at these first, partly because they are expected to have a more adverse impact on emissions from agriculture than import or export measures that contribute to the overall level of measured market price support (MPS). This is because, in contrast with market price support measures such as tariffs, they do not provide incentives for consumers in protecting countries to reduce their consumption of potentially polluting agricultural commodities4. There is also a political-economy reason for focus on these subsidies. As they must be financed by governments, they are likely to receive more scrutiny from Ministries of Finance than measures such as tariffs that change producer returns indirectly by raising domestic prices relative to world prices. The export taxes used in some developing countries have a similar feature, with Ministries of Finance often providing support, rather than opposition, for these measures. 4 Our emphasis on these commodities differs from studies such as Mayrand et al. (2003) in which coupled subsidies are classified as most-distorting. While we agree with this assessment for trade impacts, we do not for environmental impacts. Whether trade barriers will raise or lower emissions is an empirical question. If protection is applied mainly to countries with little comparative advantage, where production at world prices is far below consumption, the reduc- tion in demand may exceed the reduction in supply. Whether this is the case also depends on the elasticities of supply and demand. 11 COUPLED SUBSIDIES the share of coupled subsidies that are subject to environ- mental conditionality. A third is to identify those subsidies that support the use of fertilizer. The final column, for Data on coupled subsidies paid by government for pro- GSSE, is provided to allow comparison between subsidies ducers are presented in Table 6. The purpose of this table paid directly to farmers and payments made to strengthen is to disaggregate total subsidies into those that influence the enabling environment for agriculture (see OECD output (coupled subsidies) and those that do not (decou- 2018a). pled subsidies). Another purpose is to provide estimates of TABLE 6. ANNUAL COUPLED SUBSIDIES AND GSSE PAID BY GOVERNMENTS, AVERAGE FOR 2014-16 (US$ BILLION) Narrow Broader All Environmental Environmental Fertilizer Coupled Decoupled GSSE Conditionality Conditionality Subsidies Subsidies Subsidies Australia 0.2 0.1 0.0 0.5 0.4 1.0 Brazil 0.0 2.9 0.0 4.5 0.0 2.7 Canada 0.0 0.0 0.0 1.7 0.0 1.8 Switzerland 0.8 2.6 0.0 1.4 1.1 0.8 Chile 0.0 0.1 0.0 0.4 0.0 0.4 China 0.0 7.2 0.0 53.7 5.9 39.9 Colombia 0.0 0.0 0.0 0.9 0.0 0.6 Costa Rica 0.0 0.0 0.0 0.0 0.0 0.0 EU28 8.8 65.0 0.0 36.2 44.4 12.9 Indonesia 0.0 0.0 1.4 1.9 0.0 1.2 India 0.0 0.0 11.0 28.0 0.0 12.2 Iceland 0.0 0.0 0.0 0.1 0.0 0.0 Israel 0.0 0.0 0.0 0.2 0.0 0.2 Japan 2.4 5.8 0.0 4.8 3.0 8.4 Kazakhstan 0.0 0.0 0.0 1.1 0.0 0.5 Korea 0.8 0.9 0.1 1.0 0.8 2.9 Mexico 1.1 1.1 0.0 3.7 0.0 0.8 Norway 0.1 0.1 0.0 1.4 0.0 0.2 New Zealand 0.0 0.0 0.0 0.0 0.0 0.4 Philippines 0.0 0.0 0.0 0.0 0.0 0.0 Russia 0.0 0.0 0.0 0.0 0.0 0.0 Turkey 0.1 0.1 0.3 3.1 0.0 2.9 Ukraine 0.0 0.0 0.0 1.1 0.0 0.2 USA 5.1 24.9 0.0 16.8 11.2 8.7 Vietnam 0.0 0.0 0.0 0.5 0.0 0.7 South Africa 0.0 0.0 0.0 0.2 0.0 0.3 Total 19.4 110.8 12.8 163.3 66.8 99.7 Source: OECD (2018a). Note: Narrow environmental conditionality is based on OECD flags for environmental conditions on input use. Broader conditionality measures include condi- tions such as cross-compliance requirements in the EU. Coupled subsidies include payments based on outputs, payments based on inputs and payments based on activity levels such as area or livestock numbers. Decoupled subsidies are measures intended to avoid creating incentives to change output, specifically categories E (production not required); F (Payments based on non-commodity criteria) and G (miscellaneous payments) of the OECD (2016, p23). GSSE measures are payments designed to influence the environment for agriculture without pay- ments directly to farmers. For consistency, we use the OECD measure that fertilizer subsidies are zero in China, although some other sources identify these subsidies to fertilizer-producing enterprises as being a subsidy to farmers (Huang, Gulati and Gregory 2017). 12 Modeling the Impacts of Agricultural Support Policies on Emissions from Agriculture The first column of the table shows the value of subsidies The total amount of coupled subsidies in Column 4 of provided subject to environmental conditions on input use, Table 6 is heavily influenced by support to farmers in as designated by the OECD. This was only US$19.4 bil- China and India, because a large amount of this type of lion of total subsidies of US$163 billion. Using a broader support has been transferred to decoupled form in the measure of environmental conditionality that includes US and the EU (see Mamun, Martin, and Tokgoz 2019). broader requirements such as cross-compliance condi- There is some controversy about the extent to which tions in the EU and the USA results in US$110 billion of subsidies provided to Chinese agriculture are, in fact, support subject to some form of conditionality, over 80 coupled with output at the producer level as many were percent of which was in the US and the EU. This suggests originally introduced in ways that appeared to be lump that the coverage of conditionality is substantial, only the sum in nature (Huang et al. 2011). In India, as we will effectiveness of this conditionality remains a source of see, positive support from coupled subsidies operates in uncertainty. an environment where output prices have been depressed by border policies. This broader measure of conditionality was created by expert reviews5 of the extent of environmental condition- The final column of Table 6 shows that governments also ality in policies and countries covered by the OECD’s spent substantial amounts on support designed to improve Monitoring and Outlook Studies (2018a). It includes the enabling environment for agriculture, such as invest- all measures for which OECD’s “cookbook” or country ments in agricultural research and development and rural expert opinion confirmed that support under the speci- infrastructure. Many of these investments, and particu- fied policy is subject to environmental conditionality. The larly investments in agricultural research and develop- results in the table make clear that most of such support is ment, have very marginal high rates of return at current provided by the EU, with the United States as the second- investment levels (Alston et al. 2009, Zhang and Fan 2004, most important provider of this type of support. As we Fan, Cho and Rue 2018, Laborde et al. 2019), suggesting considered quite general conditionality schemes such as that the returns to greater investment in these activities the EU’s “cross-compliance” conditions (OECD 2018b), would be extremely high. By contrast, much of the ben- the resulting US$148 billion in support subject to envi- efit to society of coupled subsidies is lost through costly ronmental conditions is much larger than the $19 billion increases in production. While the measures reported indicated by OECD as being subject to input conditional- under GSSE clearly influence output levels and emission ity for environmental purposes. levels, they largely do so by overcoming market failures, such as under-investment by private agents in research Fertilizer subsidies amount to around US$13 billion per and development and/or provision of infrastructure. year in the table with 85 percent of the subsidies included While their impact on output is generally very substan- in the countries covered by the OECD accounted for by tial, their impact on producer returns and on emissions is India and most of the remainder by Indonesia. Expand- ambiguous, depending on factors such as the impact on ing the sample to countries not covered by the OECD output prices and output levels. would likely increase the importance of this measure, although it seems unlikely that it would become greatly more important as a share of overall agricultural support. The team undertook these reviews for all 26 economies covered by the OECD 5 Monitoring and Outlook dataset. Team members were Lars Brink, Joe Glauber, Will Martin, Valeria Pineiro and Simla Tokgoz. Agricultural Incentives 13 IMPORT MEASURES For consistency with the approach used in modeling trade flows in IFPRI’s global computable general equilibrium model, MIRAGRODEP, applied in this study, we use While the goal of the Uruguay Round of WTO negotia- the OECD measures of agricultural trade distortions to tions was to convert all agricultural trade measures affect- indicate the protection rate applying on non-preferential ing imports into tariffs (Hathaway and Ingco 1995), many trade flows, and then scale down protection on preferen- different—and frequently non-transparent—agricultural tial trade flows (Hoekman, Martin, and Braga 2009) in trade measures remain, including the Tariff Rate Quo- line with the tariffs on those flows6. This results in a data- tas introduced during those negotiations. This greatly base on protection that captures these two key dimensions complicates estimating the degree of agricultural protec- of global agricultural protection. Data from this database tion. Furthermore, successive rounds of preferential trade for import protection in broad groups of countries and for negotiations have resulted in very substantial differences select individual countries are given in Table 7. between the tariffs levied on different suppliers (Guim- bard et al. 2012). The pattern of import protection observed in Table 7 has potentially important implications for emissions. Of the Unfortunately, a key result of this is that standard tariff commodities with the largest emission intensities—beef, measures provide little guidance on the rate of protection milk and rice—milk and rice have particularly high aver- to agriculture or to individual agricultural commodities. age rates of protection. Average protection rates on beef Another challenge for measuring agricultural protection are substantially lower, at around 18 percent, but this rate is the presence of trade measures on exports. Of these, of protection is vastly higher than for less protected, low export taxes or export restrictions designed to lower emission commodities, such as vegetables and fruit, and domestic prices are the most common, although export wheat. On average, protection rates for emission-intensive subsidies designed to raise domestic prices have been commodities are much higher in rich countries than in widely used in agriculture in the past. developing countries. Rates of protection for these com- modities vary considerably between countries, with rates The primary approach that economists have followed to over 100 percent for rice in traditional high-protection deal with these problems is to estimate the average rate of countries such as Japan and the Republic of Korea, but protection or taxation by comparing the domestic price also in developing countries such as Colombia and the of a good with the price of a comparable traded good, Philippines. Rates of protection to beef are high in tradi- with both prices adjusted for any transportation or mar- tional high-protection economies such as the EU28, Japan keting costs between the points at which they are observed and Korea, but also in Indonesia, Russia and Vietnam. (OECD 2016). While this approach is suitable for measur- ing the average rate of protection for agriculture, it does not allow us to account for the presence of trade flows subject to protection rates that differ from the average— such as those within free trade areas or customs unions. To deal with this problem, we have combined data from the OECD database on agricultural support measures with data on preferential tariff rates obtained from the GTAP v10 database (see Aguiar et al. 2016 for the broad structure of this database). 6 More specifically, we use the ratio of the power of the tariff on import of the good from country i, (1+ti), to the power of the tariff from country j. 14 Modeling the Impacts of Agricultural Support Policies on Emissions from Agriculture TABLE 7. AGRICULTURAL IMPORT PROTECTION IN COVERED COUNTRIES, % Farm Milk Beef Pork/ Rice Sugar Veg/ Wheat Poultry Fruits World 16.3 26.2 17.8 19.8 27.5 26.5 8.3 7.3 Developed 18.9 39.5 23.0 29.3 39.0 20.6 8.8 2.9 Developing 14.7 21.7 13.5 12.4 25.1 29.3 7.6 8.0 Australia 0.0 0.1 Brazil 4.4 4.5 0.3 17.9 0.0 4.4 5.9 Canada 3.7 76.3 1.6 4.7 0.2 Chile 0.5 1.8 3.5 China 23.4 60.0 15.4 13.6 45.2 71.2 9.8 52.5 Colombia 25.7 30.5 4.1 27.8 170.6 34.6 1.6 9.8 Costa Rica 17.3 6.7 0.0 74.6 152.8 36.2 EFTA 33.0 39.9 73.1 103.7 8.2 18.9 7.5 46.7 EU28 6.8 0.6 34.3 4.9 16.4 15.3 7.9 2.1 Indonesia 18.8 4.8 30.7 53.8 73.4 69.8 2.4 India 20.5 21.5 8.5 2.1 6.0 7.1 21.7 18.9 Israel 14.6 53.2 22.4 12.8 2.1 11.5 Japan 41.6 107.2 38.5 52.0 220.5 35.1 50.9 Kazakhstan 6.2 0.9 19.2 18.9 1.1 24.7 1.7 0.1 Rep Korea 108.1 129.6 47.0 92.3 119.2 7.0 81.8 1.2 Mexico 2.4 4.7 4.4 5.4 14.8 0.8 New Zealand 0.8 3.3 2.1 Philippines 27.5 1.1 10.0 32.2 140.8 37.8 2.7 Russia 12.9 20.9 32.5 15.3 7.3 33.8 2.4 0.2 Turkey 19.9 4.2 144.9 38.8 19.2 14.1 10.7 Ukraine 4.5 7.3 2.0 10.4 4.7 44.0 0.8 5.4 USA 2.3 13.3 2.0 0.1 32.0 0.2 Vietnam 21.0 4.0 54.2 6.0 0.4 75.8 4.1 2.6 Regional aggregates Asia 14.0 23.9 8.7 5.0 6.9 64.3 5.9 3.4 Central America 28.6 61.1 6.3 52.2 31.9 7.9 9.6 0.0 Former Soviet U 5.6 6.4 7.7 9.1 3.8 3.5 6.4 0.5 Latin America 5.1 4.6 1.8 10.0 4.1 21.8 1.4 7.1 Middle E & N Africa 6.0 4.2 3.7 6.8 9.9 4.8 4.4 10.2 Sub-Saharan Africa 14.2 9.0 7.7 13.6 28.3 15.8 6.7 7.2 South Africa 5.5 6.4 0.1 36.5 5.2 Source: Authors’ computations based on OECD data. Note: Most blank cells refer to zero protection while a few refer to minor products not covered by the OECD database. Agricultural Incentives 15 EXPORT MEASURES measures in Table 7. However, they are non-trivial for several commodities and countries. On average, they are largest for India and for Ukraine, countries where policy In many countries, protection or taxation is observed appears to have reduced food prices to consumers at the for exported commodities. This implies the existence of expense of farm returns. India’s policies appear to have export taxes when domestic prices are artificially depressed imposed implicit export taxes on a wide range of com- and export subsidies when domestic prices are increased modities—including dairy, beef, oilseeds, vegetables and by the intervention. In these cases, we used the price gap fruit, rice and wheat. By commodity, export taxes are larg- estimated by the OECD (2018a) to create a measure of est for rice, with sizeable implicit export taxes in two major the implicit export tax or subsidy applying to those trade exporters—India and Kazakhstan. They are also sizeable flows. Results for these trade flows are given in Table 8. for oilseeds in many countries, sometimes because of a desire to subsidize oilseed crushing at the expense of oil- The export tax/subsidy measures presented in Table seed producers. 8 are, on average, substantially lower than the import TABLE 8. EXPORT TAXES AND/OR SUBSIDIES, % Farm Milk Beef Pork/ Rice Sugar Veg/ Wheat Poultry Fruits World 2.6 0.6 3.8 0.0 8.9 -0.5 1.4 3.7 Developed 0.2 -0.2 1.1 Developing 4.6 2.9 8.7 0.1 10.0 -0.6 1.9 16.8 Canada -0.2 -22.0 EFTA -1.4 -2.4 0.0 -0.4 Indonesia 0.2 9.1 India 15.3 25.1 37.3 11.1 -19.8 33.6 21.1 Israel 0.1 -0.2 Kazakhstan 7.4 40.9 2.4 Philippines 0.0 Russia 6.6 6.9 Ukraine 17.5 24.4 25.9 Vietnam 14.2 2.1 22.6 Regional aggregates Asia 4.4 0.3 15.0 Central America 0.5 5.8 Latin America 11.9 4.5 1.5 25.3 ME & N Africa 0.0 0.0 Sub-Saharan Africa 5.6 13.9 10.4 8.5 Source: Authors’ Calculations. Note: Most blank cells imply zero export tax rates. A few may refer to cases where the product was regarded as too minor to include in the OECD estimates of agricultural support. Either way, the blank cells can be ignored without great cost to the accuracy of the overall assessment. 16 Modeling the Impacts of Agricultural Support Policies on Emissions from Agriculture COUPLED SUBSIDIES VS partly because MPS includes both positive support from tariffs and negative support from direct or implicit export MARKET PRICE SUPPORT taxes. While the combination of positive and negative support contributes to higher economic costs from MPS The two components of all the support that influences measures than from subsidies (Anderson, Martin and output decisions are coupled subsidies and market price Valenzuela 2006), it results in a lower average rate of sup- support. Coupled subsidies generally tend to increase out- port from border measures than would otherwise be the put without lowering demand in the subsidizing countries, case. while market price support increases supply in protecting regions but, at the same time, reduces demand for agri- Also notable are the high rates of the protection in some cultural products in those countries by raising prices to cases, with countries like Iceland, Norway and Switzer- consumers. A rough indicator of the magnitude of each land, having much higher than average rates of both cou- of these supports is provided by dividing the value of pro- pled subsidies and border support. A second interesting ducer support provided by the value of output at world feature is the combination of positive protection from prices. coupled subsidies and negative protection from border measures in countries like India, Kazakhstan, Ukraine A surprising feature of Table 9 is the relatively small gap and Vietnam. These countries typically are net export- between the average rate of support provided by coupled ers—or potential net exporters—of certain key commodi- subsidies of 5.1 percent and the average support provided ties and apparently choose to use export barriers to lower by market price support (MPS) of 7.9 percent. This is the cost of food to domestic consumers. TABLE 9. COUPLED SUBSIDIES VS MARKET PRICE SUPPORT, 2014-16 (%) Coupled Subsidies Price Support Australia 1.1 0.0 Brazil 2.9 1.1 Canada 4.0 6.2 Switzerland 26.2 68.6 Chile 3.0 0.1 China 4.4 12.7 Colombia 4.0 14.8 Costa Rica 0.0 10.8 EU28 8.6 4.9 Indonesia 2.0 33.2 India 6.8 -13.2 Iceland 38.0 67.0 Israel 3.3 15.1 Japan 10.7 71.9 Kazakhstan 9.0 -1.9 Korea 4.8 88.8 Mexico 7.0 2.1 Norway 61.8 68.0 New Zealand 0.1 0.6 Philippines 0.0 31.0 Russia - 10.1 Continued on the next page Agricultural Incentives 17 Coupled Subsidies Price Support Turkey 6.4 29.5 Ukraine 3.5 -11.5 USA 4.6 2.8 Vietnam 1.3 -3.7 South Africa 1.2 2.1 Total 5.1 7.9 Source: OECD (2018a). Note: Both forms of support are expressed as a share of value of production at world prices. 18 Modeling the Impacts of Agricultural Support Policies on Emissions from Agriculture SECTION FOUR ANALYZING THE IMPACTS OF AGRICULTURAL SUPPORT To assess the impacts of current agricultural support, we examine the implications of moving from current support levels to a hypothetical situation in the absence of any global intervention. For this analysis, we use IFPRI’s MIRAGRODEP model (Laborde, Robichaud and Tokgoz 2013), which is an extension of the widely used MIRAGE model of the global economy (Decreux and Valin 2007). The underlying database used for the analysis is Prerelease 3 of the GTAP v10 database for 2014. The data on agricultural support were adjusted in line with the measures discussed in Sec- tion 3 for agricultural border measures and subsidies that influence output or input decisions (coupled subsidies). The model was augmented with a post-solution model based on the estimates discussed in Section 2 linking GHG emissions to output and inputs of agricultural activities determined in the model. The combined model was then used to assess the impacts of policy reform on emissions of CO2, CH4 and N2O, and these results combined to generate a total CO2 equivalent. The macroeconomic assumptions used for the analysis were designed to be relatively “neutral” to avoid situations where macroeconomic adjustments such as real exchange rate changes outweigh the impacts of interest, and to allow us to focus on the impacts of agricultural support policies on emissions. These assumptions were: 1. no dynamic effects of investment decisions (the static version of the model was used); 2. aggregate real public expenditures are kept constant and a consumption tax is adjusted to keep the government budget balance fixed as a share of GDP; 3. land use is constant to focus on emissions from agricultural production; 4. total employment is constant. Our approach of holding land use constant is consistent with many other studies in this area, such as Henderson and Lankoski (2019), and allows us to focus on changes in emissions from agricultural production, without needing to address the impacts of land use change, which are very context specific. Having estimates of the impacts on agricultural emissions is an important building block towards a full understanding of the impacts of reform. In this paper, we begin by considering the impact of removing coupled subsidies, and then turn to border measures. 19 IMPACTS ON EMISSIONS OF be close to one percent lower. In developed countries, farm output would be 1.7 percent less and, in develop- COUPLED SUBSIDIES ing countries, it would be 0.5 percent less. The smaller impact on developing country output reflects much lower To understand the impacts of current subsidies, we con- agricultural subsidies provided by governments of poorer sider in Table 10 the percentage changes in key farm developing countries (Anderson, Martin and Valenzuela outputs from removal of coupled subsidies. As discussed, 2006), but also the limited coverage of developing country we assess the impact through a counterfactual, that is, assistance in the OECD database on which we relied in what would be the level of output in the absence of those preparing this study. subsidies. Without subsidies, global farm output would TABLE 10. PERCENTAGE CHANGES IN OUTPUT FOLLOWING REMOVAL OF COUPLED SUBSIDIES, % Farm Beef Dairy Rice Pork/Poultry World -0.9 -0.7 -0.6 -0.9 -0.6 Developed -1.7 -1.1 -1.3 -0.3 -1.2 Developing -0.5 -0.2 0.3 -1.0 -0.3 Australia 1.7 1.4 1.1 1.0 0.5 Brazil 0.3 0.4 0.0 -1.5 1.1 Canada 1.7 1.7 0.2 0.3 1.7 Chile -1.0 -0.1 -0.1 0.4 -0.6 China -1.1 -0.1 0.1 -1.6 -0.4 Colombia 0.5 -0.1 -0.1 -0.1 -0.1 Costa Rica 0.5 0.0 0.3 0.4 0.0 EFTA -6.7 -8.9 -4.9 -5.3 -7.3 EU28 -3.4 -3.5 -1.5 -1.4 -2.5 Indonesia -0.3 -0.3 0.6 -0.4 0.1 India -1.7 -2.1 0.2 -2.9 -1.2 Israel 0.4 -0.6 -0.1 1.4 -0.5 Japan -2.9 -3.2 -3.8 -0.4 -0.3 Kazakhstan -0.8 -1.3 -0.1 0.1 -1.7 Rep Korea 0.0 0.1 0.5 -0.3 0.1 Mexico -3.3 -7.0 0.0 1.0 -4.2 New Zealand 1.3 2.4 0.9 0.9 2.0 Philippines 0.2 0.5 0.5 0.0 0.1 Russia -1.6 -0.5 -3.5 -0.8 -2.3 Turkey -3.9 -2.2 0.9 2.7 0.8 Ukraine 0.4 -2.5 0.2 1.0 -1.3 USA 0.0 0.1 -0.2 1.0 0.7 Vietnam 0.2 0.7 0.2 0.2 -0.2 Regional averages Asia 0.4 0.5 0.4 0.3 0.5 Continued on the next page 20 Modeling the Impacts of Agricultural Support Policies on Emissions from Agriculture Farm Beef Dairy Rice Pork/Poultry Central America 0.4 0.3 0.4 0.1 0.1 Former Soviet U 0.6 0.1 1.1 0.2 0.3 Latin America 0.5 0.3 0.2 0.4 0.0 Middle East & N Africa 0.7 0.8 0.3 0.9 0.1 Sub Saharan Africa 0.3 0.0 0.4 0.1 0.2 South Africa 1.0 0.2 0.4 0.0 0.8 Source: Authors’ model results. TABLE 11. IMPACTS ON EMISSIONS OF ABOLISHING COUPLED SUBSIDIES, KT OF CO2 EQUIVALENTA All Crop Enteric Manure Rice Synthetic Residues Fermentation Fertilizer World -34420 -2915 -6016 -3871 -1041 -10138 Developed -18116 -1079 -4107 -2987 -206 -4942 Developing -16304 -1836 -1909 -884 -834 -5197 Australia 1880 99 711 497 -1 280 Brazil 967 -62 886 538 -9 -337 Canada 1669 295 236 195 0 593 Chile -40 0 -9 -18 0 2 China -6011 -792 -45 -294 -459 -1758 Colombia 43 0 -15 -7 -3 11 Costa Rica 12 0 3 1 0 5 EFTA -1009 -15 -369 -258 0 -112 EU28 -17141 -1278 -4177 -3304 20 -4162 Indonesia -4 -32 -35 -7 -13 51 India -10459 -1032 -1989 -764 -266 -3085 Israel 12 1 -2 -3 0 5 Japan -1296 -99 -147 -85 -229 -305 Kazakhstan -237 -11 -88 -63 -28 -6 Rep Korea 29 -1 6 5 0 -1 Mexico -5640 -139 -2886 -1613 1 -317 New Zealand 692 2 350 249 0 63 Philippines 55 3 28 20 -9 9 Russia -2428 -10 -725 -491 0 -687 Turkey -1927 -248 -114 -88 -2 -901 Ukraine 282 108 -60 -70 0 203 USA -514 -72 8 204 4 -611 Vietnam 118 8 51 18 -4 34 Regional aggregates Asia 1540 60 718 412 -27 263 Central America 185 1 94 45 -2 23 Continued on the next page Analyzing the Impacts of Agricultural Support 21 All Crop Enteric Manure Rice Synthetic Residues Fermentation Fertilizer Former Soviet U 749 22 266 163 -1 142 Latin America 1123 136 468 207 -6 94 Middle East & N Africa 1616 112 291 232 1 237 Sub Saharan Africa 992 6 484 366 -7 91 South Africa 321 23 42 43 0 37 Source: Authors’ computations based on MIRAGRODEP simulation results. a Note: Total GHG emissions in “kt of CO2 equivalent” are composed of CO2 totals excluding short-cycle biomass burning (such as agricultural waste burning and Savannah burning) but including other biomass burning (such as forest fires, post-burn decay, peat fires and decay of drained peatlands), all anthropogenic CH4 sources, N2O sources and F-gases (HFCs, PFCs and SF6). Note: kt = thousand tonnes. The impacts of current coupled subsidies on emissions in countries and regions where agriculture was originally from agriculture are presented in Table 11. It shows that lightly protected, not protected, or subject to export taxa- without those subsidies and, hence, with lower output, tion. If emission intensities were the same in all regions, there would be less global emissions in the amount of 34 the only thing that would matter for global GHG emis- million tons of CO2 equivalent, or around -0.6 percent sions is the impact of reform on global output. However, lower. Interesting, the largest impact on emissions comes we know that emission intensities for agricultural prod- from fertilizer subsidies (explaining about one third of the ucts vary across countries. Frequently, but not always, they impact). Impacts of existing subsidies on enteric fermen- tend to be much lower in the rich countries than in poor tation form the second largest component. Without cou- countries (Mamun, Martin and Tokgoz 2019). Hence, we pled subsidies, emissions would fall the most in China, the also need to examine changes in output by country. EU, Mexico, and other countries that currently provide substantial subsidies to agriculture. Unsurprisingly also, A key result in Table 12 is that, in the absence of agri- countries providing little support through coupled subsi- cultural trade protection, global farm output would be dies, like Australia, the impact would be less and, if sub- 0.1 percent lower than the actual level. This result shows sidies would be removed, they would even see an increase that current agricultural trade protection helps expand in emissions as their output would increase in response global agricultural output only slightly. It pulls additional to improved farm competitiveness, owing to higher world non-land resources into the sector, but at the same time prices resulting from removal of subsidies elsewhere. resources are moved out of the sector elsewhere, that is, in countries with low or no protection facing lower world IMPACTS ON EMISSIONS OF prices. The effect is much smaller than for coupled subsi- dies, despite the greater magnitude of market price sup- BORDER DISTORTIONS port. This might, at first, seem surprising but reflects the fact that protection reduces consumption and raises farm Our first experiment in this section considers elimina- output in the countries with border measures. The effect tion of all agricultural border measures, both import and on global output is an empirical question, with the answer export. We first consider the impact on agricultural out- depending heavily on the importance of the protect- put levels by commodity, then the impact on global GHG ing countries as consumers relative to their importance gas emissions, to then turn to the impacts on emissions as producers. As noted in the introduction, if countries from agriculture by product and country. Table 12 pre- imposing protection were almost completely dependent sents the effects on farm output for agriculture overall and on imports for their consumption, the effect on world out- for key emission-intensive commodities. put would be much the same as with a consumption tax in those regions. Given that most of the high protection is in In the absence of trade protection, output levels would be net importing countries, it is perhaps not surprising that lower in initially protected regions, but would be higher the impact of removing border measures is close to zero. 22 Modeling the Impacts of Agricultural Support Policies on Emissions from Agriculture TABLE 12. OUTPUT CHANGES FROM ABOLITION OF ALL BORDER MEASURES, % All Farm Beef Dairy Rice Pork/Poultry World -0.1 -0.2 -0.9 -0.4 0.3 Developed -0.6 -3.0 1.4 -3.3 -1.3 Developing 0.1 2.7 -3.5 -0.2 1.3 Australia 20.9 31.2 40.2 22.3 2.7 Brazil 11.1 18.1 -1.1 -2.9 19.4 Canada -2.8 8.3 -46.6 0.1 13.2 Chile 5.6 9.9 10.1 0.8 8.9 China -3.6 -3.6 -39.6 -1.1 -0.1 Colombia -2.2 1.6 -8.4 -43.8 -1.4 Costa Rica -0.5 4.4 -2.8 -49.6 -22.1 EFTA -22.0 -36.5 -17.5 47.5 -39.8 EU28 -1.3 -12.8 7.1 -20.1 -2.1 Indonesia -5.2 -24.2 -0.8 -9.0 -5.0 India 2.4 32.0 3.5 6.2 0.7 Israel -6.0 -20.5 -19.2 106.9 -6.5 Japan -21.8 -32.6 -64.7 -4.4 -22.7 Kazakhstan -0.1 -2.5 -1.2 23.6 -15.7 Rep Korea -10.8 -18.3 -47.7 -13.4 -6.6 Mexico -2.2 5.0 -1.7 -2.4 -10.3 New Zealand 25.5 46.5 26.3 -33.2 -15.8 Philippines -5.0 -12.0 67.3 -27.4 -5.7 Russia -8.8 -11.7 -14.1 -11.5 -10.6 Turkey -14.9 -67.2 5.6 5.9 -11.8 Ukraine 16.9 40.7 21.7 -28.5 2.6 USA 4.9 3.8 8.7 29.6 4.8 Vietnam 5.0 -42.4 27.9 10.3 -2.2 Asia 3.5 -5.1 1.3 4.0 16.2 Central America 3.3 0.8 5.8 1.9 0.1 Former Soviet Union 2.2 16.7 -1.0 -4.9 -4.2 Latin America 5.2 18.9 5.6 1.0 1.5 Middle East & N Africa -0.2 -6.0 1.4 -4.2 0.4 Sub Saharan Africa 0.1 -0.6 0.5 -0.9 -0.1 South Africa 1.5 0.2 2.7 5.0 1.0 Source: Authors’ computations based on MIRAGRODEP simulation results. Note: Result are presented for overall farm output and selected commodities. Border measures include import tariffs, export taxes and nontariff measures applied at the border. Agricultural production patterns across countries change with a modest overall expansion in developed countries as output would be lower for most of the emission-inten- outweighing a small average decline in output in develop- sive commodities produced by high-income countries with ing countries. The combination of the expansion in over- relatively low emission-intensities, while output expands all farm output with the larger expansion of beef and milk in developing countries. This pattern is seen, broadly, for production appear to explain most of the impact of bor- beef, rice and pork/poultry. For milk it is slightly different, der measures on overall GHG emissions from agriculture. Analyzing the Impacts of Agricultural Support 23 Looking at the country-level changes, the results for over- The impacts of border measures on emissions are pre- all farm output follow an expected pattern, with lower sented in Table 14 and, as before, by looking at the simu- farm output in currently highly protected economies. lated level of emissions in the absence of agricultural trade such as Japan and Korea, and higher output in lightly pro- protection. The table shows an overall increase in global tected economies such as Australia, Brazil, New Zealand GHG emissions of 128 million tons of CO2 equivalent, and the Ukraine. For some individual emission-intensive or 2.1 percent, with a 1.9 percent increase in developed commodities, output levels would change substantially countries and a 2.2 percent increase in developing coun- without the border measures. For instance, in India there tries. This increase in emissions is despite lower global are production of beef (including buffalo meat) would be output (see Table 10) and indicates that in the absence 32 percent higher. Also output of key agricultural com- of current border measures GHG emissions from agricul- modities in Australia and New Zealand would be signifi- ture would in fact be higher, as a result of changes in the cantly higher. By contrast, output of all these products composition and location of output in ways that lead to would be lower in Japan and Korea, as expected. There higher emissions. are also considerable variations in outcomes by product Table 14 shows that about 90 percent of the higher emis- within countries, with Canada seeing a modest increase sions from agriculture would come from livestock-related in beef production, a large increase in pork and poultry emissions (enteric fermentation and manure) generated production and a decline of almost 50 percent in dairy in Australia, Brazil, India, the United States and other production. Latin America. These are the countries and regions where output of livestock products is expected to be higher in Table 13 shows the simulation estimates of the impact the absence of current border measures. In Brazil and of agricultural border measures on real national income India, higher emissions would mainly come from greater and farm income. For the world as a whole, both national output of beef, including buffalo (see Table 12). These income and farmer income would be higher in the higher levels of emissions explain almost all the increase absence of agricultural trade protection. Real national from developing countries, as well as most of the global income would be higher in all regions as greater efficiency increase in emissions. Higher beef production would would outweigh terms-of-trade deteriorations in import- result through different channels. The increase in emis- ing countries. Farmers in the most protected economies, sions from livestock production in Brazil reflects the lim- such as Japan, Korea and the EFTA countries, would ited protection provided to animal-source food products enjoy much lower incomes, while competitive exporters in Brazil and the expansion of the livestock sector in such as Australia, Brazil, Canada, New Zealand, and response to greater market opportunities that would be the United States would experience substantially higher created in the absence of import barriers. The increase in real farm incomes. For most other countries, the changes India reflects the removal of export barriers reported in in national farm income are relatively small, suggesting Table 6. In contrast to these cases, GHG emissions gener- that—where they arose—the biggest political challenges ated in China, the EU, Russia and Turkey would be lower, would likely involve balancing between gainers and losers, as in these cases output of livestock products would be less such as dairy farmers and other producers in Canada. in the absence of agricultural trade protection. 24 Modeling the Impacts of Agricultural Support Policies on Emissions from Agriculture TABLE 13. CHANGES IN REAL NATIONAL INCOME AND FARM INCOME IF ALL BORDER MEASURES ON AGRICULTURAL COMMODITIES WOULD BE REMOVED, % Real Income Real Farmer Income World 0.3 0.9 Developed 0.3 0.9 Developing 0.4 0.9 Australia 0.4 16.8 Brazil 0.4 6.6 Canada 0.3 2.4 Chile 0.0 3.5 China 0.6 -2.5 Colombia 0.5 0.7 Costa Rica 1.4 6.1 EFTA 0.6 -16.4 EU28 0.1 -0.1 Indonesia 0.5 -1.7 India 0.0 3.5 Israel 0.4 -4.9 Japan 0.6 -14.8 Kazakhstan 0.2 4.1 Rep Korea 3.2 -19.4 Mexico 0.2 -0.2 New Zealand 2.5 25.6 Philippines 1.2 -4.9 Russia 0.3 -0.4 Turkey 1.0 -3.8 Ukraine 1.2 33.9 USA 0.0 5.5 Vietnam 2.8 23.8 Asia 0.2 3.8 Central America 0.1 3.5 Former Soviet U 0.3 3.4 Latin America 0.3 6.0 Middle East & N Africa 0.1 0.2 Sub-Saharan Africa 0.1 0.4 South Africa 0.2 2.9 Source: Authors’ computations based on MIRAGRODEP simulation results. Note: border measures include price-based measures such as tariffs, export taxes, import subsidies, and nontariff measures that change the price of domestic goods relative to world prices. Analyzing the Impacts of Agricultural Support 25 TABLE 14. IMPACTS ON EMISSIONS OF ABOLISHING BORDER DISTORTIONS, KT OF CO2 EQUIVALENT Crop Enteric Synthetic Total Residues Fermentation Manure Rice Fertilizer World 127635 4129 91043 39624 -1193 1203 Developed 25597 3115 11644 9139 201 3042 Developing 102037 1013 79399 30486 -1394 -1839 Australia 26913 223 14756 9995 5 683 Brazil 66288 1174 40800 20208 -60 1688 Canada 838 62 -91 357 0 380 Chile 857 1 438 377 0 14 China -51961 -3029 -17404 -9390 197 -9002 Colombia -1448 -111 -494 -203 -179 -311 Costa Rica -8 -6 -8 -45 -3 19 EFTA -3391 -46 -1299 -978 0 -283 EU28 -17619 -570 -7612 -5016 -150 -2788 Indonesia -7669 -460 -3256 -2131 -673 -1047 India 61442 1639 38188 13270 600 4706 Israel -229 -4 -79 -87 0 -4 Japan -5188 -18 -1883 -1529 78 -43 Kazakhstan 129 255 -192 -163 8 152 Rep Korea -2421 -58 -707 -496 -36 -224 Mexico 2071 -26 1941 316 0 -73 New Zealand 13700 2 7121 4834 0 1189 Philippines -4124 -447 -712 -575 -1760 -679 Russia -6838 1580 -4461 -2858 -25 1092 Turkey -11865 -291 -4940 -4426 12 -773 Ukraine 9120 1941 2141 1083 -5 2284 USA 19603 1941 5819 4829 329 3036 Vietnam -4432 211 -3211 -1901 169 299 Asia -93 360 -3108 1090 430 255 Central America 1008 12 533 227 -2 132 Former Soviet U 8587 47 4687 3054 -11 331 Latin America 37646 -200 26226 11477 -43 -42 Middle East & -2615 -77 -1384 -1058 -36 -75 N Africa Sub-Saharan -1412 -47 -831 -698 -39 136 Africa South Africa 745 72 65 61 0 149 Source: Authors’ calculations. Notes: Only key sources are included, with smaller items such as Pesticides and Direct Energy use included in the total, but not presented separately. Crop Residues and Burning Crop residues have been combined to save space, as have Manure Management, Manure on Pastures and Manure on soils. 26 Modeling the Impacts of Agricultural Support Policies on Emissions from Agriculture JOINT IMPACT OF BORDER The finding that in the absence of both coupled subsidies and border protection GHG emissions from agriculture MEASURES AND COUPLED would be higher could be somewhat surprising and poten- SUBSIDIES tially controversial. It would imply that current agricul- tural support would in fact help lower emissions compared If both border measures and coupled subsidies on out- with a situation without such support. In general, there is put of key agricultural commodities would be removed at no strong reason to expect that measures introduced pri- the same time the impacts on output and emissions would marily for political-economy reasons, and, hence, unre- be close to the sum of the two components taken sepa- lated to climate change mitigation, would have either a rately. Table 15 shows this scenario with modest declines positive or negative impact on GHG emissions. One in overall output, and substantial reallocations of produc- might expect that abolition of such measures might—by tion from currently protected countries to those that are increasing overall economic efficiency—reduce emissions unprotected, or even taxed, rather than one of uniform to the extent that it leads to the adoption of lower-cost, shifts between developed and developing countries. Thus, and generally more efficient, approaches to production. output of beef would be substantially higher in Australia, Brazil, Canada, Colombia, India, New Zealand and However, as noted in Mamun, Martin and Tokgoz (2019), Ukraine, and drastically lower in Europe and Japan. compared with developing countries, agricultural pro- tection tends to be higher in developed countries, where Results for the impact on emissions of the abolition of emission intensities tend to be lower. Without existing both border measures and coupled subsidies are presented protection of the agricultural sector, by unwinding the in Table 16. These results reveal that the effects of abol- tendency to relocate production in the high-income coun- ishing market access barriers and coupled subsidies are tries, emissions could well increase as the model results somewhat different from the sum of their separate effects. show. While the results are much more specific than this, The result of these two reforms is an increase in emissions with sharp differences in emission intensities within both of 102 million tons of CO2 equivalent or 1.7 percent of the rich and the poor country groups, the mentioned the base level. Much of the increase is associated with the dichotomy explains most of the simulated global outcome. increases in output of bovine meat in Brazil and India. As pointed out earlier, these findings only refer to the As was the case with the removal of market access distor- impacts of government support on GHG emissions from tions, the largest influence on the outcome is an increase agricultural production. We do not account for changes in in emissions from enteric fermentation and from livestock land use caused by the incentives to agricultural produc- manure. The biggest contribution comes again from tion. Outcomes could change significantly when consid- increased emissions in Brazil and India, along with smaller ering land use and deforestation. Hence, this is important increases in countries such as Australia, New Zealand and priority for further research, both in the sense of how cur- the United States. The result is ameliorated to a modest rent agricultural support may influence emissions through degree by reductions in emissions from synthetic fertilizers impacts on land use, as well as of how subsidy reform and rice production, primarily in developing countries. could redirect incentives such that they promote agricul- tural productivity growth through sustainable intensifica- tion and reducing the ecological footprint of agricultural land use. Analyzing the Impacts of Agricultural Support 27 TABLE 15. OUTPUT CHANGES FROM ABOLITION OF ALL BORDER MEASURES & COUPLED SUBSIDIES, % Farm Beef Dairy Rice Pork/Poultry World -1.1 -0.8 -1.4 -1.4 -0.3 Developed -2.4 -4.2 0.0 -3.4 -2.6 Developing -0.4 2.8 -3.0 -1.2 1.2 04_OECDSample -2.1 -2.5 -2.3 -3.0 -1.2 Australia 23.2 33.3 42.8 24.2 3.2 Brazil 11.7 19.8 -1.1 -4.5 21.3 Canada -1.0 10.5 -46.2 0.5 15.6 Chile 4.7 10.6 10.2 1.4 8.1 China -4.7 -3.5 -39.2 -2.9 -0.5 Colombia -1.7 1.6 -8.2 -43.9 -1.5 Costa Rica 0.0 4.7 -2.5 -48.6 -22.1 EFTA -30.0 -45.9 -24.9 41.9 -46.1 EU28 -5.1 -17.2 5.2 -20.8 -5.0 Indonesia -5.5 -24.4 0.0 -9.4 -4.9 India 1.0 29.6 3.8 2.9 -0.6 Israel -5.3 -20.9 -18.1 112.0 -7.5 Japan -24.5 -35.2 -67.4 -4.8 -22.6 Kazakhstan -0.8 -3.7 -1.3 24.8 -18.0 Rep. Korea -10.5 -17.9 -46.6 -13.4 -6.2 Mexico -5.7 -2.6 -1.6 -1.3 -14.9 New Zealand 27.3 51.3 27.2 -33.0 -15.4 Philippines -4.3 -10.8 75.0 -27.2 -5.5 Russia -10.4 -12.1 -18.0 -11.9 -13.1 Turkey -18.9 -69.4 7.7 9.5 -10.2 Ukraine 16.7 32.3 22.2 -27.0 0.4 USA 5.1 3.9 8.7 31.0 5.6 Vietnam 5.3 -41.8 28.6 10.6 -2.3 Asia 4.1 -4.6 1.8 4.5 17.6 Central America 3.8 1.1 6.4 2.0 0.2 Former Soviet Union 2.8 17.4 0.0 -4.8 -4.1 Latin America 5.9 20.1 6.2 1.4 1.6 Middle East & N Africa 0.5 -5.0 1.7 -3.0 0.5 Sub-Saharan Africa 0.4 -0.6 0.8 -0.6 0.0 South Africa 2.5 0.4 3.4 4.9 1.7 Source: Authors’ computations based on MIRAGRODEP simulation results. 28 Modeling the Impacts of Agricultural Support Policies on Emissions from Agriculture TABLE 16. IMPACTS ON EMISSIONS OF ABOLISHING COUPLED SUBSIDIES & BORDER DISTORTIONS, KT OF CO2 EQUIVALENT Crop Enteric Synthetic Total Residues Fermentation Manure Rice Fertilizer World 102071 1257 88780 37691 -2331 -7511 Developed 7590 1728 7529 6086 -33 -1811 Developing 94481 -471 81251 31605 -2298 -5700 Australia 28986 282 15683 10636 5 884 Brazil 71705 1207 44319 21961 -72 1651 Canada 2168 249 208 618 0 744 Chile 837 0 449 364 0 14 China -58045 -3870 -17224 -9508 -329 -10944 Colombia -1381 -111 -483 -200 -181 -305 Costa Rica 5 -6 -3 -43 -3 23 EFTA -4146 -57 -1571 -1169 0 -370 EU28 -36517 -1814 -12817 -9064 -140 -6948 Indonesia -7644 -492 -3270 -2118 -676 -1008 India 52513 777 35948 12410 236 2366 Israel -212 -3 -79 -88 0 0 Japan -5353 -16 -1858 -1505 -171 -28 Kazakhstan -103 241 -280 -225 -17 133 Rep Korea -2359 -59 -693 -481 -36 -223 Mexico -4039 -160 -1239 -1456 0 -381 New Zealand 14796 3 7673 5218 0 1307 Philippines -3978 -441 -647 -532 -1756 -668 Russia -8786 1460 -4926 -3244 -25 526 Turkey -12220 -308 -5158 -4606 5 -789 Ukraine 9092 2011 1950 987 -5 2429 USA 18800 1680 5830 5079 333 2298 Vietnam -4307 223 -3166 -1881 179 328 Asia 1663 421 -2304 1703 444 399 Central America 1232 13 660 288 -3 149 Former Soviet U 9488 66 5092 3319 -12 421 Latin America 40510 -94 27978 12246 -47 18 M East & -1192 1 -1046 -784 -30 99 N Africa Sub-Saharan -473 -36 -354 -334 -32 185 Africa South Africa 1030 90 107 102 0 179 Source: Authors’ computations based on MIRAGRODEP simulation results. Analyzing the Impacts of Agricultural Support 29 REMOVING ALL PRODUCTION 15, where existing support is removed without compensa- tion to those producers who lose, or taxation of the win- SUBSIDIES AND MARKET ners, this scenario yields higher farm output worldwide, PRICE SUPPORT BUT driven by higher output in developed countries. Output, MAINTAINING DIRECT INCOME but generally lower output in developing countries. In SUPPORT TO FARMERS lightly protected countries such as Australia and Brazil, farm output is only slightly bigger than under a scenario without any form of agricultural support. In the latter Two simulations were performed for this assessment of case, higher world prices would induce more production the implications of removing support to farm production taking place in these countries. Economies like Canada while compensating those farmers who lost income from and Japan, with wide variations in agricultural support this change. The first element replicates the scenario pre- rates, would see sharp declines in output from highly sup- sented in the previous section, but here we combine this ported activities such as dairy production. with the provision of lumpsum subsidies for losers and a tax increase for winners, such as to maintain average farm The impact on emissions associated with this scenario incomes at their initial level. This second element of this are presented in Table 18. Emissions from agricultural experiment involves uniform subsidies to maintain farm production are higher compared with the baseline case incomes in countries where subsidies and rates of protec- of current levels of support. However, the difference in tion were initially high, and uniform taxes in countries emission increase is much less (about half) compared with where production support and protection was initially the scenario in which all support is removed (as reported low. Although the simulation using uniform taxes/trans- in table 16). Hence, the present scenario would yield a fers to maintain farm incomes at their initial levels is quite smaller increase in emissions, despite the greater overall hypothetical, it produced results that are quite informative increase in global farm output. As is evident in Table 18, about channels of impact. this results mainly from a smaller increase in emissions from livestock (enteric fermentation and manure), particu- The impacts of this counterfactual scenario on output larly in Australia and Brazil. Those two countries would by main agricultural commodity are shown in Table 17. see large increases in livestock output and emissions from Compared with the scenario results presented in Table beef under the scenario when all support is removed. 30 Modeling the Impacts of Agricultural Support Policies on Emissions from Agriculture TABLE 17. OUTPUT IMPACTS OF REMOVING ALL SUPPORT WITH UNIFORM TRANSFERS TO MAINTAIN FARM INCOMES, % All Farm Beef Dairy Rice Pork/Poultry World 0.7 0.3 0.8 2.7 0.9 Developed 2.0 -1.1 4.9 2.7 8.0 Developing 0.0 1.8 -3.9 2.6 0.2 Australia 0.8 1.9 7.9 -9.6 3.2 Brazil 0.0 3.9 -6.8 1.2 -9.5 Canada -3.8 7.1 -48.3 7.3 0.1 Chile 1.8 8.8 6.4 2.5 0.4 China -0.5 1.5 -31.2 4.3 2.6 Colombia -2.4 1.6 -8.9 -1.7 -44.2 Costa Rica -4.7 -1.7 -9.8 -27.5 -55.6 EFTA 24.8 -6.3 45.1 16.9 375.0 EU28 2.0 -8.5 14.4 1.4 -10.6 Indonesia -5.7 -23.0 -3.1 -5.9 -9.4 India 0.2 28.1 3.0 -0.9 1.6 Israel -2.1 -9.9 -22.9 -4.4 215.6 Japan 2.5 19.2 -35.2 21.3 13.0 Kazakhstan -2.5 -4.8 -2.5 -19.7 19.8 Rep. Korea 15.0 8.3 -21.3 17.7 -7.0 Mexico -1.4 5.2 2.1 -8.7 4.7 New Zealand -1.4 24.6 -9.5 -24.3 -48.4 Philippines -0.5 -4.1 65.0 -1.1 -19.7 Russia -7.8 -7.4 -18.3 -10.6 -9.6 Turkey -5.2 -55.5 17.3 21.3 41.0 Ukraine -8.2 -12.7 -16.7 -33.4 -53.5 USA 3.0 2.4 4.4 2.9 28.1 Vietnam -5.6 -50.4 -19.3 -10.5 -3.2 Asia 3.0 -4.6 0.1 12.4 4.1 Central America 3.1 1.1 3.4 -0.3 2.1 Former Soviet Union 1.8 15.5 -2.4 -5.9 -5.4 Latin America 5.3 19.3 4.2 0.9 2.1 M East & N Africa -0.4 -5.3 -0.8 -1.2 -2.8 Sub-Saharan Africa 0.2 -0.6 -1.5 -0.7 -0.5 South Africa -2.0 -2.0 -2.2 -3.0 -6.8 Source: Author’s computations based on MIRAGRODEP simulation results. Analyzing the Impacts of Agricultural Support 31 TABLE 18. IMPACTS ON EMISSIONS OF REMOVING ALL SUPPORT WITH COMPENSATION, KT OF CO2 EQUIVALENT All Crop Enteric Ferm Manure Rice Synth Fert Residues World 56232 1642 47711 19126 -2267 -5965 Developed 10968 2664 2530 3671 -14 -512 Developing 45264 -1022 45181 15455 -2252 -5453 Australia -260 -54 51 -266 3 -86 Brazil 8823 226 6395 2987 -45 -404 Canada 767 258 -309 -14 0 742 Chile 516 -3 334 199 0 0 China -25295 -1888 -8295 -847 -17 -6312 Colombia -1542 -112 -534 -223 -180 -347 Costa Rica -213 -8 -131 -105 -3 14 EFTA 1605 -26 790 558 0 -210 EU28 -2412 -760 1246 1391 -110 -3415 Indonesia -9302 -682 -4094 -3069 -1054 -212 India 49470 524 35304 12225 325 816 Israel -67 0 -42 -50 0 14 Japan 611 4 196 652 -165 -105 Kazakhstan -421 184 -389 -299 -18 113 Rep Korea 3170 -34 219 416 -32 -155 Mexico 2147 -104 2198 529 1 -195 New Zealand 2805 -5 1318 824 0 633 Philippines -2469 -306 -251 -177 -1363 -475 Russia -7308 1648 -4391 -2744 -23 522 Turkey -6283 -239 -2962 -2264 24 -914 Ukraine -104 1398 -1281 -1578 -7 1465 USA 11990 1631 3411 2854 313 1561 Vietnam -8332 -237 -4139 -2922 -245 -459 Asia -1719 343 -3949 74 429 447 Central America 926 14 469 193 -1 142 Former Soviet U 7480 62 3975 2562 -12 454 Latin America 38994 -194 27084 11827 -28 33 M East & -2727 -27 -1672 -1393 -27 50 N Africa Sub-Saharan -4389 -34 -2608 -1998 -30 197 Africa South Africa -229 59 -233 -216 0 120 Source: Author’s computations based on MIRAGRODEP simulation results. 32 Modeling the Impacts of Agricultural Support Policies on Emissions from Agriculture PRODUCTIVITY increasing resilience might not yield similarly large gains in those dimensions. IMPROVEMENTS Clearly, there are many ways in which agricultural emis- The preceding analysis showed that present agricultural sions per unit of output might be reduced. If, for instance, support measures, especially coupled subsidies, raise some emissions were linked to the amount of land used, and environmental concerns as they raise GHG emissions. output per unit of land increased, then emissions per However, it should be recognized that, in most countries, unit of output would decline. Alternatively, an efficiency those subsidy policies are largely based on political-econ- improvement that reduced the amount of inputs such as omy considerations rather than for concerns about climate feed needed per unit of output would result in a decline change. Proper design of policy for effectiveness requires in emissions per unit of output where emissions are linked satisfying Mundell’s (1960) principle of assigning individ- to the feed input. By contrast, some types of technical ual policies to the goal that they address most directly. We change are strongly linked to emission reductions. will address options for repurposing agricultural support measures aligned with environmental objectives in subse- Gerber et al. (2013, pp 48-50) provide a range of quent research, but here we wish to put the environmental approaches strongly focused on emission reductions from impacts of current policies in perspective by looking at a livestock, involving a range of strategies such as using feed form of support that impacts more directly on resource additives, manipulating rumen behavior, reducing stor- use. In this vein, we also investigated in the context of age time for manure, selecting genetically improved ani- this study the effects of public support for investments mals and increasing livestock fertility rates. Some of these that would directly increase agricultural productivity or options (such as increasing fertility) both reduce emissions reduce emissions from agricultural production. and increase productivity, while others (such as feed addi- tives designed purely to reduce emissions (e.g., Mernit In the context of this study, it is important to consider 2018) may have a relatively small impact on productiv- both the impact of R&D on productivity and on emis- ity. Since emission of a powerful source of energy such sions. A striking feature of the emission intensity data as methane is clearly wasteful, there must be a presump- provided through FAOSTAT is that the average emission tion that innovations that reduce this waste by converting intensity of livestock products is generally much lower in it into valued products such as milk or meat will at least OECD countries than in non-OECD countries (Mamun, partially pay for themselves. Even if these efficiency gains Martin and Tokgoz 2019; Tubiello 2012). This suggests are not enough to make use of this technology privately that higher productivity tends to lower the emission inten- profitable, they may greatly reduce the support needed to sity of agriculture. One potentially important approach adopt them, perhaps as part of a package involving condi- to reducing emissions from agriculture is through invest- tional support in return for use of these inputs. ments designed to increase productivity and reduce greenhouse gas emissions from agriculture. Investments in A similar range of strategies can be envisaged for crop agricultural R&D that raise agricultural productivity also production. For rice, Neue (1993) identifies a range of have the key advantage of being a powerful instrument strategies for reducing emission intensities. Millar et al for poverty reduction (Ivanic and Martin 2018). They (2014) identify a similar range of strategies for reducing can also contribute to developing a more sustainable and N2O emissions from other cereal crops, using approaches resilient agriculture with a smaller land footprint (Shime- such as selecting the right application rate, using the right les et al. 2018). Traditionally, agricultural research and fertilizer formulation, and optimizing the timing and development has been focused strongly on lowering costs placement of fertilizers. The alternate wetting and drying of production, and has been highly successful in doing (ADW) approach for rice is seen as reducing water use, that (Alston et al. 2009), but there seems little reason that raising productivity and reducing emissions (Lampayan et broadening this focus to include reducing emissions and al. 2015; Allen and Sander 2019). Analyzing the Impacts of Agricultural Support 33 Considerable care needs to be taken in specifying the The second type of policy preserves the gains from reduc- nature of productivity change when examining its impact tions in emission intensity but faces the risk that it will not on GHG emissions. If emissions are linked to input- be sufficiently attractive to producers for it to be widely use—which is likely for important emission categories adopted. The two cases can perhaps be most usefully such as enteric fermentation in cattle—then widely- thought of as limiting cases that bound a wide range of used approaches to specifying productivity growth such intermediate possibilities. as purely factor-augmenting technical change will not capture such reductions in input intensity. Instead, one By way of simple experiments, we considered (i) a pro- would need a form of productivity gain that saves both on ductivity change that reduced emissions by 30 percent the use of factors and intermediate inputs. Productivity without reducing intermediate or factor input demands, changes of this broad type are well known in the litera- and (ii) a productivity change (MFP) that reduced both ture (OECD 2001). While such productivity changes need the factor and intermediate input demands for all agricul- special weights when index approaches are being used to tural sectors by 30 percent. The 30 percent was chosen for measure implications for national productivity, this prob- illustrative purposes because it is widely regarded as being lem does not arise when model-based approaches are in the order of magnitude of gains feasible in important being used. areas such as reducing generation of greenhouse gases by enteric fermentation (Mernit 2018). The results for As it would be impossible to analyze all possible forms percentage reductions in emissions under these scenarios of productivity change that might apply, we consider two with support unchanged are presented in Table 19. relevant polar cases: The results in the first column of Table 19 reflect the 1. Productivity change that saves all factors and pure emission reductions, not associated with any other intermediate inputs, and increase in productivity. For the set of countries covered 2. Productivity change that reduces emissions with- by the OECD analysis and, for comparability, by our out substantially increasing productivity. productivity simulations, the reduction in emissions from agriculture is 28.8 percent, very close to the 30 percent The first type of productivity change will reduce emis- reduction in the emission intensity of each activity. The sion intensities roughly in proportion to the increase in small discrepancy reflects the presence of some emis- productivity. Its effect on overall emissions is, however, sions that are linked to the use of peat land, without being likely to be influenced by the so-called rebound effect, linked to agricultural output. The reduction in emissions where the reduction in the cost of the good resulting from for developed countries is almost as large as for the OECD the productivity increase raises demand for the good and sample, at 27.7 percent, reflecting (slightly) incomplete hence reduces the benefit in terms of emission reduction. coverage of developed countries by the OECD. Globally, Freire-Gonzalez and Puig-Ventosa (2015) examine the emissions are reduced by 19.5 percent, rather than 30 importance of this phenomenon and advocate combin- percent, primarily because of the incomplete coverage of ing productivity change and energy taxes to minimize it. world agriculture by the OECD sample. TABLE 19. REDUCTIONS IN EMISSIONS FROM AGRICULTURE FROM PRODUCTIVITY SHOCKS (% CHANGE IN CO2 EQ) Pure Emission-Reducing Factor and Input Saving World -19.5 -9.5 Developed -27.7 -11.2 Developing -17.1 -8.9 OECD Sample -28.8 -8.6 Source: Author’s computations based on MIRAGRODEP simulation results. 34 Modeling the Impacts of Agricultural Support Policies on Emissions from Agriculture The second column refers to the case where emission livestock sector—increases the supply of livestock prod- intensities of each activity are also reduced by 30 percent, ucts. Another cause is the double impact of productivity but these reductions are accompanied by savings of fac- change on output—a 30 percent increase in productiv- tors and intermediate inputs. In this situation, there are ity increases output for unchanged inputs by 30 percent two determinants of emission outcomes—the reduction and increases the incentive to use additional inputs in this in emissions associated directly with the reduction in the activity (Martin and Alston 1997). At the same time, how- emission intensity of each activity, and the increase in out- ever, the increase in the efficiency of feed use in the live- put associated with the famous “rebound” or Jevons effect stock sector reduces the demand for feed from maize and associated with the increase in productivity. This effect similar commodities, which end up expanding less than seemed surprisingly large given the relatively low elastici- other crop and livestock activities. This result illustrates ties of demand for agricultural products in the model, so the complexity of productivity increases of this type— we investigated the changes in output by sector associated that save both inputs and factors—because of their com- with it. The associated changes in output are given in pounding effect along value chains. Table 20. The large changes in output occurred despite large falls in world prices of agricultural products (22.5 Another striking feature of the table is the very large percent on average across all farm products) that were increases in output of agricultural fibers. This reflects the needed to offset the large increase in the effective price of competition between agricultural fibers such as cotton and agricultural products associated with the increase in pro- synthetic fibers in the production of textiles. There are ductivity (see Martin and Alston 1997). also important, but relatively smaller, differences in output effects between crops and livestock. Livestock output rises The OECD sample column is easiest to interpret because somewhat more, partly because the elasticity of demand our productivity simulation affects all the countries for livestock products is larger, and partly because of the included in it. The overall increase in farm output is strik- increase in the productivity of the feed-producing sector. ingly large given the 30 percent cost reduction and the Despite the larger increase in output from the relatively relatively low elasticities of demand for most agricultural emission-intensive livestock sector, agricultural emissions products. One cause of this large increase in output is the fall by 9.6 percent because of the large savings in emis- increase in the efficiency of grain production, which— sions within these sectors. in addition to the increase in productivity within the TABLE 20. PERCENTAGE CHANGES IN OUTPUT BY SECTOR AND REGION, 30% MFP INCREASE World Developed Developing OECD Sample Dairy 17.4 26.7 7.8 28.4 Fibers 23.0 49.2 18.3 39.4 Maize 8.0 7.8 8.1 13.6 Beef 15.4 22.0 9.3 23.9 Pork & Poultry 22.7 25.4 21.0 29.5 Other Crops 12.7 22.7 7.2 23.2 Rice 15.4 21.1 14.8 26.9 Wheat 9.8 24.1 2.3 23.0 All Crops 14.0 21.7 11.9 24.5 All Farm 16.4 23.9 12.8 26.3 All Livestock 19.0 24.9 14.1 27.8 Source: Author’s computations based on MIRAGRODEP simulation results. Analyzing the Impacts of Agricultural Support 35 The very large reductions in emissions associated with pure emission-reducing innovations (see Table 15) rela- tive to those associated with productivity-enhancing innovations that only incidentally reduce emissions have important policy implications. While raising productivity in ways that reduces input use can lower emissions, it is clear from this analysis that the rebound effect of this type of productivity increase is quite large. This suggests that research that focusses primarily on emission reduction could be quite important in achieving overall reductions in emissions. The result for productivity increases that save on all inputs will change once the analysis can be extended to include changes in land use change. If land-saving technology leave agricultural use. When productivity gains allow for enables reductions in the global agricultural footprint, it avoided deforestation, the gains will be larger. By contrast, will be possible to harvest gains from carbon sequestration the gains from pure emission-reducing improvements in that are in addition to those included here—although the technology are unlikely to change nearly as much as those extent of these gains will depend on the types of land that from productivity gains that save all inputs. 36 Modeling the Impacts of Agricultural Support Policies on Emissions from Agriculture SECTION FIVE CONCLUSIONS The analysis presented in this paper examines the implications of current levels of sup- port to farmers on global agricultural emissions. To assess this impact, we examined the impact of eliminating the current measures. In this initial assessment, we focused only on emissions from agricultural production, and constrained land use to remain constant. We built on the OECD measures of agricultural incentives, and developed a new database and model relating emissions to agricultural outputs that accounts for the specificity of emissions by production sector and by source of emissions. Our analysis leads to the conclusion that current subsidies coupled to production induce both higher global agricultural output and emissions. We quantified this impact by simulating a scenario where these measures are removed worldwide. Without the subsidies the global farm output volume would be 0.9 percent lower. The reduction in output would be smaller for some of the most emission-intensive commodities, such as beef (-0.7 percent) and dairy products (-0.6 percent). The CO2 equivalent of global emissions from agriculture would be 0.6 percent lower in this scenario, partly because of the smaller declines in output of the most emission-intensive products and partly because the relocation of production in some cases resulted in increases in the emis- sion intensity of production. When both coupled subsidies and border measures would be removed, global farm output would drop further (by 1.1 percent), driven by lower output in the developed countries. Paradoxically, despite the bigger decline in output, global emissions would increase in this scenario as a result of production shift from relatively low-emission- intensity countries to somewhat higher emission intensity countries, such as Brazil. The increase in emissions would be smaller, however, than in the case when only the border measures would be abolished. This outcome is driven by the elimination of the coupled subsidies that incentivize emission-intensive agricultural activities. The upshot is that, on balance, current agricultural subsidies and trade protection as such do not drive up GHG emissions from agricultural production. The subsidies do, but the trade protection does not. It should be noted that the impacts on emissions of current agri- cultural support measures are small relative to emission levels. Hence, eliminating all 37 support would do little for climate change mitigation; it from the productivity-increasing simulations because the might even increase emissions from agriculture. productivity increases generated a substantial rebound effect with especially large increases in output in livestock These findings are preliminary and further research is sectors. Future work involving land use change is likely to needed to understand the true impacts, especially since the provide more favorable impacts for productivity changes present scenario analysis did not consider in the impacts that save all inputs because of the consequent reductions on land use change. Furthermore, the findings should not in land needed for agriculture. be taken as conclusions about the effectiveness (or lack thereof) of current agricultural support policies. Present The results of this work need to be interpreted with con- agricultural subsidy policies in most countries are largely siderable caution. Given the time constraint under which based on political-economy considerations and rarely for this report was prepared, we were forced to rely almost their impacts on GHG emissions. Proper assessment of entirely on the measures of agricultural support provided policy effectiveness requires assigning policies to the goal by the OECD. While these measures cover a very large that they are to pursue most directly. share of global output, they omit some important agri- cultural producers such as Argentina and many devel- In this vein, we also investigated the effects of investments oping countries whose collective output is considerable. that increase agricultural productivity or reduce emis- Extending the coverage of these measures is an important sions from agricultural production. A key feature of the priority for future research. Another important priority productivity-increasing simulations is that they involve for the future is to comprehensively consider the impacts saving of intermediate inputs as well as factors, since of reforming agricultural incentives on land use change pure factor-augmenting technological change would not and particularly the substantial emissions associated with generally reduce emission intensities. The productivity- deforestation. Finally, our tentative conclusion that simply increasing technological changes were compared with reducing current agricultural distortions has a relatively experiments in which the emission-intensities of products limited impact on emissions—and may even increase were reduced by the same 30 percent without increases them—points to a need to investigate the use of the mul- in productivity. The pure reductions in emission intensi- tiple instruments needed to achieve multiple goals such ties generated much greater reductions in emissions from as economic efficiency, emission reduction, resilience and agriculture that were roughly three times as large as those poverty reduction. 38 Modeling the Impacts of Agricultural Support Policies on Emissions from Agriculture REFERENCES Aguiar, A., Narayanan, B. & McDougall, R. (2016), ‘An Decreux, Y. and Valin, H. (2007), ‘MIRAGE, Updated Overview of the GTAP 9 Data Base.’ Journal of Global Eco- Version of the Model for Trade Policy Analysis Focus on nomic Analysis 1(1): 181-208. Agriculture and Dynamics’ CEPII Working Paper 2007- https://jgea.org/resources/jgea/ojs/index.php/jgea/ 15, CEPII, Paris. article/view/23 Edwards, M. and Trancik, J. (2014) ‘Climate Impacts Allen, J. and Sander, B. (2019), The Diverse Benefits of Alter- of Energy Technologies Depend on Emissions Timing’ nate Wetting and Drying (AWD). International Rice Research Nature Climate Change 4(5): 347–52. Institute, Los Baños, Philippines. www.ccafs.cgiar.org. Fan, Shenggen, Cho, E. and Rue, C. (2018) ‘Returns to Alston, J. Andersen, M., James, J. and Pardey, P. public agricultural and rural investments in China’ China (2011), ‘The Economic Returns to US Public Agricul- Agricultural Economic Review 10(2):215-23, tural Research’ American Journal of Agricultural Economics https://doi.org/10.1108/CAER-11-2017-0211 93(5):1257-77. FAO (2017), Milk and Milk Products, FAO Outlook, Food Alston, J., Chan-Kang, C., Marra, M., Pardey, P. and and Agriculture Organization, Rome Wyatt, T. (2000a) A Meta-Analysis of Rates of Return to http://www.fao.org/fileadmin/templates/est/COMM_ Agricultural R&D, Ex Pede Herculem? IFPRI Research MARKETS_MONITORING/Dairy/Documents/ Report 113, International Food Policy Research Institute, Food_Outlook_June_2017__Dairy_.pdf Washington DC. FAO (2019), FAOSTAT Emissions Database, Agriculture, Alston, J., Marra, M., Pardey, P. and Wyatt, T. (2000b), http://www.fao.org/faostat/en/#data/GT ‘Research Returns Redux: a Meta-analysis of the Returns to Agricultural R&D’ Australian Journal of Agricultural and Fellmann, T., Witzke, P., Weiss, F.,Van Doorslaer, B, Resource Economics 44:(2):185-215. Drabik, D., Huck, I., Salputra, G., Jansson, T. and Leip, A. (2018), ‘Major Challenges of Integrating Agriculture Alston, J., Pardey, P., James, J. and Andersen, M. (2009), into Climate Change Mitigation Policy Frameworks’ Miti- ‘The Economics of Agricultural R&D’ Annual Review of gation and Adaptation Strategies for Global Change 23:451–468. Resource Economics 1:537-66. DOI 10.1007/s11027-017-9743-2 https://doi.org/10.1146/annurev.resource.050708.144137 Freire-González, J. and Puig-Ventosa, I. (2015), ‘Energy Anderson, K., Martin, W. and Valenzuela, E. (2006), ‘The Efficiency Policies and the Jevons Paradox’ International relative importance of global agricultural subsidies and Journal of Energy Economics and Policy 5(1):69-79 market access’ World Trade Review 5(3): 357–76. Gerber, P.J., Steinfeld, H., Henderson, B., Mottet, A., Boadi, D., Benchaar, C., Chiquette, J. and Massé, D. Opio, C., Dijkman, J., Falcucci, A. and Tempio, G. (2013), (2004), ‘Mitigation strategies to reduce enteric methane Tackling climate change through livestock – A global assessment of emissions from dairy cows: Update review’ Canadian Jour- emissions and mitigation opportunities. Food and Agriculture nal of Animal Science 84: 319–335. Organization of the United Nations (FAO), Rome. 39 Guimbard, H., Jean, S., Mimouni, M. and Pichot, X. Huang, Jikun, Wang, Xiaobing, Zhi, Huayong, Huang, (2012), ‘MACMAP-HS6 2007, An Exhaustive and Con- Zhurong and Rozelle, Scott (2011) ‘Subsidies and distor- sistent Measure of Applied Protection in 2007’ Economie tions in China’s agriculture: evidence from producer-level Internationale 2(130): 99-121. data’ Australian Journal of Agricultural and Resource Economics 55:53–71. Hathaway, D. and Ingco, M. (1995), ‘Agricultural liberali- zation and the Uruguay Round’ in Martin, W. and Win- Ivanic, M. and Martin, W. (2018), ‘Sectoral Productiv- ters, L.A. eds. (1996), The Uruguay Round and the Developing ity Growth and Poverty Reduction: National and Global Countries, Cambridge University Press, Cambridge. Impacts’ World Development 109: 429–39. Haque, M.N. (2018), ‘Dietary manipulation: a sustainable Jensen, H., Pérez Domínguez, I., Fellmann, T., Lirette, P., way to mitigate methane emissions from ruminants’ Jour- Hristov, J. and Philippidis, G. (2019), ‘Economic Impacts nal of Animal Science and Technology 60(15): 1-10. of a Low Carbon Economy on Global Agriculture: The Bumpy Road to Paris’ Sustainability 11(8), 2349. Harvey, L. (1993), ‘A guide to global warming potentials’ Energy Policy 21(1): 24-34. Laborde, D. and Mamun, A. (2019), ‘Global Database on Nominal Rate of Protection in Agriculture: 1955-2017’, Henderson, B. and Lankoski J. (2019), ‘Evaluating the envi- Powerpoint Presentation, International Food Policy ronmental impact of agricultural policies”, OECD Food, Research Institute, Washington DC, 10 December. Agriculture and Fisheries Papers, No. 130, OECD Pub- lishing, Paris. http://dx.doi.org/10.1787/add0f27c-en Laborde, D., Robichaud, V. and Tokgoz, S. (2013), ‘MIRA- GRODEP 1.0: Technical Documentation’ AGRODEP Himics, M., Fellmann, T., Barreiro-Hurléa, J., Witzke, Technical Note, IFPRI, December. H.P., Pérez Domínguez, I., Jansson, T., and Weiss, F. (2018), ‘Does the Current Trade Liberalization Agenda Lampayan, R., Rejesus, R., Singleton, G., Bouman, B. Contribute to Greenhouse Gas Emission Mitigation in (2015), ‘Adoption and Economics of Alternate Wetting Agriculture?’ Food Policy 76:120–9. and Drying Water Management for Irrigated Lowland Rice’ Field Crops Research 170:95-108. Hoekman, B., Martin, W. and Braga, C. eds (2009), Trade Preference Erosion: Measurement and Policy Response, Palgrave- Mamun, A., Martin, W. and Tokgoz, S. (2019), ‘Reform- MacMillan and the World Bank, New York and Washing- ing Agricultural Subsidies for Improved Environmental ton DC. Outcomes’ Discussion Paper, International Food Policy Research Institute, Washington DC. Houghton, J., Meira Filho, L., Callander, B., Harris, N., Kattenberg, A. and Maskell, K. eds. (1996), Climate Change Martin, W. and Alston, J. (1997), ‘Producer surplus with- 1995: The Science of Climate Change, Cambridge University out apology?: evaluating investments in R&D’ Economic Press, Cambridge, for the Intergovernmental Panel on Record 73(221):146-58, June. Climate Change. Mayrand, K., Dionne, S., Paquin, M., Pageot-LeBel, I. Huang, J., Gulati, A. and Gregory, I. (2017), Fertilizer Sub- (2003), The Economic & Environmental Impacts of Agri- sidies-Which Way Forward? IFDC and FAI, Muscle Shoals, cultural Subsidies: An Assessment of the 2002 US Farm Bill & AL Doha Round, Unisféra Intl Centre, https://unisfera.org/en/publications_details.sn?pub=5 40 Modeling the Impacts of Agricultural Support Policies on Emissions from Agriculture Mernit, J. (2018), ‘How Eating Seaweed Can Help Cows OECD (2019b), ‘A Global Economic Evaluation of GHG to Belch Less Methane’ Yale Environment 360, 2 July. Mitigation Policies for Agriculture’ Background Paper for https://e360.yale.edu/features/how-eating-seaweed- Joint Working Party on Agriculture and the Environment, can-help-cows-to-belch-less-methane COM/TAD/CA/ENV/EPOC(2018)7/FINAL, OECD, Paris. Millar, N., Doll, J. and Robertson, G.P., (2014), ‘Man- agement of Nitrogen Fertilizer to Reduce Nitrous Oxide Searchinger, T., Malins, C., Dumas, P., Baldock, Glauber, (N2O) Emissions from Field Crops’, Climate Change and J., Jayne, T., Huang, J., and Marenya, P. (2020), Revising Agriculture Fact Sheet Series—MSU Extension Bulletin Public Agricultural Support to Mitigate Climate Change, E3152 , Michigan State University, November. World Bank, Washington DC. Mundell, R. (1960), ‘The Monetary Dynamics of Inter- Searchinger, T., Waite, R., Hanson, C. and Ranganathan, national Adjustment under Fixed and Flexible Exchange J. (2019), Creating a Sustainable Food Future: A Menu of Solu- Rates’ Quarterly Journal of Economics 74(2): 227-57. tions to Feed Nearly 10 Billion People by 2050, World Resources Report, World Resources Institute, Washington DC. Neue, H. (1993), ‘Methane emission from rice fields: Wet- land rice fields may make a major contribution to global Shimeles, A., Verdier-Chouchane, A. and Boly, A. eds. warming’ BioScience 43 (7): 466-73. (2018), Building a Resilient and Sustainable Agriculture in Sub- Saharan Africa, Palgrave Macmillan, Cham, Switzerland OECD (2001), Measuring Productivity OECD Manual: Meas- urement of Aggregate and Industry-Level Productivity Growth, Tubiello, F., (2019), ‘Greenhouse Gas Emissions Due to OECD, Paris. Agriculture’, in Ferranti, P., Berry, E.M., and Anderson, J.R. (eds.), Encyclopedia of Food Security and Sustainability, vol. OECD (2016), OECD’s Producer Support Estimate and Related 1, Elsevier, Amsterdam, 196–205. Indicators of Agricultural Support, OECD, Paris. UN (1992), United Nations Framework Convention on OECD (2018a), Agricultural Policy Monitoring and Evaluation Climate Change, United Nations, New York. 2018, OECD Publishing, Paris. http://dx.doi.org/10.1787/agr_pol-2018-en. Vermeulen, S. (2016), ‘Agriculture is Integrated into the Paris Agreement’CGIAR Program on Climate Change, OECD (2018b), European Union: Estimates of Support Agriculture and Food Security. to Agriculture, OECD, version of 15 Feb 2018 “The https://ccafs.cgiar.org/research/annual-report/2015/ Cookbook” agriculture-is-integrated-into-paris-unfccc-agreement http://www.oecd.org/unitedstates/producerandconsum- ersupportestimatesdatabase.htm Zhang, Xiaobo and Fan, Shenggen (2004), ‘How produc- tive is infrastructure? A new approach and evidence from OECD (2019a), Agricultural Policy Monitoring and Eval- rural India’ American Journal of Agricultural Economics 86(2): uation 2019, OECD Publishing, Paris, 492–501. https://doi.org/10.1787/39bfe6f3-en. 41 1818 H Street, NW Washington, D.C. 20433 USA Telephone: 202-473-1000 Internet: www.worldbank.org/agriculture Twitter: @WBG_agriculture