Examining and Supporting Agricultural Interventions to Reduce Deforestation … · 2018. 10. 10. · from deforestation and forest degradation in developing countries.” The second

Examining and Supporting Agricultural Interventions to Reduce Deforestation in Brazil

Avery Simon Cohn

A dissertation submitted in partial satisfaction of the

Requirements for the degree of

Doctor of Philosophy

Environmental Science, Policy, and Management

in the

Graduate Division

of the

University of California, Berkeley

Committee in charge:

Professor Kate O’Neill, Chair Professor James Bartolome Professor Michael O’Hare Professor David Zilberman

Fall 2012

Examining and Supporting Agricultural Interventions to Reduce Deforestation in Brazil

Abstract

Examining and Supporting Agricultural Interventions to Reduce Deforestation in Brazil

Avery Simon Cohn

Doctor of Philosophy in Environmental Science, Policy, and Management

University of California, Berkeley

Professor Kate O’Neill, Chair

The title of this dissertation is “Examining and Supporting Agricultural Interventions to Reduce Deforestation in Brazil.” It is a collection of three research chapters. Each is a case study that stands alone and directly complements the other cases. Chapter One examines the international politics of targeting agriculture to Reduce Emissions from Deforestation and Forest Degradation (REDD) and how the ideas, approaches, and insights of an epistemic community of scholars has helped to frame the debate. Chapter Two explores whether in the absence of decisive multilateral actions, Brazil can effectively protect its forests without indirectly harming those of the rest of the globe. Chapter Three examines scientific, regulatory, and enforcement elements to reduce the risks of implementing a policy like the one I examine in Chapter Two. Taken together, these cases provide a cross section of normative, theoretical, empirical, regulatory, and political dimensions of emerging efforts to intervene in agriculture to prevent deforestation. Insights from each chapter and from the broader dissertation are intended to inform governance of the land use change process and to contribute to environmental social science scholarship that examines and informs these efforts. Chapter One: “How the Changing ‘Drivers of Deforestation’ have Shaped and May Shift REDD Principles” In 1992, the United Nations Conference on Environment and Development (UNCED) spawned a panoply of environmental regimes that have abjectly failed in

the eyes of some and changed the world in profound and difficult to discern ways in the eyes of others (O'Neill, 2009). Of the regimes that emerged from Rio ‘92, the biggest and best known is the United Nations Framework Convention on Climate Change (UNFCCC). This regime now has so many facets that groups have called for a cap on concurrent negotiating sessions out of concern for nations with small delegations (UNfairplay, 2011). In recent UNFCCC negotiations, REDD has been the biggest story, securing preliminary support, and spawning spin-off working groups in a period where broader progress in the UN climate negotiations has cooled considerably. Compared to Brazil’s own efforts and quite a number of national and sub-national initiatives, REDD has been slow to formalize rules and put them to the test. In this way, it would be wrong to study REDD “on the ground.” Indeed, when I did the interviews for this chapter, a number of my subjects implored me to look elsewhere for insights into REDD policy in action. However, the UN-REDD regime has long been and remains a crucial forum for the development, adoption and revision of guiding principles for the wider REDD world.1 Many of the government officials, industry groups, NGOs, and influential scholars shaping REDD initiatives come together at REDD meetings. Over the years, these meetings have been crucial places for ideas to coalesce, to evolve, and to spread.2 The texts and working papers that have resulted from this process form an archaeological record of the rapid epistemological evolution of preventing deforestation over the past five years. It is this history of ideas that is the core of this Chapter. A recent artifact of REDD is language in the 2010 Cancún decision exhorting parties to conceptualize deforestation as a problem of the whole landscape and the whole globe.3 This is in contrast to a more unitary previous focus on the forestlands themselves at a more logical scope (for rule of law), the nation state. 1 In a blog post following Rio+20, environmental law scholar Dan Farber argued that informal outcomes have trumped formal outcomes at many UN meetings on environmental governance - http://legalplanet.wordpress.com/2012/06/24/rio20-and-network-governance/. 2 A great example is the idea of compensated reduction itself, the idea for the sites and modes of governance of deforestation that is now called REDD. It was launched at a side even at the 2003 Barcelona COP to the UNFCC. For more, see Schlamadinger, B., Johns, T., Ciccarese, L., Braun, M., Sato, A., Senyaz, A., et al. (2007). Options for including land use in a climate agreement post-2012: improving the Kyoto Protocol approach. Environmental Science & Policy, 10(4), 295-305. 3 Four interrelated components of the Cancún decision have the potential to transform the REDD regime by shifting regime away from payments to landholders theory of change. Three of the four components come in Section C. The fourth component comes in Appendix II, a further explanation of portions of Section C. The first component is paragraph 68. 68 is a hortatory paragraph, it contains no requirements, only exhortations. It reads:

Encourages all Parties to find effective ways to reduce the human pressure on forests that results in greenhouse gas emissions, including actions to address drivers of deforestation;

Paragraph 68 contains two notable components. First is the notion that “all parties” have the capacity to “reduce the human pressure on forests.” Although the paragraph does not specify the forests to which the statement refers, the placement of this statement in Section C suggests that the focus is forest in developing countries. Section C is devoted to issues “relating to reducing emissions from deforestation and forest degradation in developing countries.” The second germane component of Paragraph 68 is that it exhorts all parties to take “actions” to “address drivers of deforestation.” The next component of the decision of note is Paragraph 72, another part of Section C. Paragraph 72 reads:

Also requests developing country Parties, when developing and implementing their national strategies or action plans, to address, inter alia, the drivers of deforestation and forest degradation, land tenure issues, forest governance issues, gender considerations and the safeguards identified in paragraph 2 of appendix I to this decision, ensuring the full and effective participation of relevant stakeholders, inter alia indigenous peoples and local communities;

Paragraph 72 begins with the word “also” because it continues on from paragraphs 70 & 71, a long list of requests for developing countries parties in order to prepare for REDD policy. The notable element of paragraph 72 is again the phrase “drivers of deforestation.” In 72, the implication is that addressing “drivers of deforestation” is a requisite ingredient for effective “national [REDD] strategies or action plans.” The final element of note in the Cancún decision is Appendix II. Appendix II is the entire a set of tasks generated by the Cancún decision on REDD for the Subsidiary Body for Scientific and Technological Advice (SBSTA) to take up going forward.3 SBSTA task (a) is the portion of Appendix II of note. It request that SBSTA:

Identify land use, land-use change and forestry activities in developing countries, in particular those that are linked to the drivers of deforestation and forest degradation, identify the associated methodological issues to estimate emissions and removals resulting from these activities, and assess the potential contribution of these activities to the mitigation of climate change, and report on the findings and outcomes of this work to the Conference of the Parties (COP) at its eighteenth session on the outcomes of the work referred to in this paragraph

in the scientific field that coined the term—Land Change Science (LCS). This history constitutes the first part of the chapter. In the second part of the chapter, I take up the question of how LCS has helped to shape REDD by developing the amorphous concept of drivers. The final portion of the chapter concludes with a synopsis of the results of interviews with REDD experts I conducted on the implication of the drivers theme for UN-REDD and wider REDD going forward. To do the research, I rely on a synthesis of UN-REDD party and observer interviews, a review of the drivers of deforestation scientific and grey literatures, and a review of the literature on forest and climate governance. Specifically, I examine the following questions and find the following:

• How and why and did LCS help to shape the guiding principles of the REDD regime?

o I show how the findings that LCS scholars make and the data that

they generate are just one way that LCS has helped to enable and shape REDD. The methods, norms, principles, and theories of LCS have also influenced REDD substantially. Through these pathways, LCS is not merely measuring what REDD aims to manage, but helping to construct REDD’s guiding philosophy of what the regime can accomplish and how to do it. These elements of LCS, not commonly associated with how science shapes policy, have helped to steer REDD towards reframing the agents who cause deforestation and reframing the appropriate places to intervene to stop the problem. Together, these two elements have been crucial for the regime’s rise.

• How did the word “drivers” enable the influence of LCS on REDD

principles?

o A corollary of the aforementioned finding is that framing matters; in particular, the use of evocative metaphor can make a substantial difference. In the popular lexicon, the word “drivers” means many things to many people. These meaning are distinct from the many things to many people that drivers signifies in LCS itself. These two sets of meanings are now intersecting to produce a third understanding of the term via the negotiations underway in the REDD regime. All language is metaphor. Scientists select metaphors to

explain causal processes. The metaphors we pick powerfully influence how our findings move through the world.

• Why did the 2010 Cancún decision of the COP to the UNFCCC urge a shift

in REDD’s founding guiding principles to expand the sites and strategies of REDD governance from the forests themselves to the “drivers” of deforestation emanating from across landscapes and across borders?

o The shift came as compromise between nations concerned that UN-REDD would not strongly enough target deforestation and other nations concerned about maintaining sovereignty and access to REDD finance. Though there was debate about the strength of the language, there was little debate about the broader content.

• How may the 2010 Cancún decision shift the guiding principles of REDD

concerning sites and strategies of governance?

o The final finding of Chapter One is that the COP to the UNFCCC is now hosting a political process to define drivers of deforestation and, perhaps, collapse away some of the conceptual breadth and flexibility that the term presently offers. As these deliberations proceed, they will serve to test the durability and inertia of the regime under a potential restructuring of its guiding principles the root of the deforestation problem and effective sites of intervention. The regime has reached a point where it may need to address, head on, the very principles that aided its birth and that have guided its rise.

The work has also sparked the following two questions for future research:

• How can the case of REDD’s shifting principles inform the practice of earth systems science?

• How do the insights from this case study compare with other findings on the

role of scientific norms in shaping the cognitive dimensions of global environmental governance?

Chapter Two: “Global Trade Can Provide and Scatter GHG Mitigation from Cattle Ranching Intensification Policies in Brazil”4 As Chapter One details, the REDD process in the UN has only inched along towards defining and addressing causes of deforestation; in particular, it has not addressed agricultural causes. The Brazil PPCDAM, by contrast, is much further along in this respect. However, even the PPCDAM has only just begun to address the role of agriculture in causing deforestation and in implementing interventions to address these causes. Thus far, one intervention involves a new line of subsidized credit through the the Program for Low Carbon Agriculture (ABC), which is intended to encourage low carbon practices by extending special loans. However, the program has faced barriers in distributing the loans effectively (Stabile, Nepstad, & Azevedo, 2012). In another intervention, the Environment Ministry (MMA) has lent support to third party efforts to ensure deforestation-free supply chains, but these efforts too are only just underway.5 Agricultural interventions, are thus, by and large, yet to come. For this project, we compare two potential agricultural interventions, a tax and subsidy, which each have the potential to achieve one of the most widely discussed intervention goals—the conversion of Brazil’s low productivity, pastured-based cattle systems into higher productivity pasture-based systems. Hereafter we refer to these interventions as cattle ranching intensification programs (CRIPs). Specifically, we examine how tax CRIPs and subsidy CRIPs affect the propensity of modeled cattle producers of Brazil to adopt intensive alternative technologies. In turn, we examine how these changes shift the simulated local landcover, global landcover, and GHG emissions relative to business-as-usual. Unlike previous research, we simulate the impacts of national policies on global land use with a model that takes into account global trade. The use of a global model is motivated by the slow pace of global agreements on land use in the UNFCCC. Because it is unclear when, if ever, the UNFCCC will be able to issue a unified global policy, there is a need to investigate the potential for

4 This chapter is joint with Aline Mosnier, Hugo Valin, Petr Havlik, Michael Obersteiner, Erwin Schmid, Mario Herrero, and Michael O’Hare. 5 Each of the last three Environment Ministers of Brazil Marina Silva, Carlos Minc and now Izabella Teixeira have encouraged soy industry groups such as Brazilian Vegetable Oil Industries Association (ABIOVE) to sign on to the Soy Moratorium. For more see, The Soy Moratorium Will Be Renewed to Combat Deforestation in the Amazon (Portuguese). (2011, 5/25). Globo Rural Online. Retrieved from http://revistagloborural.globo.com/Revista/Common/0,,EMI236052-18095,00-MORATORIA+DA+SOJA+SERA+REPACTUADA+PARA+ENFRENTAR+O+DESMATAMENTO+NA+AMAZONIA.html

national climate mitigation strategies that can deliver GHG benefits even in the present world of disjointed land use, forest, and climate policy. It is particularly crucial to understand global trade’s effect on the GHG-effectiveness of Brazilian CRIPs because in recent years, the market for Brazilian cattle products has become increasingly global (Millen, Pacheco, Meyer, Rodrigues, & Arrigoni, 2011). Previous research examining GHG mitigation potential of cattle ranching intensification in Brazil has not examined how international trade could influence the GHG potential of Brazilian climate policies (Gouvello, 2010; Lapola et al., 2010; Martha, Alves, & Contini, 2012). Perhaps one reason for this research gap was the lack of a model with sufficient detail of Brazil and with suitable linkages to global land use, agriculture, and food consumption dynamics. Most of the research activity associated with this project was developing such a model. I devised a detailed representation of Brazilian agricultural systems and cattle ranching intensification technologies in the Global Biomass Optimization Model (GLOBIOM), a partial equilibrium economic model of the competition between land use activities with a coupled greenhouse gas accounting model. I built and incorporated a spatial model of agricultural input and output costs and I developed a representation of grassland underutilization. Then, using the updated GLOBIOM, I examined the implementation of both intensification subsidies and taxes for the most extensive Brazilian cattle ranchers. Once the model preparation was complete, we used it to examine the following questions, and got the following results:

• Could policies promoting increased productivity in the cattle systems of Brazil reduce domestic and global deforestation?

o Our analysis finds that cattle ranching intensification policies in

Brazil could deliver global GHG benefits. Some of these benefits would be scattered beyond Brazil’s borders. While this bodes well from a global GHG mitigation perspective, it is of note that Brazil might not gain any tangible benefits from these mitigation efforts under the current system of GHG accounting.

• Are previous studies correct to suggest that policies to increase the

productivity of Brazilian cattle systems would “spare land” for crops and forests?

o Previous studies on the climate effects of intensification policies find GHG mitigation estimates similar to our GHG mitigation estimates. However, our findings suggest that others’ estimates are right for the wrong reasons. Previous studies simulate or project substantial conversion of pastureland to cropland (Gouvello, 2010; Lapola, et al., 2010; Martha, et al., 2012). These findings contribute to rosy projections of agricultural GDP and climate mitigation through the expansion of sugarcane bioenergy systems. Because of these projections, policy papers often invoke agricultural expansion as a key ancillary motivation for cattle ranching intensification. Our analysis suggests that both taxes and subsidies cap cattle ranching expansion, but do little to induce the conversion of existing cattle lands to agriculture. This is likely due to the economic and bio-physical marginality of these lands. On such lands, cattle systems regularly outcompete agricultural land uses.

• By what mechanisms and processes would these policies alter the Brazilian landscape and the global land system?

o Even in the absence of global land use and GHG emissions

regulations, Brazilian land use interventions have high global GHG mitigation potential. A subsidy is likely to cause mitigation outside of Brazil by displacing cattle production by other nations. Lower production costs increase the competitiveness of Brazilian beef on the international market. This new Brazilian beef displaces beef produced in other nations. In aggregate, the beef displaced would have contributed more emissions from deforestation than the new Brazilian beef. We find benefits of unilateral climate policy where the literature has primarily described costs.

• If these policy interventions were to work as modeled, would they be cost

effective?

social costs of implementing CRIPs (including adoption costs, government costs, and costs other private parties) dwarf private costs. These costs are still reasonably low relative to other policies, but their scale relative to private costs merits attention from the standpoint of political feasibility and durability.

Our future research will probe the sensitivity of our outcomes to assumptions about energy, market access, adoption behavior and patterns of trade. We are also developing a project to validate our ranching sector modeling with observational data. Chapter Three: “The viability of cattle ranching intensification in Brazil as a strategy to spare land and mitigate greenhouse gas emissions”6 The third chapter of my dissertation begins where the second one left off. It examines what policymakers would need to know and to do to reduce world GHGs through cattle ranching intensification programs in Brazil. In a nutshell, this project looks at the ways in which real world patterns of land use, challenges of policy design, and issues of implementation are likely to deviate from their land use model analogs. I develop a schematic of the steps for successful CRIPs, and I examine whether the scientific literature provides evidence that these steps could occur. I find that several factors will challenge the effectiveness of Brazilian land sparing schemes: (i) the global trade of Brazilian agricultural products; (ii) non-agricultural motivation for cattle ranchers to hold land; (iii) science gaps concerning the propensity of ranchers to intensify; and (iv) the need for more frequent and accurate earth observation of agricultural systems. We examine the following questions and find the following results:

• What needs to go right for CRIPs in Brazil to be effective at reducing greenhouse gas emissions?

6 This chapter is joint with Maria Bowman, David Zilberman, and Kate O’Neill. It is published as a working paper by Climate Change, Agriculture, and Food Security. The citation is Cohn, A., Bowman, M., Zilberman, D., & O'Neill, K. (2011). The viability of cattle ranching intensification in Brazil as a strategy to spare land and mitigate greenhouse gas emissions. Copenhagen, Denmark: CCAFS.

o Interventions that accelerate the adoption of higher yield production practices are a necessary but insufficient condition to ensure desirable GHG mitigation. Boosted output must reduce producer prices such that the profitable area of extensive cattle ranching shrinks. This then must lead to reduced deforestation and must free land for other productive uses. In addition, direct emissions increases must not offset the GHG benefits from land sparing. Finally, the benefits from the mitigation must exceed the loss to social welfare through intended and unintended consequences of the intervention.

• What are some risky and problematic assumptions that could make

successful model results, such as those we obtain in Chapter Two, difficult to replicate in practice?

o CRIPs clearly have high mitigation value. However, we find them to

be high risk as well. Many steps that must go right for these policies to deliver climate benefits could easily go wrong. The following factors are particularly important in determining the risk profile of CRIPs: (i) determinants of adoption propensity; (ii) endogeneity of intensification with land tenure; (iii) ability to ascertain the land use change associated with extensive pasture; and (iv) ability of competitive markets to “pass-through” lower cattle prices and reduce the area of profitable cattle ranching.

• What enforcement, data, and scientific activities could increase the

likelihood of effective CRIPs?

o Better agricultural data, land use data, governance, and understanding into the effectives of incentives on rancher behavior could all reduce the risk of CRIPs. Given the importance of these policies to succeed, it is crucial to pursue all of these activities.

for my family and for Lala

Table of Contents

List of Figures and Tables v

List of Abbreviations vii

Preface x

Introduction xvii Brazilian Deforestation Policy on the Rio+20 Stage xvii My Approach xxiii Interlinked Case Studies xxiv Chapter One: “How the Changing ‘Drivers of Deforestation’ have Shaped and May Shift REDD Principles” xxiv Chapter Two: “Global Trade Can Provide and Scatter GHG Mitigation from Cattle Ranching Intensification Policies in Brazil” xxix Chapter Three: “The viability of cattle ranching intensification in Brazil as a strategy to spare land and mitigate greenhouse gas emissions” xxxii Conclusions & Next Steps xxxiv

Acknowledgements xxxv

Chapter One: How the “Changing Drivers of Deforestation” have Shaped and May Shift The Principles Guiding REDD 1

Motivation, Framing, & Methods 1 A Brief Overview of REDD 2 Conceptualizing Deforestation: From Colonial Science to Cancún 7

Overview 7 Concern for Deforestation: 16th Century until 1970 7 Research Emerges on Causes of Deforestation 8 “Seeing” the Complex Causes of Deforestation 9 Parsing Causes (and its effects) 11 “Causes” to “Drivers” 13 Spotlighting Worthy Causes 15

Changing the Frame: From State-Led to Landholder-Led Deforestation 16 Deforestation Policies: 1985-2009 19

Overview 19 Public-Public 19 Private-Private 22 Private-Public: REDD 23 History 24

iii

Uptake 26 The Cancún Decision Broadens REDD Norms 28

Cancun in context 29 “Drivers” and Change? 29

Will Defining Drivers Transform REDD? 31 REDD Out Ahead 32 Agriculture 32 Issue Creep 33 Trade 33 Polycentricity 33 Finance 33 Demand-side and Supply Chain 34 Whole Landscape 34 Reference Levels 35

Concluding Thoughts 36

Chapter Two: Global Trade Can Provide and Scatter GHG Benefits from Cattle Ranching Intensification Policies in Brazil 39

Introduction 39 Policy Context 39 The Literature Gap 40 Our Contributions 41

Results 42 Baseline Simulation 43 Intensification Policy Scenarios Framework 46 Scenario Results 47 Social Cost of GHG Mitigation 55

Discussion 56 Feasibility and effectiveness of unilateral climate policies 56 Technology Adoption Patterns are Central to Intensification Efficacy 57 Better Governance Needed for Effective Land Sparing 59

Conclusions 60 Methods 61

GLOBIOM Overview 61 Demand Constraints 62 Input Data 62 Brazil Modifications 62 Intensification Costs 64 GHGs 64

Chapter Three: The viability of cattle ranching intensification in Brazil as a strategy to spare land and mitigate greenhouse gas emissions 66

Introduction 66 Definitions 69

Intensification 69 Land Sparing 70 Cattle Ranching Intensification Programs 71

The Essential Elements for CRIPs to Reduce GHGs 71 Intensive Ranching Technologies Must be Able to Scale Up 73 Inducing Adoption of Intensification Must be Straightforward 75 Theorizing Ranch-Level and Regional Determinants of Adoption 75 Increasing Intensive Ranching Must Reduce Cattle Product Prices 79 Reducing Cattle Product Prices Must Reduce Pasture Area 80 Reducing Extensive Cattle Ranching Must Deliver GHG Benefits 82 CRIPs Must Deliver Net Environmental & Social Benefits 85 CRIPs Must Deliver Benefits Cost Effectively 86

Conclusions 88

Concluding Thoughts 90 Overview 90 Green Accounting and Earth Systems Governance 91 Case as Critique: Green Accounting and Deforestation Policy 92 Deforestation Policy Analysis to Inform Deforestation Policy 94 Validate and Communicate 96 Reflexivity 96

References 98

List of Figures and Tables Figure i-1. The Reduction in Deforestation in the Legal Amazon, 2004-2011. .... xviii Figure i-2. Chart from Nepstad et al., (2009) showing The Action Plan to Prevent

Illegal Deforestation in the Brazilian Amazon (PPCDAM). ............................ xix Figure i-3. Map from the Environmental Research Institute of the Amazon showing

Amazon deforestation. ................................................................................. xxii Figure i-4. An illustration in a report to the UNFCC urging reform of the practice of

holding concurrent negotiating streams (UNfairplay, 2011). ........................ xxv Figure 1-1. A schematic of the references levels for determining progress towards

reducing emissions from deforestation (Parker et al., 2008). ............................ 3 Figure 1-2. Categorical variations in competing proposals to govern REDD (Parker

et al., 2008). ................................................................................................... 4 Figure 1-3. World Bank Forest Carbon Facility contrasts the scale of its REDD

readiness and pilot payments with its expectations of the scale of future REDD schemes (World Bank FCPF, 2008). ................................................................ 6

Figure 1-4. A diagram of the timeline for multilateral REDD activities (Pistorius, 2012). ............................................................................................................. 6

Figure 1-5. Turner’s theoretical framework for examining the causes of global change (Turner II, 1989). ............................................................................... 10

Table 1-1. Schematic depicting framings of the problem and solutions to deforestation (1989-2012). ............................................................................ 26

Figure 1-6. Notable events in the development of forest governance. ................... 28 Figure 1-7. A constellation of new themes “drivers” evoke for REDD experts. ...... 36 Figure 2-1. Political regions of Brazil. .................................................................. 42 Figure 2-2. Distribution of cattle in Brazil by political region over time. ............... 44 Figure 2-3. Policy-induced adoption of semi-intensive alternative pasture

management systems. ................................................................................... 49 Figure 2-4. Change in Pasture Area Caused by Intensification Policies, By Region

2010-2030. ................................................................................................... 50 Figure 2-5. Beef Production Displaced to Other Model Regions Under Tax

Scenarios. ..................................................................................................... 51 Figure 2-6. Beef Production Displaced in Other Model Regions Under Subsidy

Scenarios. ..................................................................................................... 52 Figure 2-7. GHG Impacts of Tax and Subsidy Policies. ........................................ 53 Figure 2-8. Evolution of Beef and Milk Prices in Brazil Under Intensification

Policies. ........................................................................................................ 54

Figure 2-9. Global social cost of GHG mitigation from pasture intensification scenarios in Brazil (in USD/tCO2e). ............................................................... 55

Table 3-1. Brazil’s Proposed NAMAs: Pledged Emissions Reductions for the year 2020 ............................................................................................................. 68

Figure 3-1. Essential elements for cattle ranching intensification programs to mitigate GHGs. ............................................................................................. 71

Figure 3-2. Jevons’ Paradox…………………………………………………………………..……81

vii

List of Abbreviations ABC Low Carbon Agriculture Credit Program of Brazil ABIOVE Brazilian Vegetable Oil Industries Association AFOLU Agriculture, Forestry, & Other Land Uses AU Animal Unit AWGLCA Ad Hoc Working Group on Long-Term Cooperative Action Under the

Convention BATCM The Berkeley Agricultural Transport Costs Model CBD United Nations Convention on Biodiversity CCAFS Climate Change, Agriculture, and Food Security CfRN Coalition of Rainforest Nations CGIAR Consultative Group in Agricultural Research CIFOR Center for International Forest Research CLF Campbell, Lobell, & Field CO2e Carbon Dioxide Equivalent COP Conference of the Parties COP11 Conference of the Parties to the UNFCCC Montreal, 2005 COP15 Conference of the Parties to the UNFCC, Copenhagen, 2010 COP16 Conference of the Parties to the UNFCC, Cancun, 2010 COP17 Conference of the Parties to the UNFCC, Durban, 2011 COP18 Conference of the Parties to the UNFCC, Doha, 2012 COP9 Conference of the Parties to the UNFCCC Milan 2003 CRIPs Cattle Ranching Intensification Programs CWE Carcass Weight Equivalent EISA Energy Independence and Security Act of 2007 EMBRAPA Brazilian Agricultural Research Corporation EPIC Environmental Policy Integrated Climate Model EU ETS The European Union Emission Trading Scheme FAO United Nations Food and Agriculture Program FSC Forest Stewardship Council GATT General Agreement on Tariffs and Trade GHGs Greenhouse Gases GLOBIOM The Global Biomass Optimization Model GTAP The Global Trade Analysis Project GWP Global Warming Potential

viii

IAM Integrated Assessment Model IBAMA Brazilian Institute of Environment and Renewable Resources IFRI International Forest Resources and Institutions Program IIASA The International Institute for Applied Systems Analysis ILRI The International Livestock Research Institute ILUC Indirect Land Use Change INPE The Brazilian National Space Agency ISEE The International Society of Ecological Economics ITTO International Tropical Timber Association IWG-IFR Informal Working Group on Internal Finance for REDD+ LCFS Low Carbon Fuel Standard LUC Land Use Change LUCS Land use change science LULUCF Land Use, Land Use Change, and Forestry MBA Masters of Business Administration MMA Environment Ministry of Brazil NAMAs Nationally Appropriate Mitigation Activities NCAR The National Center for Atmospheric Research NGO Non-Governmental Organization PNG Papua New Guinea PNMC The National Climate Action Plan of Brazil PPCDAM The National Plan to Prevent Deforestation PREPCOM Preparatory Meeting of the UNCED PRODES Program to Calculate Deforestation in the Brazilian Amazon RED Reduced Emissions from Deforestation REDD Reduced Emissions from Deforestation and Forest Degradation REDD+ Reduced Emissions from Deforestation, Forest Degradation, and

Enhancement of Forest Carbon Stocks REDDplus Reduced Emissions from Deforestation, Forest Degradation, and

Enhancement of Forest Carbon Stocks Rio+20 The 2012 United Nations Conference on Sustainable Development SBSTA Subsidiary Body on Science and Technical Advice SOM Supporting Online Material t ton TEEB The Economics of Ecosystems and Biodiversity UN United Nations UN-REDD United Nations REDD Programme UNCED The 2012 United Nations Conference on Sustainable Development UNCSD United Nations Conference on Sustainable Development

UNFCCC United Nations Framework Convention on Climate Change UNFF United Nations Forum on Forests USD United States' Dollars USRFS US Renewable Fuel Standard WB FCPF World Bank Forest Carbon Partnership Facility

Preface This telling of the arc of my Ph.D. begins in a conference on the economics of biofuels at the University of California, Berkeley Faculty Club, where I was watching a seminar by Tim Searchinger (2007). It was October 2007, and I was starting my first semester on campus. If you follow biofuels policy, you will probably recognize Searchinger’s name as the first author of an influential Science paper on biofuels and deforestation from February 2008 (T. Searchinger et al., 2008).7 The presentation we saw that day summarized the Science paper and occurred within a week of submission of the paper to Science for review (T. Searchinger, et al., 2008). Searchinger presented results from a global simulation model to demonstrate how, by increasing rates of deforestation, biofuels mandates in the United States could ultimately cause more climate harm than good. In the model, the mandate diverts crops away from food to fuel and causes three outcomes: food consumption decreases slightly in the poorest nations of the world, crop yields increase slightly wherever global markets reach, and conversion of new lands to agriculture increases sharply along the forest frontiers of South America and Southeast Asia. The resultant emissions, primarily from the deforestation effect, completely swamps the emissions that biofuels might help to avoid if they did nothing but displace fossil transportation fuels. Searchinger dubbed this phenomenon Indirect Land Use Change (ILUC). Some of the attendees at the workshop met Searchinger’s presentation with skepticism. One participant quipped that it was unfair to penalize biofuels producers for what he called the butterfly effect. By this he was referring to the idea that if a butterfly flaps it wings at the right time and in the right place that this could cause substantive and potentially far-flung consequences. To him, the geographic disconnect between biofuels production in the U.S. and the tropical land use change that Searchinger had simulated looked like a butterfly effect. How could the actions of farmers in the Midwest cause deforestation half a world away? Others felt Searchinger had cooked the books to get such a large effect.

7 This paper is currently the most heavily cited biofuels paper ever, with over 1,700 citations as of August 2012.

Searchinger’s finding had swift and substantial policy impacts. In the audience that day was the late Alex Farrell, at the time a newly tenured professor in the Energy and Resources Group at Berkeley. Alex was also the chief architect of the California Low Carbon Fuel Standard, the then still-being-planned mechanism for California biofuels policy (Farrell et al., 2007). Soon thereafter, Alex began lobbying the California Air Resources Board in earnest to account for indirect land use change emissions as part of the biofuels carbon footprint. Alex had been conscious of potential land use change effects of biofuels for some time already. In his 2006 Science article, to date the second-most heavily cited paper on biofuels, Alex and his co-authors cautioned in that their findings that biofuels could cause more good than harm were conditional on the magnitude of land use change resultant from biofuels policies. But Farrell et al. (2006) didn’t try to estimate the effect for that article. Searchinger’s publication helped to catalyze the California Air Resource Board to mandate that ILUC = be included in the agency’s lifecycle greenhouse gas accounting methodology for biofuels. Meanwhile in December 2007, though Searchinger’s paper still hadn’t been accepted for publication, a group of NGO representatives successfully lobbied for a stipulation in the Energy Independence and Security Act of 2007 (EISA) similar to what Farrell had won in California (US Congress, 2007). EISA was enacted with a requirement that the emissions associated with land use change had to be accounted for, and that it had to be included in the carbon footprint of every biofuel. In turn, the carbon footprint of each biofuel determined whether it qualified for inclusion in the mandate and, if so, to what extent. At the time, I was fascinated by the meteoric rise of ILUC from a thought experiment to a meaningful element of state and federal policy. I decided that as a case study about science in policy, ILUC deserved my attention. And at the time, this sort of environmental politics work was what I planned to do for my dissertation. I started collecting and classifying quotes on the ILUC controversy and gave a presentation in the March 2008 International Studies Association that used the science studies, Science and Technology Studies, and Global Environmental Politics literatures to situate the debate (Cohn, 2008). Meanwhile, I had been getting substantial exposure to the process of environmental assessment in Alex Farrell’s lab group, as well as in Arpad Horvath’s Lifecycle Analysis class. I was involved in both because I felt that my science politics work required technical literacy. But during that time, I realized that I did not just want the technical literacy, I wanted to do environmental assessment research as a part of my dissertation.

xii

There was a twist to my approach, however, that continues to this day—my work is focused on better understanding how institutions shape and production activities and utilize environmental assessment itself. I find this to be a terrain of crucial applied importance, of high potential for theoretical contributions, and an ideal way to apply my social science training. I also retain an interest in environmental politics with a particular focus on the interface with environmental assessment and land change processes. In particular, one theme has been important to my work since I attended Searchinger’s presentation in 2008—the environmental footprints of fuels and, in fact, anything else that people consume is not intrinsic, but rather contingent on markets, choices, and institutions. In Searchinger, the land use change modeled is a market-mediated effect. If the markets were to have different characteristics the effects would be different. Over the course of the early 2000s, this became a the topic of a methods debate in Lifecycle Assessment. One particularly vexing issue is that performance-based policies themselves affect markets and, therefore, they themselves affect the environmental performance of the production activities that they target. Production externalities are therefore not just a function of industrial activities, but also a result of the policies themselves. Internalizing externalities into the price does not predictably reduce externalities. Around this time I put together a qualifying exams list organized around the problem of the performance of performance-based policies. I read literature on Lifecycle Analysis, ILUC Analysis, Environmental Performance metrics, and Environmental Governance with a committee of Professors Arpad Horvath, Michael O’Hare, Alastair Iles, and Dara O’Rourke as exam chair. This preparation led to a methods paper on the use of LCA to examine compare the environmental performance of technology alternatives with a group of colleagues from the Renewable and Appropriate Energy Laboratory8. Our paper was inspired by a provocative Science paper using LCA to show that burning biomass for electric cars would “outperform” refining biomass into ethanol for internal combustion engines (J. E. Campbell, Lobell, & Field, 2009). Along with Derek Lemoine, I organized a team of students to put together a response where we showed how the performance benefits the authors modeled for bioelectricity cars are not actually intrinsic to bioelectricity production, but instead

8 Derek Lemoine, Rich Plevin, Adam Brandt, Andy Jones, Sintana Vergara, and Dan Kammen

xiii

sensitive to two assumptions about the regulatory and market context of the bioenergy use (Lemoine et al., 2010). First, the policies that bring bioelectricity to market influences the mix of electricity feedstocks that bioelectricity displaces. If biomass is used to satisfy a renewal portfolio standard (RPS), it would displace other renewables and increase the GHG intensity of electricity. If biomass is used outside of an RPS, it might displace conventional electricity feedstocks and in most cases9 reduce the GHG intensity of electricity. Second, producing bioelectricity would not necessarily spark the use of more electric cars as the original analysis assumed. This is because currently, electrified transport is limited by the high costs of batteries, not the availability of electricity. Like all other electricity, bioelectricity is a part of a common pool, the electric grid. For bioelectricity to displace gasoline, increased production of bioelectricity would need to spark the acquisition and use of electrified vehicles. However, if battery costs determine consumer uptake, producing more bioelectricity will have no effect. For me, researching biofuels and biofuels policy opened a door to asking similar questions about other kinds of land use activities. What are the mechanisms by which policies affect land use? To what assumptions are our estimates sensitive? What are the policy implications of market-mediated GHG effects from land use policies? I came to realize that the sorts of questions that Searchinger et al., Campbell et al., and Lemoine et al. raise are not limited to biofuels, but germane to land use policy broadly writ. I began thinking of multiple ways to apply these principles and approaches. For example, nascent efforts to promote lower carbon steel must grapple with questions of how to analyze and account ILUC effects. Specifically, one proposed approach to reducing steel GHGs is to replace coal-based charcoal with biomass-based charcoal. Yet the GHG benefits of the switch are sensitive to associated land use effects. And again, the policies and markets mediate the land use effects. To demonstrate the importance of this issue, I prepared a brief paper and presented a poster demonstrating how the indirect land use effects of steel substantially lessen the GHG savings from using biomass-derived charcoal to produce steel (Cohn, 2009).

9 Although if there were a carbon price, co-firing of biomass with coal (as is sometimes done in the Midwest) could slow the retirement of coal plants by allowing coal plants to compete with other, lower-carbon electricity sources.

xiv

At the start of my work on the steel ILUC paper, I had been intending to demonstrate the perils of ignoring or not accounting ILUC. By the end, I came to realize that my own engagement with ILUC is part of my broader interest in doing policy-relevant scientific inquiry to better understand the processes of land use change. For me, this boils down to: 1) understanding why land users, firms, and policymakers use land and spark land use change as they do; 2) accounting these activities in environmental assessments; and 3) using insights from global environmental politics to direct and support both of these previous two activities. Together, these three themes are focus of this dissertation. With the encouragement of Mike O’Hare, I decided to apply this approach, born out of the biofuels debates, to a topic of salience for biofuels policy and beyond—the dynamics of Brazilian agricultural systems. This focus marked a return to my first job after my master’s degree wherein I wrote a report investigating opportunities to reduce the environmental impacts of agricultural production in Northern Brazil.10 It was a field course during my master’s degree, several years before I saw Searchinger’s seminar, that got me started in Brazil. I had just agreed to spend a summer doing work on conservation and development with coffee farmers in El Salvador. At the time, I had a bit of background in field ecology, but I hadn’t spent much time on farms in general, let alone in the tropics. So I jumped at the chance for hands-on-learning by enrolling in a field course entitled “Conservation and Development in Amazonia.” Taught by professors Daniel Nepstad, David McGrath, and Lisa Curran, the course began with bimonthly lectures and ended with a three-week-long field practicum. With ferry boats, jeeps, canoes, and bushwacking, we explored a rich landscape in flux and met the people shaping it.

10 The report, prepared for the Nature Conservancy, was specifically to assess the potential for agricultural certification to serve as a conservation tool in the region. A key conclusion of the report was the lack of science and data concerning land dynamics in the region.

Our explorations were primarily in Santarém, Pará, a vivacious and remote city in a region at the confluence of the Amazon and Tapajós Rivers. The professors had remarkable expertise and social networks in the region and this made the experience irreplaceably textured. The region is home to a colorful tapestry of riverine communities of malarial gold-miners, liberation theologist environmental activists, isolated trans-migrants, illegal loggers, subsistence fishers, and a wild West style sense of both lawlessness and possibility. Yet amid of all of this character and color, the electric green soy fields and the sparkling new storage and processing facilities stood out to me. Just one month before our visit, a soy terminal facility owned by the agribusiness giant Cargill, and used for processing soybeans bound for China and Europe, suddenly appeared on the banks of the Amazon.This rise in soy production across the Southern Amazon region turned out to be the exclamation point at the end of the largest and most rapid expansions of agriculture that the planet has ever seen. It was the fastest expansion of the last twenty years of the twentieth century. With innovative new soy varieties optimized for the length of a tropical day, the weakest Brazilian Real of the decade, and voracious demand for soy products around the world (but particularly in Europe and China), the Brazilian soybean seemed to be an unstoppable juggernaut (Fearnside, 2001; Goldsmith & Hirsch, 2006). But even during our visit, the trend was beginning to evaporate—2003 marked the local peak of soy’s rise. The causes of this rapid rise and decline are tricky to disentangle, but recent research points primarily to the strengthening of the Brazilian currency and also to the National Plan to Prevent Deforestation in the Brazilian Amazon (PPCDAM), an audacious government-wide effort to protect forests (Assunção, e Gandour, & Rocha, 2012). Although it was the rapid, land use change that first sparked my research on agriculture in Northern Brazil, working in this new, more nuanced context is more rewarding. I am interested in this work because explaining land use change processes, including the influence of the incipient policies themselves, can support effective policymaking. Additionally, although other tropically forested nations have adopted innovative policies to reduce deforestation, Brazil is a first mover of its kind.11 Other crucial tropically forested countries such as Indonesia face similar hurdles. My hope is that 11 The first wave of effective forest conservation occurred in countries such as Costa Rica where advance were important, but where the scale of the problem is quite different.

xvi

the experience in Brazil can serve as a learning laboratory for efforts to govern land use elsewhere. Brazil is also a leader in national-level climate law in general and efforts to mitigate through land use pathways in particular. And despite taking a fairly hardline stance to maintaining a narrow scope for UN efforts to reduce deforestation, Brazil is charting an ambitious domestic policy course to address deforestation at the landscape level with the PPCDAM. The implementation of environmental policies poses technical challenges in any context, and land use change-based climate mitigation strategies are no exception. Feedbacks, spillovers, the inability to replicate interventions or control for spatio-temporally variant conditions, and multi-scale institutions undermine effort to assess the effectiveness of land use policies. In various ways, my dissertation grapples with the assessment of complex land use systems, the extent of the use of these assessments in Brazilian land use policy, the role that these policies might be playing and will play in the land change process, and the political dimensions of making these efforts work.

xvii

Introduction

Brazilian Deforestation Policy on the Rio+20 Stage Rio de Janeiro, Brazil (June 20, 2012) -- Rio+20, the United Nations (UN) Conference on Sustainable Development (UNCSD), is in full swing, but for the past few mornings, I have been commuting by metro from a youth hostel in Ipanema to downtown Rio de Janeiro for the meetings of the International Society of Ecological Economics (ISEE). Aiming to boost attendance and to influence the UN process, ISEE scheduled its meeting to fall between the preparatory opening (PREPCOM) and the high-level closing of the UN meeting. The heads of state haven’t arrived yet, but the city has declared school holidays to try to ease traffic as its streets teem with tens of thousands of visitors for the summit. To get to ISEE, I’ve been boarding subway Line One in Ipanema and zipping up the western edge of Guanabara Bay. Here the metro divides the old footprint of the city and the Aterro do Flamengo, a sinuous landfill that claimed the bay’s edge and expanded the Zona Sul shoreline when constructed in the 1960s. This week, the Aterro bustles with people—the Cupola dos Povos, the Peoples’ Summit runs along its length. From Flamengo to Catete, throngs of Peoples’ Forum attendees spill in and out of the metro. Yesterday, a group of indigenous Brazilians in headdresses and grass skirts waited purposefully on the platform, carrying spears. They were headed for a march organized by the global rural social movement, La Via Campesina. On these rides, between the crush of the commuters, I’ve caught sufficient glimpses to asynchronously see the entirety of a short Rio+20 welcome video playing on the platforms and produced by the Government of Brazil. On the loop, flocks of tropical birds, the iconic Pão de Açucar Mountain, and Iguaçu Falls slide across the screen along with bright orange flashing text: “Brazil has reduced illegal deforestation by 77 percent.” One of the Earth Summit’s fatal flaws seems to be its breadth, so the laser focus on this particular statistic startled me. I was also particularly attuned to noticing this statistic because it is part of a larger theme—public policies for the prevention of deforestation in Brazil—which is the organizing theme of this dissertation.

xviii

Curious, I did a Google search and traced the origin of the figure to a press conference by President of Brazil, Dilma Rousseff, herself just two weeks ago to acknowledge Earth Day in Brazil. There, Dilma told reporters that in 2012, the area of deforestation in the legal Amazon was less than one quarter of the area deforested in 2004. The decline from 2004 to 2012 was 77 percent.

Figure i-1. The Reduction in Deforestation in the Legal Amazon, 2004-2011. In 2004, an area of tropical forest in the Brazilian Amazon the size of Massachusetts, or over 27,000 square kilometers, was lost to deforestation. Since then, deforestation rates in the biome have fallen steadily. Nevertheless, cumulative deforestation continues to increase. Since 1989, an area the size of Virginia has been lost. Declining agricultural profitability over the period, and a coordinated effort by the Federal Government of Brazil to slow deforestation have contributed to the decline (Assunção et al., 2012)

Last week, Dilma devoted most of her weekly radio program, Café com a Presidenta, to describe Brazil’s success at fighting deforestation and to suggest that the experience could serve as a model for the Rio+20 agenda. In the conversation, Dilma frames the deforestation statistic as the chief environmental component of the evidence that Brazil can lead the Earth Summit green economy agenda by example. The framing is that deforestation has dropped while gross domestic

xix

product (GDP) has increased and poverty has declined.12 Dilma may also be using the statistic to symbolize Brazil’s burgeoning capacity for effective governance and international leadership.

Figure i-2. Chart from Nepstad et al., (2009) showing The Action Plan to Prevent Illegal Deforestation in the Brazilian Amazon (PPCDAM).

The PPCDAM features a stepwise target to reduce illegal deforestation in the Legal Amazon of Brazil. Thus far, the plan has been on target. While the green economy was supposed to be a fixture of the Earth Summit, delegates have been in a debilitating stalemate, unable to proffer even a vague definition. For Dilma, it is more straightforward. Her green economy appears to

12 Critics, including one government official from the Amazon state of Pará whom I met at the summit, point out that 1) the reduction in deforestation is occurring in just the Amazon biome, and (2) that the GDP of Brazil has always been dominated by São Paulo State, far from the Amazon.

have three axes—economic growth, poverty alleviation, and environmental protection in the form of reduced Amazonian deforestation.13 Thus, Dilma frames the green economy as something that the Government of Brazil is deliberately constructing by introducing poverty reduction and environmental protection constraints on economic growth. In Dilma’s rhetoric, government activities not only shape the poverty alleviation axis and the environmental regulation axis, but they determine them. During Café com a Presidenta, Dilma explained the case of the decline in the deforestation rate this way:

President Dilma: To reduce illegal deforestation 77 percent from the level of 2004, Brazil launched the Action Plan to Prevent and Control Deforestation in the Brazilian Amazon. Brazil, which has the privilege to possess the largest area of tropical forests in the world, can be proud to be protecting these forests more and more.

Luciano Seixas: What was done to achieve that outcome, Madam President?

President Dilma: Look, Luciano, the good news is the result of a strong enforcement campaign by the government—punishing and impeding the illegal deforestation. It is the result of coordination between IBAMA14, the Armed Forces, the Federal Police, and state governments.15

The policy that Dilma mentions, the PPCDAM, is not a garden-variety environmental policy. Rather it is an audacious approach, coordinated across numerous ministries, to phase out deforestation in the Amazon biome of Brazil by marshaling an innovative set of interventions. Begun in 2004, the PPCDAM was formulated to have three main approaches—direct prevention of deforestation16,

13 Critics have challenged this framing as both reductionist and hypocritical. Some argue that while of great importance, Amazonian deforestation is an insufficient stand-alone for environmental progress. Others argue that the Government of Brazil still does far too much to abet deforestation. In the past year, groundbreaking on the construction of the Belo Monte dam on the Xingu River and the weakening of the county’s forest code have been controversial and not-unrelated flashpoints. Both cases are examples of direct government actions likely to increase deforestation rates in the country. 14 The Brazilian Institute of Environment and Natural Resources, a federal government ministry that enforces environmental laws 15 Seixas, L. (2012). June 11th. Coffee with the President 16 i.e. the activities to prevent illegal removal of trees from private and public lands.

xxi

regularization of land tenure17, and agricultural interventions to reduce deforestation.18 The PPCDAM is being introduced in phases. Thus far, the vast bulk of activities fall in the category of direct deforestation prevention (Federal Government of Brazil, 2004). Every two days, the Brazilian National Space Agency (INPE) receives satellite images of the entire Amazon biome. They have developed algorithms to monitor these images for indicators of deforestation. INPE sends the incident report to IBAMA. IBAMA has seven helicopters for preventing deforestation. They use them to ferry enforcement teams to the active deforestation sites, prioritized by clearing size. When there are more than seven sites, they prioritize enforcement based on clearing size. Upon landing, teams investigate and make arrests and confiscate equipment where warranted. Despite complaints that the PPCDAM is authoritarian19, the budget for many of its enforcement activities remains meager20 and Dilma’s interview overstates PPCDAM’s efficacy at reducing deforestation. For example, recent research demonstrates that fluctuations in agricultural prices also have a pronounced effect on deforestation rates. When agricultural production costs in Brazil are high relative to the world market, agricultural output and deforestation naturally slow (Assunção, et al., 2012; Richards, Myers, Swinton, & Walker, 2012). The PPCDAM may be having an effect on the deforestation rate, but it is certainly not responsible for all of Brazil’s reduced deforestation over the period from 2004 to 2012. The most authoritative analysis to date attributes roughly 50 percent of the deforestation reduction to PPCDAM deforestation enforcement (Assunção, et al., 2012).21

17 A great deal of land in Brazil lacks clear, indisputable tenure, or tenure of any kind. Efforts to regularize land tenure would work grant titles and resolve title disputes. 18 I define agricultural interventions to prevent deforestation as any government activities with goal of reducing deforestation that seek to alter agricultural production practices, the configuration of agricultural supply chains, or the consumption of agricultural goods. Agricultural interventions could include efforts that target producers directly credit, input taxes and subsidies, research & development, and output taxes and subsidies. They also could include indirect interventions like agricultural infrastructure construction and planning, and even trade and fiscal policies. 19 http://www.eenews.net/climatewire/print/2012/06/22/3 20 The Brazilian government has estimated that its total annual budget to fight deforestation is roughly $500 million. This works out to roughly $200-700 per hectare of deforestation. 21 However, even this analysis may be overestimating PPCDAM’s effectiveness. Rather than examining the relationship that enforcement events have on deforestation outcomes, the analysis instead presumes that enforcement is more likely to have acute impacts in counties where less forest remains. However, there is no intuitive basis to suggest a link between forest scarcity and the acuity

xxii

Given these doubts about enforcement efficacy and as the Brazilian Real again weakens against the dollar, the importance of the land tenure and agricultural intervention portions of the policy are likely to heighten. In the very recent past, Brazil was the lowest cost large-scale agricultural producer in the world. If the currency falls again, deforestation in the country could again soar. The hope then would be that the land tenure and agricultural intervention components of the PPCDAM could help to sop up additional demand for deforestation if agricultural production again becomes more cost competitive.

Figure i-3. Map from the Environmental Research Institute of the Amazon showing Amazon deforestation. The green area of the map in the lower right shows the limits the political unit called the Amazon Biome. This area covers roughly 59% of Brazil. The bold black outline in the map at lower right shows the extent of the political unit, the Legal Amazon. The map at left shows area of

of the most prevalent enforcement type—satellite monitoring. Remote sensors are algorithmically triggered to observe deforestation events; they do not respond to forest scarcity. A newer enforcement strategy, county-level embargoes, is nominally focused on municipalities with high areas of deforestation. These high areas may co-vary with low quantities of remaining forest cover. However, even in such cases it would be unclear whether lower deforestation in such counties was due to a lack of forest left to cut or effective enforcement.

xxiii

forest cover within the Amazon biome (green), areas of recent deforestation until 2010 (red), and areas classified as non-forest within the biome (yellow). Note that the official deforestation statistics for the biome are based on annual satellite measurements of the loss of the green areas. Meanwhile, forest regrowth has begun in some of the red areas. An analysis conducted in 2008 estimated that over 20 percent of this type of area is the in process of regaining forest cover.

This strong rationale for intervening in agriculture to reduce deforestation is met with several problems: little experience in how to implement such a plan; daunting data and scientific requirements; and normative questions about the appropriate degree of centralization, not just for land use governance, but also for land use decision-making and, with it, questions about the broader land use economy. These uncharted terrains are the motivation for this dissertation. My aim is to examine and support agricultural interventions to prevent deforestation with the goal of informing policies to end deforestation, to build environmental social science, and to contribute to earth systems governance more broadly writ.

My Approach Few successful policy efforts to end deforestation exist and theoretical insights are lacking, particularly regarding effective agricultural interventions. Public goods are at stake, but these are not the prototypical community forest or local fishing grounds. Through markets and policy networks these public goods are closely linked to people, institutions, and landscapes all over the world. This is a commons that sprawls well beyond any jurisdiction—linking landscapes from Iowa, to Mato Grosso, Brazil to Kalimantan, Indonesia. It also propagates along supply chains from the churrascerias22 of São Paulo, to the Greenpeace stickers on Adidas boxes on the high streets of Germany, to United Nations negotiations, to the laboratories of biofuels start-ups in Northern California, and to the steel mills of Southeastern China. Stylized global overviews are on their own insufficient to capture the complexities, the patterns, the market characteristics, the technologies, the politics, and the behaviors crucial to target agricultural causes of deforestation. Locally bounded case studies miss key global processes. My approach is therefore to focus on a set of research cases that address crucial scientific, political, and geographic elements of the challenge.

22 Brazilian barbeque restaurants.

xxiv

Interlinked Case Studies The title of this dissertation is “Examining and Supporting Agricultural Interventions to Reduce Deforestation in Brazil.” It is not a monograph-style dissertation, but rather a collection of three research chapters. Each is a case study that stands alone and directly complements the other cases. Chapter One examines the international politics of targeting agriculture to Reduce Emissions from Deforestation and Forest Degradation (REDD) and how the ideas, approaches, and insights of an epistemic community of scholars has helped to frame the debate. Chapter Two explores whether, in the absence of decisive multilateral actions, Brazil can effectively protect its forests without indirectly harming the rest of the globe. Chapter Three examines scientific, regulatory, and enforcement elements to reduce the risks of implementing a policy like the one I examine in Chapter Two. Taken together, these cases provide a cross section of normative, theoretical, empirical, regulatory, and political dimensions of emerging efforts to intervene in agriculture to prevent deforestation. Insights from each chapter and from the broader dissertation are intended to inform governance of the land use change process and the scholarly communities studying and informing these efforts.

Chapter One: “How the Changing ‘Drivers of Deforestation’ have Shaped and May Shift REDD Principles” Twenty years ago, here in Rio, the United Nations Conference on Environment and Development (UNCED) spawned a panoply of environmental regimes that have abjectly failed in the eyes of some, and changed the world in profound and difficult to discern ways in the eyes of others (O'Neill, 2009). Of the regimes that emerged from Rio ’92, the biggest and best known is the United Nations Framework Convention on Climate Change (UNFCCC). This regime now has so many facets that groups have called for a cap on concurrent negotiating sessions out of concern for nations with small delegations (UNfairplay, 2011). In recent UNFCCC negotiations, REDD has been the biggest story, securing preliminary support, and spawning spin-off working groups in a period where broader progress in the UN climate negotiations has cooled considerably.

xxv

Figure i-4. An illustration in a report to the UNFCC urging reform of the practice of holding concurrent negotiating streams (UNfairplay, 2011). This illustration intends to evoke the challenges faced by small nations to follow concurrent negotiating streams resultant from the ever-growing activities under the UNFCCC. REDD itself has more than five separate negotiating streams.

Compared to Brazil’s own efforts and quite a number of national and sub-national initiatives, REDD has been slow to formalize rules and put them to the test. In this way, it would be wrong to study REDD “on the ground.” Indeed, when I did the interviews for this chapter, a number of my subjects implored me to look elsewhere for insights into REDD policy in action. However, the UN-REDD regime has long been and remains a crucial forum for the development, adoption and revision of guiding principles for the wider REDD world.23 Many of the government officials, industry groups, NGOs, and influential scholars shaping REDD initiatives come together at REDD meetings. Over the years, these meetings have been crucial places for ideas to coalesce, to evolve, and to spread.24 The texts and working 23 In a blog post following Rio+20, environmental law scholar Dan Farber argued that informal outcomes have trumped formal outcomes at many UN meetings on environmental governance - http://legalplanet.wordpress.com/2012/06/24/rio20-and-network-governance/ . 24 A great example is the idea of compensated reduction itself, the idea for the sites and modes of governance of deforestation that is now called REDD. It was launched at a side even at the 2003

xxvi

papers that have resulted from this process form an archaeological record of the rapid epistemological evolution of preventing deforestation over the past five years. It is this history of ideas that is the core of this Chapter. A recent artifact of REDD is language in the 2010 Cancún decision exhorting parties to conceptualize deforestation as a problem of the “whole landscape” and the entire globe.25 This is in contrast to a previous, more unitary focus on the forestlands themselves on the nation state level. Barcelona COP to the UNFCC. For more, see Schlamadinger, B., Johns, T., Ciccarese, L., Braun, M., Sato, A., Senyaz, A., et al. (2007). Options for including land use in a climate agreement post-2012: improving the Kyoto Protocol approach. Environmental Science & Policy, 10(4), 295-305. 25 Four interrelated components of the Cancún decision have the potential to transform the REDD regime by shifting regime away from payments to landholders theory of change. Three of the four components come in Section C. The fourth component comes in Appendix II, a further explanation of portions of Section C. The first component is paragraph 68. Paragraph 68 is a hortatory paragraph; it contains no requirements, only exhortations. It reads:

Encourages all Parties to find effective ways to reduce the human pressure on forests that results in greenhouse gas emissions, including actions to address drivers of deforestation;

xxvii

In the text of the Cancún decision and in the shorthand of UN-REDD insiders, the proposed shift of REDD guiding principles is known as the introduction of “drivers of deforestation” or “drivers” for short. Interested in the origins of this shift to drivers, I began by seeking out the origins of the concept of drivers of deforestation in the scientific field that coined the term—Land Change Science (LCS). This history constitutes the first part of the chapter. In the second part of the chapter, I take up the question of how LCS has helped to shape REDD by developing the amorphous concept of drivers. The chapter concludes with a synopsis of the results of interviews I conducted with REDD experts on the implication of the drivers theme for UN-REDD and wider REDD going forward. To do the research, I rely on a synthesis of UN-REDD party and observer interviews, a review of the drivers of deforestation scientific and grey literatures, and a review of the literature on forest and climate governance. Specifically, I examine the following questions and find the following:

• How and why and did LCS help to shape the guiding principles of the REDD regime?

o I show how the findings that LCS scholars make and the data that

Technological Advice (SBSTA) to take up going forward.25 SBSTA task (a) is the portion of Appendix II of note. It request that SBSTA:

xxviii

problem. Together, these two elements have been crucial for the regime’s rise.

• How did the word “drivers” enable the influence of LCS on REDD

principles?

o A corollary of the aforementioned finding is that framing matters; in particular, the use of evocative metaphor can make a substantial difference. In the popular lexicon, the word “drivers” means many things to many people. These meanings are distinct from the many things to many people that drivers signifies in LCS itself. These two sets of meanings are now intersecting via the negotiations underway in the REDD regime to produce a third meaning. All language is metaphor. Scientists select metaphors to explain causal processes. The metaphors we pick have a powerful influence how our findings move through the world.

• Why did the 2010 Cancún decision of the COP to the UNFCCC urge a shift

• How may the 2010 Cancún decision shift the guiding principles of REDD

concerning sites and strategies of governance?

o The COP to the UNFCCC is now hosting a political process to define drivers of deforestation and, perhaps, collapse away some of the conceptual breadth and flexibility that the term presently offers. As these deliberations proceed, they will serve to test the durability and inertia of the regime under a potential restructuring of its guiding principles.. The regime has reached a point where it may need to address, head on, the very principles that aided its birth and that have guided its rise.

xxix

The work has also sparked the following two questions for future research:

• How can the case of REDD’s shifting principles inform the practice of earth systems science?

• How do the insights from this case study compare with other findings on the

role of scientific norms in shaping the cognitive dimensions of global environmental governance?

Chapter Two: “Global Trade Can Provide and Scatter GHG Mitigation from Cattle Ranching Intensification Policies in Brazil”26 As Chapter One details, the REDD process in the UN has only inched along towards defining and addressing causes of deforestation; in particular, it has not addressed agricultural causes. The Brazil PPCDAM, by contrast, is much further along in this respect. However, even the PPCDAM has only just begun to address the role of agriculture in causing deforestation and in implementing interventions to address these causes. Thus far, one intervention involves a new line of subsidized credit through the Program for Low Carbon Agriculture (ABC), which is intended to encourage low carbon practices by extending special loans. However, the program has faced barriers in distributing the loans effectively (Stabile, Nepstad, & Azevedo, 2012). In another intervention, the Environment Ministry (MMA) has lent support to third party efforts to ensure deforestation-free supply chains, but these efforts too are only just underway.27 Agricultural interventions are thus, by and large, yet to come. For this project, we compare two potential agricultural interventions, a tax and subsidy, which each have the potential to achieve one of the most widely discussed intervention goals—the conversion of Brazil’s low productivity pasture-based cattle systems into higher productivity pasture-based systems. Hereafter we refer to these interventions as cattle ranching intensification programs (CRIPs).

26 This chapter is joint with Aline Mosnier, Hugo Valin, Petr Havlik, Michael Obersteiner, Erwin Schmid, Mario Herrero, and Michael O’Hare. 27 Each of the last three Environment Ministers of Brazil Marina Silva, Carlos Minc and now Izabella Teixeira have encouraged soy industry groups such as Brazilian Vegetable Oil Industries Association (ABIOVE) to sign on to the Soy Moratorium. For more see, The Soy Moratorium Will Be Renewed to Combat Deforestation in the Amazon (Portuguese). (2011, 5/25). Globo Rural Online. Retrieved from http://revistagloborural.globo.com/Revista/Common/0,,EMI236052-18095,00-MORATORIA+DA+SOJA+SERA+REPACTUADA+PARA+ENFRENTAR+O+DESMATAMENTO+NA+AMAZONIA.html

xxx

Specifically, we examine how tax CRIPs and subsidy CRIPs affect the propensity of modeled cattle producers of Brazil to adopt intensive alternative technologies. In turn, we examine how these changes shift the simulated local landcover, global landcover, and GHG emissions relative to business-as-usual. Unlike previous research, we simulate the impacts of national policies on global land use with a model that takes into account global trade. The use of a global model is motivated by the slow pace of global agreements on land use in the UNFCCC. Because it is unclear when, if ever, the UNFCCC will be able to issue a unified global policy, there is a need to investigate the potential for national climate mitigation strategies that can deliver GHG benefits even in the present world of disjointed land use, forest, and climate policy. It is particularly crucial to understand global trade’s effect on the GHG-effectiveness of Brazilian CRIPs because in recent years, the market for Brazilian cattle products has become increasingly global (Millen, Pacheco, Meyer, Rodrigues, & Arrigoni, 2011). Previous research examining GHG mitigation potential of cattle ranching intensification in Brazil has not examined how international trade could influence the GHG potential of Brazilian climate policies (Gouvello, 2010; Lapola et al., 2010; Martha, Alves, & Contini, 2012). Perhaps one reason for this research gap was the lack of a model with sufficient detail of Brazil and with suitable linkages to global land use, agriculture, and food consumption dynamics. Most of the research activity associated with this project was developing such a model. I devised a detailed representation of Brazilian agricultural systems and cattle ranching intensification technologies in the Global Biomass Optimization Model (GLOBIOM), a partial equilibrium economic model of the competition between land use activities with a coupled greenhouse gas accounting model. I built and incorporated a spatial model of agricultural input and output costs and I developed a representation of grassland underutilization. Then, using the updated GLOBIOM, I examined the implementation of both intensification subsidies and taxes for the most extensive Brazilian cattle ranchers. Once the model preparation was complete, we used it to examine the following questions, and got the following results:

• Could policies promoting increased productivity in the cattle systems of Brazil reduce domestic and global deforestation?

xxxi

o Our analysis finds that cattle ranching intensification policies in Brazil could deliver global GHG benefits. Some of these benefits would be scattered beyond Brazil’s borders. While this bodes well from a global GHG mitigation perspective, it is of note that Brazil might not gain any tangible benefits from these mitigation efforts under the current system of GHG accounting.

• Are previous studies correct to suggest that policies to increase the

productivity of Brazilian cattle systems would “spare land” for crops and forests?

• By what mechanisms and processes would these policies alter the Brazilian

landscape and the global land system?

o Even in the absence of global land use and GHG emissions regulations, Brazilian land use interventions have high global GHG mitigation potential. A subsidy is likely to cause mitigation outside of Brazil by displacing cattle production by other nations. Lower production costs increase the competitiveness of Brazilian beef on the international market. This new Brazilian beef displaces beef produced in other nations. In aggregate, the beef displaced would have contributed more emissions from deforestation than the new Brazilian beef. We find benefits of unilateral climate policy where the literature has primarily described costs.

xxxii

• If these policy interventions were to work as modeled, would they be cost

effective?

o Previous analyses suggest that cattle ranching intensification in Brazil is one of the lowest cost, highest volume strategies to mitigate global GHGs (Eliasch, 2008; McKinsey & Co., 2009; Gouvello, 2010; Lapola, et al., 2010). We find that while there are some low-cost locales for intensification, the private cost curve slopes up steeply in the last quartile of ranches. Also of note is that for many ranches, social costs of implementing CRIPs (including adoption costs, government costs, and costs other private parties) dwarf private costs. These costs are still reasonably low relative to other policies, but their scale relative to private costs merits attention from the standpoint of political feasibility and durability.

Chapter Three: “The viability of cattle ranching intensification in Brazil as a strategy to spare land and mitigate greenhouse gas emissions”28 The third chapter of my dissertation begins where the second one left off. It examines what policymakers would need to know and do to reduce world GHGs through cattle ranching intensification programs in Brazil. In a nutshell, this project looks at the ways in which real world patterns of land use, challenges of policy design, and issues of implementation are likely to deviate from their land use model analogs. I develop a schematic of the steps for successful CRIPs, and I examine whether the scientific literature provides evidence that these steps could occur. I find that

28 This chapter is joint with Maria Bowman, David Zilberman, and Kate O’Neill. It is published as a working paper by Climate Change, Agriculture, and Food Security. The citation is Cohn, A., Bowman, M., Zilberman, D., & O'Neill, K. (2011). The viability of cattle ranching intensification in Brazil as a strategy to spare land and mitigate greenhouse gas emissions. Copenhagen, Denmark: CCAFS.

xxxiii

several factors will challenge the effectiveness of Brazilian land sparing schemes: (i) the global trade of Brazilian agricultural products; (ii) non-agricultural motivation for cattle ranchers to hold land; (iii) science gaps concerning the propensity of ranchers to intensify; and (iv) the need for more frequent and accurate earth observation of agricultural systems. We examine the following questions and find the following results:

• What needs to go right for CRIPs in Brazil to be effective at reducing greenhouse gas emissions?

o Interventions that accelerate the adoption of higher yield production

practices are a necessary but insufficient condition to ensure desirable GHG mitigation. Boosted output must reduce producer prices such that the profitable area of extensive cattle ranching shrinks. This then must lead to reduced deforestation and must free land for other productive uses. In addition, direct emissions increases must not offset the GHG benefits from land sparing. Finally, the benefits from the mitigation must exceed the loss to social welfare through intended and unintended consequences of the intervention.

• What are some risky and problematic assumptions that could make

successful model results, such as those we obtain in Chapter Two, difficult to replicate in practice?

o CRIPs clearly have high mitigation value. However, we find them to

• What enforcement, data, and scientific activities could increase the

likelihood of effective CRIPs?

o Better agricultural data, land use data, governance, and understanding into the effectives of incentives on rancher behavior

xxxiv

could all reduce the risk of CRIPs. Given the importance of these policies to succeed, it is crucial to pursue all of these activities.

Conclusions & Next Steps The basic issues that President Dilma raised in her radio interview and the areas of inquiry in this dissertation have not faded at all in importance. Fundamental avenues for future inquiry remain—there is still much to learn in order to spare the land. Next month, I am starting a postdoctoral fellowship in which I will continue to explore these issues using interdisciplinary methods, with a foundation in lessons from the past and an eye toward the future. In the concluding thoughts section at the end of this dissertation, I discuss how my dissertation and my future work aim to build on the green accounting model of earth systems governance in concrete ways. I use the case of land use governance to show that accountings of physical processes, if they are to accurately generate theories of policy effectiveness, badly need complementary environmental social science. This is because social factors, including policy design, policy adoption, cognition, culture, and prices, can all shift the relationships between production activities and environmental outcomes. The study of these phenomena will require the cooperation of policymakers, and the use of mixed methods; it cannot be done in silos. The effort will require quantitative research in communication with qualitative research, model-to-model comparisons, and numerous other efforts to validate and translate within and beyond the academe. Finally, scholars and policy makers alike will need to be reflexive in order to surmount the challenges of this ambitious agenda. With all of the above, efforts to end deforestation can begin a new paradigm of earth systems governance.

xxxv

Acknowledgements Berkeley, California (August 10, 2012) -- I didn’t know it at the time, but this dissertation began on a paper placemat over a long lunch with my parents in May 2004. I’d just been offered a consulting position to research agricultural interventions to prevent deforestation in Brazil and the task felt both electric and daunting. Three hours later, the three of us were still chattering excitedly about research approaches, theoretical lenses, and how to do applied research well. I am immeasurably fortunate that this sort thing was a regular feature of my life. I will always be grateful for the way both of my parents, Ted and Barbara, nurtured the scientist in me from a very early age. I am grateful for so much more too, and I love you both dearly. I also love and appreciate the friendship of my sister Adrienne, my brother Harris, and my brother-in-law James. We have always shared fun and adventures and you’ve been there for me whenever I need it. Much love to Bill too—it has been a pleasure to get to know you. Thanks for all my relatives young and old for your engaged support. Thank you for not sneaking into my graduation! My time at Berkeley has been everything I could have hoped a Ph.D. could be. Thanks to the Department of Environmental Science, Policy, & Management (ESPM) and especially my division of the Department, Society & Environment. This is a place where people push the boundaries of science and disciplines not just because they can, but because they care. Memorable ESPM activities include meetings with the Science, Technology & the Environment graduate group, serving on the Vision Committees of 2011 and 2012, attending Clare Kremen and Justin Brashares’ seminar on agriculture and biodiversity, The Bartolome Rangelands Reading Seminar, and joining up with InterdisciplinariTEA. Huge thanks go to the following ESPM faculty. Dara O’Rourke for his mentorship, and friendship; Alastair Iles for being a generous and encouraging PI and quals committee member; David Winickoff for support and encouragement; James Bartolome for sage and good-natured advice and feedback on the dissertation, and Nancy Peluso for having the best potlucks. I am of course especially grateful to my advisor, Kate O’Neill. Thank you for allowing me the freedom to follow my passions, thank you for being firm when you sensed I needed to reign things in, and thank you for giving me and other students in your group a glimpse into your

xxxvi

strategies for being a successful and respected scholar (translation chart of British academic English and all). Thanks also to friends and peers at ESPM- Manisha Anantharaman, Patrick Baur, Graham Bullock, Elizabeth Havice, Abby Martin, Kabir Peay, Mark Phillbrick, and Seth Shonkoff, and Alexis Yelton. Even before I started at Berkeley, I attended meetings of the Biofuels Interest Group, a lab group organized by the late Alex Farrell of the Energy and Resources Group. This was the just the beginning of an amazing experience as an honorary ERGie (ERG is the Energy and Resources Group at UC Berkeley). My studies were enriched by the projects with the Renewable and Appropriate Laboratory, lunches and a successful research grant with Dick Norgaard, and thought provoking early mornings at the ERG Climate Lab. Thanks to friends and colleagues Renata Andrade, Alex Farrell, Kevin Fingerman, Anand Gopal, Stacy Jackson, Andy Jones, Dan Kammen, Derek Lemoine, Morgan Levy, Sergio Pacca, Rich Plevin, Deepak Rajagopal, Laura Schewel, Niels Tomijima, Sintana Vergara, and Scott Zimmerman. I also feel lucky to have found friends, colleagues, and mentors on the rest of campus. Thanks to David Zilberman and Arpad Horvath for being committee members and mentors, Maria Bowman for being a colleague, and a friend, and Mike O’Hare for being a committee member, a mentor, and a field hand who paid not only his own way, but my way too. The Bay Area Tropical Forest Network has been a wonderful place to make friends, expand my intellectual horizons, and drink beer. Thanks to Rhett Butler and Holly Gibbs for founding it. It was a pleasure to get to know Rachael Garrett, Lisa Kelley, Hannah Salim, Matt Potts, and Kim Carlson. I grew up in the area and lived here before graduate school. Being around old friends and their network was a special treat. Thanks to Cooper, Ramsey, Mike, Kevin, Ben, and Jonah for being the best of friends from high school onward. Thanks to my roommates and friends Ramsey, Michelle, Nell, Jill, Drew, Colin, Cynthia, Sharon, Julia, Brenna, Joyce, and Emily. Thanks, to the Wesleyan crew, the Sol crew, and the O Bigode crew. Thanks also to John, Jon, Kabir, Zack, Matt, and Brody for adventures off the coast. Finally my most heartful thanks to Lala Wu. Your love and support means everything to me.

xxxvii

Chapter One: How the “Changing Drivers of Deforestation” have Shaped and May Shift The Principles Guiding REDD

Motivation, Framing, & Methods In December 2005, a team of politicians from Papua New Guinea and Costa Rica presented a framework to “Reduce Emissions from Deforestation” at the Conferences of the Parties (COP) to the United Nations Framework Convention on Climate Change (UNFCCC) Montreal in 2005. During the previous year, a group from Columbia University had done careful legwork for the proposal. The intellectual roots for the proposal go back further, to a team of land change scientists who presented the concept at the Milan COP to the UNFCCC in 2003 (M. Santilli et al., 2003; B. Schlamadinger et al., 2005). The Santilli et al. (2003) proposal has deeper roots still. It coheres closely with a longstanding tendency in Land Change Science (LCS) concerning how to discuss the causes of deforestation and strategies to mitigate and slow deforestation. The innovation in the Columbia proposal was to frame Reduced Emissions from Deforestation and Forest Degradation (REDD) as a regime that would do two crucial things—enroll nation-states with the purpose of reforming private sector actvities. By doing so, the proposal specified for REDD a root of the problem (the private sector) and a site for intervention (the public sphere) that were distinct from previous roots and sites, distinct from each other, and conceivably effective (For more see Table 1-1). Thus the REDD proposal, known as compensated reduction, nimbly addressed two main pitfalls that had plagued foregoing attempts at international forest governance—the moral hazard of paying developing nations to police themselves and the corollary debate about a fair price to pay. In this paper, I trace this, innovation, from its roots in early LCS and into the policy arena. I demonstrate how the concept of drivers of deforestation was instrumental in enabling the search for theories that would facilitate the adoption of the policy. However, as the policy path continues to unfold, the regime has perhaps reached a point where it will need to address, head on, the very principles that aided its birth. I explain how this happened by drawing on interviews that reveal how the term “drivers of deforestation” came to appear in the COP to the UNFCCC. I then close by discussing the opportunities and challenges associated with this development.

Some experts I interviewed for this piece see the “drivers” language as a harbinger of a shift towards what some are calling Plan B REDD. This is the multi-faceted assembly of actions to achieve REDD occurring outside the formal UNFCCC process. Some see the ‘drivers’ concept as ushering in a shift in REDD positions, including funding. Norway, for example has just released a tender to spend billions on addressing agricultural drivers of deforestation. Thus through actions, symbolism, and as a case of international environmental politics the appearance of “drivers of deforestation” language merits the focus I’m giving it. This focus, for me anyway, has helped to organize my own thoughts on agricultural drivers of deforestation, a topic of great policy importance, but that has not been addressed as a case of global environmental politics. The pathway leading to this rather rapid and remarkable change is a crucial component of this piece. This timeline serves two purposes. First it provide readers context for my findings. I also use a version of it to elicit insights from the experts I interviewed in the piece. The process was a sort of participatory process tracing. I developed an initial timeline and then sought affirmation and critique from my interview subjects. I used a mix of methods for this piece. I conducted participant observation at several meetings29, I interviewed 18 experts, and reviewed the scientific literature on causes, of deforestation. The methods also include a review of publicly available official decisions, agendas, and commentaries from the UNFCCC website. The bulk of the interview subjects are REDD specialists from parties to the UNFCCC (12). An additional four are NGO representatives working on the REDD. The final two are scholars closely involved in researching REDD policy.

A Brief Overview of REDD REDD is an environmental governance regime30 with the aim of reducing emissions from land use in developing countries. The first public presentation advocating for

29 Meetings attended were “The Role of Commodity Roundtables & Avoided Forest Conversion in Subnational REDD+,Agriculture, Food Security & Greenhouse Gas (GHG) Accounting”, San Diego, California, September 2011; The COP to the UNFCCC Durban, South Africa, December 2011; and the United Nations Conference on Sustainable Development, Rio de Janeiro, Brazil, June 2012. 30 Here I use regime in broad sense after the definition proposed by O’Neill (2009). She defines regimes as collections of rules, organizations, norms, and principles. She bases her definition on works by Krasner (1983), Downie (2005), and Hasenclever (1997). For more please see O'Neill, K. (2009). The environment and international relations. New York, New York: Cambridge University Press. Krasner, S. D. (1983). International regimes. Ithaca, New York: Cornell Univ Press. Hasenclever, A., Mayer, P., & Rittberger, V. (1997). Theories of international regimes. New York,

REDD was presented by Brazilian lawyer Márcio Santillli in 2003 at the ninth COP to the UNFCCC in Milan, Italy (M. Santilli, et al., 2003). In 2005, at the Eleventh COP to the UNFCCC in Montreal, Canada, the governments of Costa Rica and Papua New Guinea proposed the inclusion of REDD in the UNFCCC. The proposal won support and the parties began work to concretize REDD. REDD enjoyed a good deal of progress within the climate regime from 2005-2010. At the invitation of the UNFCCC, parties and observers made scores of proposals for formalizing REDD. The Bali Action Plan adopted REDD as part of GHG mitigation efforts in 2007, the 2009 Copenhagen Accord included a REDD mechanism, and the 2010 Cancún Agreements made REDD an official element of the UNFCCC process.

Figure 1-1. A schematic of the references levels for determining progress towards reducing emissions from deforestation (Parker et al., 2008). The basis for REDD performance payments is relative to previous emissions and not absolute. To determine relative emissions requires counterfactual assumptions of the amount of deforestation emissions that would have occurred in the absence of policies to government efforts to prevent deforestation. This counterfactual level is also referred to as a reference level. Emission reductions are the difference between observed deforestation emissions and those of the reference level.

The competing REDD proposals inaugurated a set of as yet on-going debates about REDD architecture. Key issues included the appropriate sorts of activities eligible for compensation under REDD (scope), how to determine how much credit these

New York: Cambridge University Press. Downie, D. L., Krueger, J., & Selin, H. (2004). Global Policy for Hazardous Chemicals. Global Environmental Policy: Institutions, Law and Policy, 125-145.

activities should receive (reference level), who should pay how much for what and when (finance), and who should receive how much payment when (distribution) (Parker, Mitchell, Trivedi, & Mardas, 2008). Figure 1-2 depicts one rendition of these issues and how they relate. Figure 1-1 depicts the basic premise for one of the more arcane of these activities, the calculation of reference levels. .

Figure 1-2. Categorical variations in competing proposals to govern REDD (Parker et al., 2008).

In 2005 parties and observers were invited to submit proposals for the mechanisms and modes of the REDD regime. The basic question on which the submitted proposals differed: Who should pay how much to whom for what?

In addition to REDD inside the corridors of the UNFCCC, to understand what REDD is accomplishing, it is important to look beyond the decisions won within the formal UNFCCC process. From 2005-2010 REDD has undergone a diversification of the “sites”31 of REDD governance as well. The World Bank Forest Carbon Partnership Facility (FCPF) and the UN-REDD Programme32 both began to distribute finance not exclusively linked to deforestation. For a chart of distributed and projected REDD finance please see Figure 1-3. A large number of voluntary REDD entities also emerged. Meanwhile, the scope has expanded from reduced emissions from deforestation (RED), to reduced emissions from deforestation and forest degradation (REDD), to reduced emissions from deforestation and forest degradation plus the enhancement of forest carbon stocks (REDD+).33

31 According to O’Neill (2009), “’Sites’ of governance are not literal locations, but rather arenas of governance within the broader structure of global governance in which actors interact and make decisions. For more see, O'Neill, K. (2009). The environment and international relations. New York, New York: Cambridge University Press. 32 The UN-REDD Programme is an organization housed in the UN to “assist developing countries prepare and implement national REDD+ strategies” 33 For simplicity’s sake, I refer to all government led efforts from 2003-2012 as REDD. This includes RED, REDD, and REDD+ as REDD. I do not consider REDD to signify non-governmental REDD.

Figure 1-3. World Bank Forest Carbon Facility contrasts the scale of its REDD readiness and pilot payments with its expectations of the scale of future REDD schemes (World Bank FCPF, 2008).

Despite the debate and the broader outgrowth of sites and modes34 of REDD, it is remarkable that, to present, REDD has by and large35 retained as guiding principles the structure outlined in its founding principles—an entity interested in reducing deforestation pays governments to distribute finance to reform the activities of private landholders in such a way as to reduce GHGs (hereafter I refer to this structure as the IGL principle or just IGL36).37 For the purposes of this paper, I presume that REDD entails the IGL principle.

Figure 1-4. A diagram of the timeline for multilateral REDD activities (Pistorius, 2012). The 2005 decision to consider REDD led to early workshops, then development of formal rule making processes, and the establishment of institutions proffering support for readiness and capacity building.

34 According to O’Neill (2009), “‘Modes’ of governance are ways of crafting and implementing environmental regulations and initiatives – whether it be through the negotiation of treaties or the development of private-sector led voluntary certification systems.” For more see, O'Neill, K. (2009). The environment and international relations. New York, New York: Cambridge University Press. 35 The exception to this rule are voluntary or non-governmental REDD activities. I exclude this in order to focus on the attractiveness of REDD sites and modes of governance during its initial uptake. 36 “I” stands for interested entity, “G” stands for government, and “L” stands for landholder. 37 The most obvious exception to this rule is non-governmental or voluntary REDD. I exclude this type of REDD from my discussion of REDD 2005-2012. In my discussion of future directions for REDD I examine NGO REDD and voluntary REDD in greater depth.

Conceptualizing Deforestation: From Colonial Science to Cancún

Overview This section traces how and why the science of deforestation helped launch and popularize the REDD regime. Here I present a brief history of deforestation science to trace how scholarship dedicated to examining the causes of deforestation came to join early normative works decrying forest loss. As the focus on causes intensified, it was the LCS scholars who began working to explain and theorize the multiple causes of the land use change patterns that their research was demonstrating. This required developing approaches to describing causal inference and the lack of it.38 Here I focus on several elements of this approach that have proved important for the influence of deforestation scholarship on REDD. First, the group retreated from the use of strong causal language. They began to employ an evolving classification framework to parse causes into various classes such as proximate and ultimate causes. Little consensus has ever existed on the meaning of these classes. Instead, the classes allow for the coexistence of competing theories. Later LCS began using the term “drivers” in place of the word “causes.” Drivers evokes causality to a lay audience, but also dilutes the strength of the scholarly claim to a level that the system complexity warrants. This duality, coupled with the theoretical churn in the field’s early days allowed scholars to spotlight causes that they found worthy.

Concern for Deforestation: 16th Century until 1970 The verb deforest first appeared in the English language in the 16th century (Oxford English Dictionary, 2012a). As early as then, it had two meanings, one institutional and the other physical. Under the institutional meaning deforestation meant, “to reduce from the legal position of forest to that of ordinary land; to make no longer a forest.” In Europe in the period, forests were often royal land. The act of deforesting involved administratively freeing land for alternative use. The second definition is the one widely used today—the physical removal of trees from forest lands. This

38 I conceptualize this group as an epistemic community and I refer to it as LCS. In what many consider the seminal work on epistemic communities, Haas (as cited in O’Neill, 2009), describes them as, “transnational networks of knowledge-based communities that are both politically empowered through their claims to exercise authoritative knowledge and motivated by shared causal and political beliefs.” (p. 349) For more, see Haas, P. M. (1990). Saving the Mediterranean: the politics of international environmental cooperation. New York, New York: Columbia University Press. and O'Neill, K. (2009). The environment and international relations. New York, New York: Cambridge University Press.

definition has a close functional relationship to the first definition in that it also connotes the freeing of forest lands for alternative uses. Concern over deforestation is the subject of extensive writing and the motivation for environmental policy development throughout Europe and European colonies from the 16th century onward (J. Fairhead & Leach, 1998; Rajan, 2006; Boyd, 2010). Schama (1996) suggests that this concern may be motivated by material interests and the proclivity of European elites to view forests as sacred. The focus on deforestation led some early writers to seek out explanations for the phenomenon. French administrators in Guinea mistakenly believed that local populations were agents of deforestation and forest fragmentation when in fact the opposite was true (J. Fairhead & Leach, 1996). Villagers planted trees around their villages and over time these became the sole forest fragments in a region otherwise devoid of tree cover. The mistaken administrators of Guinea are part of a broader pattern of inept diagnosis of the causes of deforestation well into the 20th century. Concern about forest loss motivated speculation about the causes of deforestation, but not scholarly approaches to examine social processes of deforestation with sophistication and depth (J. Fairhead & Leach, 1998; Rajan, 2006).

Research Emerges on Causes of Deforestation The 1970’s and early 1980’s final saw an emergence of more rigorous scholarship into the role of human choices and institutions in shaping deforestation patterns (S. B. Hecht, 1982; Rudel, 1983; Allen & Barnes, 1985). During the period, scholars began to develop and test anthropological, micro-economic and political economic theories for explaining deforestation. The spatio-temporal focus ranged from annual household choices to the evaluation of national forest policy across decades. Yet the poor quality of forest statistics and forest observation limited these analyses in a number of respects. Remote sensing of forest cover was in its infancy at the time and forest surveys were both unreliable and subject to methodological discrepancies that undermined comparison (Grainger, 2010). Over the period from 1960 to 1985, estimates of global forest cover were not only widely discrepant, but even increasingly divergent (Allen & Barnes, 1985).

“Seeing” the Complex Causes of Deforestation The mid-1980’s marked a crucial shift in forest sensing capacity (Boyd, 2010). During this time, ecologists and earth system scientists, fueled by increasing availability of improved remote sensing data, began producing regional and global analyses of the fate of tropical forests of heightened credibility.39 As LCS scientists improved at mapping deforestation, they began to seek explanations for the patterns in their data (Grainger, 2010). The goal was to develop biophysical and socio-economically rooted theories of large-scale environmental changes. A manifesto for this development is, “The Human Causes of Global Change”in Global Change and Our Common Future: Papers From a Forum (B. L. Turner II, 1989). The article is significant for its introduction of a framework for relating human choices and social structures with changes in the earth system. Depicted in Figure 1-5, the framework delineates “human driving forces” that “underl[ie]” “proximate sources of change” in land use and industry. This chapter appears to be the birth of the concept of “drivers of deforestation” in the study of LCS.40 Turner II’s framework has been incredibly influential in the broader literature on deforestation. As I go on to argue, it presents guiding principles and a theoretical backbone for particular and popular approach to studying the deforestation process that I and others identify as LCS.

39 Jasanoff Jasanoff, S. (2001). Image and imagination: the formation of global environmental consciousness. Changing the atmosphere: Expert knowledge and environmental governance, 309-337. heralded the earth systems science breakthroughs of the period as part of “concerted action” that may have been emanating from a new “collective vision”, a new and widely held understanding of the role of people on earth. For more, see ibid. 40 Writing in 2001, Geist and Lambin adapt Turner’s schematic and make a number or rearrangements including the replacement of the term driving forces with the term drivers. See Geist, H., & Lambin, E. (2001). What drives tropical deforestation? A meta-analysis of proximate and underlying causes of deforestation based on subnational case study evidence. LUCC Report Series, 4, 116.

Figure 1-5. Turner’s theoretical framework for examining the causes of global change (Turner II, 1989). Notable innovations include the retreat from strong causal language through the use of the terms driving forces and mitigating forces and the distinction between these “driving forces” and “proximate sources of change.”

The distinctions among causes that Turner II made didn’t last long, but his choice to refer to causes as driving forces and to parse causes became a hallmark of the field. Turner II (1989) is quite clear about the distinction between proximate sources and driving forces. Proximate sources constitute observable changes in land cover such as a shift from forested land to cropland as observed in a series of aerial photographs. Driving forces are all the human factors that might cause such changes. Within driving forces, Turner II delineates agency and structure, but he does not discuss this distinction much. However, soon after the 1989 piece in a series of later first authored and co-authored pieces, Turner II returns to the 1989

driving forces framework and reprises it slightly. In ensuing works, Turner II changes his definition of proximate sources from observable social processes to “facets of society that underlie [the observable social processes]” (B. L. Turner II et al., 1990:18). This reprised framework was adopted by Ojima et al. (1994) and through Ojima et al., it came to form the basis for the analytic framework that Geist & Lambin (2001; 2002) employ in two works that stand as two of the top five most widely cited papers on deforestation. In this way, Turner II’s updated framework becomes a sort of epistemic coat rack for LCS. On it hang a great many principles, norms and theories that have developed in the field.

Parsing Causes (and its effects) A key element of the reprised Turner II framework is its use of the terms proximate and ultimate cause—a sort of epistemological olive branch. The distinction between proximate and ultimate causes is longstanding in the English language. In fact, the oldest English usage of proximate, dating to the 17th century, is the phrase, “proximate cause”. Here proximate means, “coming immediately before or after in a chain of causation, agency, reasoning, or other relation; immediate, short-term. Freq. in proximate cause. Opposed to remote or ultimate” (Oxford English Dictionary, 2012b). Proximate is thus implicitly causal. Also of note is that the lay distinction between proximate and ultimate causes is straightforward, but still not categorical.41 In theory, the degree of spatial and or temporal immediacy between a cause and effect should determine whether a cause should be classed as proximate or ultimate.42 The distinction between proximate and ultimate causes exists in many sciences, but the meaning of this classification differs across disciplines. For example, Mayr (1961), writing on the epistemology of the biological sciences draws the distinction in this way:

Proximate causes govern the responses of the individual (and his organs) to immediate factors of the environment while ultimate causes are responsible for the evolution of the particular DNA code of information with which

41 I.e. proximate might refer to one end of gradient. 42 In the sciences, however, it has long been clear that establishment of cause itself is a fraught proposition. Hume argues, for example, that there is no systematic way to distinguish chains of causation and association. For Hume, proximity is merely a necessary, but insufficient quality for causality. Though later philosophers of science have rebutted Hume’s argument, Hume’s observation concerning the complexity of demonstrating causality stands. See Hume, D. (2000). An enquiry concerning human understanding: a critical edition (Vol. 3): Oxford University Press, USA

every individual of every species is endowed. The logician will, presumably, be little concerned with these distinctions. Yet the biologist knows that many heated arguments about the “cause” of a certain biological phenomenon could have been avoided if the two opponents had realized that one of them was concerned with proximate and the other with ultimate causes.

In sum, Mayr argues that while there is no logical and no theoretical meaning to distinguishing between proximate and ultimate causes, in practical terms the distinction helps to smooth science in action.The proximate/ultimate classification can diffuse scholarly debate by allowing the coexistence of pluralistic theories of change. Thus, an alternate reading of the Mayr quote might be that the sum of causes to which he alludes constitutes his version of the sum of state variables of relevance across the biological sciences. The proximate variables fill a zone occupied by one class, e.g. microbiologists, while the ultimate variables fill a zone occupied by another class, e.g. evolutionary biologists. The distinction allows the different types of specialists to harmoniously coexist without sparring over “ultimate” causes. Parsing proximate and ultimate cause may serve different purposes in mature vs. nascent fields. No scientific field is stable, but a field as established (and focused) as biology is a stark contrast with the breadth and youth of LCS. Thus, while the distinction between microbiology (proximate) and evolutionary biology approaches (ultimate) has remained fairly clear-cut and stable in biology, the distinctions between proximate and ultimate causes in land change science have been far less stable. Over the brief history of LCS, the master list of state variables (as embodied in proposed analytic frameworks) in the field has fluctuated and shifted. These fluctuations and shifts can even be observed over very brief time periods. Recall for example that Turner II completely overhauled his own definition of “proximate” some time between 1989 and 1992. Moreover—and this is of crucial importance—the function of the proximate and ultimate classification in LUC science has not been as much about dividing state variables among specialists (as in biology between the microbiologists and the evolutionary biologists), but at times it has been to weigh in on debates about the importance of the state variables themselves. That is, parsing proximate versus ultimate enables normative framing of the systems scholars study.

“Causes” to “Drivers” In addition, and perhaps synergistically with parsing causality, Turner II derivative frameworks retreat from strong causal language. The retreat from strong causal language is perhaps best encapsulated by the rise of the use of the term “drivers” in place of the term “causes.” Here I provide an overview of the rise of term in the sciences. Then I demonstrate how the use of drivers has allowed LCS scholarship to navigate the field’s formative years without substantial controversy. The section concludes with several reasons why the dual meanings of the term are problematic. The concept of drivers of deforestation has played a crucial role in maintaining the coherence and cohesiveness of land change science by lessening the need for scholars to agree on a stable and coherent list of state variables that explain the land change process. The concept is also useful for lessening scholarly focus on distinctions between variables classified as proximate and those classified as ultimate and, crucially, attendant debates about importance. Even from its earliest introduction to the field in Turner II (1989), part of the role of drivers (then driving forces) was to imply an element of squishiness and indeterminacy that the word “cause” does not imply. Thus, as different phenomena have shifted in and out of LCS scholarship and causes have shifted from being classed as proximate or ultimate, using an umbrella term such as drivers helps to increase the sense of coherence in a field that is still undergoing defining activities. Drivers is a recent term that has emerged across scientific literature, appearing only sporadically before the early 1990s. Its use has increased tremendously during the last decade, but its distribution across fields remains uneven. Focal fields for drivers tend to be outcomes of strategic concern (terrorism, development, global change, deforestation). A commonality of all of these fields is that the motivations for study are often applied. They tend to entail forging new methodological and theoretical ground. As a consequence, drivers is commonly used to describe outcomes of complex statistical and econometric modeling where multi-causality and complexity limit clear causal stories. The term drivers seems indicative of non-mechanistic research. Additionally, in mathematical modeling of complex systems, it is common for practitioners to describe exogenous shocks as model drivers. Thus, if a model of crime and society is parameterized such that crime rates are associated with the need to hire more police officers, and the modeler is analyzing the possible outcomes of boosted crime rates, the modeler might say that the crime rates are “driving” the number of police on the street. They might even say that crime rates drive the model itself.

When used in this way, the use of the term drivers can grow problematic. The relationship between crime rates and police officers is more complex than has been depicted in the aforementioned modeling exercise (Murray, 2006). The choice to hire more police officers might reduce crime rates. Yet by specifying police officers as an endogenous variable and the crime rate as an exogenous variable, the modeler is holding the crime rate constant and thereby disallowing the number of police to affect the crime rate in ways that research shows to occur. Here, through the activity of modeling, the researchers have taken a two directionally causal (or endogenous) relationships and reduced it to uni-directionality. The use of the term drivers plays some role in this simplification. It would be wrong to say that crime rates cause police officer hiring patterns. Saying crime rates “drive” police officer hiring patterns is for some reason more permissible in the scientific literature. In a system involving three or more interacting variables, the story grows more complex. For example, though deforestation does not have any direct causal effect on government subsidies, it might also be wrong to model government subsidies to “drive” deforestation because on their own government subsidies are not responsible for any deforestation without private agents directly induced by the subsidies to deforest. Government subsidies might be a necessary condition for some portion of deforestation, however and land change science should describe this in causal terms. Formally, government subsidies should be said to cause precisely the amount of deforestation that land change scientists estimate would not have happened in the absence of the subsidies and all else equal. Unfortunately, this sort of marginal analysis is not the standard in land change science, almost certainly because it constitutes too high an empirical bar. Instead, correlates of deforestation are frequently described as drivers of deforestation. In some cases, such correlates are used to “drive” model outcomes and other suppositions. In this way, the term drivers may occlude analysis of the social causes of the LUC process. The use of the term drivers to indicate correlation with the supposition of causality is also problematic because the lay definition of drivers seems to connote strong causality. Thus an irony of the use of the term drivers to soften causality is that for a great many readers, the term may do the reverse. Evidence is the two stage transformation of studies documenting the spatio-temporal proximity of cattle pasture to a large percentage of all deforestation in the Brazilian Amazon, to studies describing this proximity as “driving deforestation”, to NGO reports saying that the cattle pasture causes the deforestation to which it is spatio-temporally proximate.

Spotlighting Worthy Causes A source of further interpretative complexity is that scholars also curate the landscape of deforestation factors by unevenly ascribing teleological significance to factors associated with deforestation. Haas (1990) notes that this practice is, by definition endemic to applied fields. The notion of teleological causes was Aristotle’s way of distinguishing causes with moral significance or purpose from the broader landscape of causes.43 In the deforestation space, this significance is often introduced through discussion of blame and/or responsibility for deforestation, for many LUC scholars an undesirable outcome. In turn, blame/responsibility may be constructed based scholars’ expectations of the potential for regulatory clout. Throughout this paper I reference spotlighting of causes of deforestation with the phrase, root of the problem. I reference spotlighting opportunities to prevent deforestation with the phrase, sites of intervention. Consider for example Zaks et al. (2009), an article on the agents to hold “responsible” for emissions from deforestation in the Brazilian Amazon. The article presumes that the entirety of the blame lies along the supply chain (and not in the actions of other actors outside the supply chain). Though the authors opt to split “responsibility” equally between producers and consumers, they argue that the framework informing the divide should be influenced by knowledge of regulatory clout:

If responsibility is given to the producer, carbon leakage can occur, and if it is assigned to the consumer not participating in a global GHG reduction agreement, the responsibility for the emissions are not taken (Andrew and Forgie 2008). Hence, several authors have put forth allocation schemes in which carbon emissions are shared between producers and consumers (Lenzen et al 2007, Rodrigues and Domingos 2008, Bastianoni et al 2004 ). These shared allocation schemes provide economic incentive to the consumer nation to favor products with the smallest environmental impacts, and thereby push producers to reduce the carbon emissions embodied in their products (Zaks et al., 2009:2)

43 Aristotle argued that four distinct types of causes exist: material (what the observed objective is physically made of), formal (the location of the object in hierarchies and structures of objects), efficient (the way the object moves through space and time), and final or teleological cause (the object’s purpose and moral characteristics). Mayr (1961) decries the use of teleological language to describe organisms’ traits as though caused by a higher purpose.

In sum, in Zaks et al. (2009), responsibility is a classification employed not strictly to denote causality, but also to spotlight an opportunity for GHG mitigation. To the extent that responsibility connotes justice, and to the extent that GHG reduction contributes utility, this is an approach where “justice as utilitarianism” freely mixes with causal inference (Okereke & Dooley, 2010). Zaks et al. (2009) could easily have invoked other justice frames or invoked other norms entirely. The point is that in the piece, the theoretical framework is a device not strictly to support causal inference, but also to spotlight causes of normative concern to the authors. Authors also use the modifiers proximate or ultimate to signal teleological significance. However, there is no convention to this use. Different frameworks for classifying variables connote entirely different significance and implications to proximate vs. ultimate variables.

Changing the Frame: From State-Led to Landholder-Led Deforestation During the 1990s, members of LCS and other thought leaders in Washington environmental NGOs and multilaterals reframed tropical deforestation from a problem of government choices to a problem of decisions by large landholders, corporate farmers, and market failure. During the 1970’s and 1980’s conservation NGOs and later developed world governments came to “see” tropical deforestation as a global problem (Boyd, 2010). This “seeing” of tropical deforestation as global problem had two main components (Boyd, 2010). First, there was the idea that tropical deforestation constituted a problem of global importance. Second was the notion that global earth systems governance could be feasible and that such governance could include governance of tropical deforestation (Jasanoff, 2001). This conceptual advance paved the way for the spotlighting of problems and solutions in the study of deforestation. As rhetoric to address tropical deforestation ramped up in the late 1980’s, the leading LCS scholarship focused attention on national governments of tropically forested countries as prime agents of deforestation (Repetto & Gillis, 1988; Mahar, 1989; Binswanger, 1991). These claims were based in some logic, but they also had some inconsistent elements. First, during the period, many tropical countries, including two of the largest, Brazil and Indonesia, still had highly active patterns of state-sponsored colonization (Rudel, 2007). These activities took the form of trans-migration projects where the state-built infrastructure and promoted the urban and rural poor to move to remote areas. Second, during the period, as now, much of tropical forests were properties of states (White & Martin, 2002; Forests for People,

2011). Thus implicating states in the deforestation of their own lands had a logical basis. However, the 1980’s also has a robust literature chronicling challenges tropical nations faced in frontier settlement and also a wider literature about the lawlessness of frontier tropical forest regions (For an example see Hochman & Zilberman, 1986). Thus the idea of the causal role of the states suffers somewhat for lack of plausibility. How exactly could a state control a frontier process when by definition they were thought to have such a weak frontier presence? By 1999, the pendulum had swung in the Beltway, and the literature on state-sponsored deforestation had been largely replaced by a new literature implicating landholders and commodity markets in deforestation.44 The earliest pieces on the subject are associated with the World Bank and appeared starting in 1993 (Lutz et al., 1993; A. S. P. Pfaff, 1996; ASP Pfaff, 1999).45 Interestingly, though the state-sponsored deforestation pieces do not use the term drivers of deforestation at all, the new crop of scholarship uses the term frequently and prominently and often at the expense of the word “causes.”46 Perhaps it was in this way that this literature could coexist with the still recent state-sponsored deforestation literature without generating much discussion or debate about the profound shift that had occurred. The use of the term “drivers” helped to cloak the profoundly different implications of the incipient literature on landholder and commodity market deforestation. The shift to focus on the private sector as the root of the problem benefited from the availability of new spatial data and had both empirical and theoretical facets. In order to examine the role of landholders in deforestation, scholars such as Pfaff (1996) and Andersen & Reis (1997) leveraged previously unavailable agricultural and deforestation data. With these data, they were able to ask previously unexamined questions about the associations between land use activities and the

44 Outside of LCS, a great deal of literature on state-sponsored deforestation continued in other circles, however. Political ecology, political economy, and economics saw the publication of classics on governments and deforestation during the period. See for example King, V. T. (1993). Politik pembangunan: The political economy of rainforest exploitation and development in Sarawak, East Malaysia. Global Ecology and Biogeography Letters, 235-244. And Dauvergne, P. (1993). The politics of deforestation in Indonesia. Pacific Affairs, 497-518. Deacon, R. T. (1995). Assessing the relationship between government policy and deforestation. Journal of Environmental Economics and Management, 28(1), 1-18. 45 It is notable that the focus on landholder-led deforestation is contemporaneous with a broader trend towards decentralization and rollback of the state. For more see Evans, P. (1997). The eclipse of the state. World Politics, 50(1), 62-87. 46 In Binswanger (1991), the word drive and its derivatives appears in just one place: “Anything that drives the land price above the capitalized value of the agricultural income stream thus makes it impossible for the poor to buy land without reducing consumption.”

deforestation processes. This line of inquiry required the development of innovative econometric methods and opened up a whole new class of theories to be tested. As table-setting for REDD, this literature comes to its logical conclusion with Margulis (2004), a World Bank published book on Brazilian deforestation. Therein, Margulis argues that scientists and policymakers need to revise their expectations that tropically forested regions cannot support profitable agriculture. The implication is that agriculture is likely to remain profitable across the tropics, independent of state activities to promote its expansion. Margulis’ book draws on empirical evidence and analytic modeling to make the case that profitable tropical agriculture would continue to grow, independent of state subsidies. Margulis’ book is also notable for the contrast of its title, Causes of Deforestation in the Brazilian Amazon, with the extensive use of the term “drivers“ in the book itself. The geographic focus of the Margulis book is Brazil, where state-sponsored colonization remained a major contributor to agricultural expansion and deforestation (Rudel, 2007). It should be noted that the two phenomena are not independent. Colonial projects provide infrastructure, labor and sometimes raw materials for large properties. Uncovering the reasoning behind the focus of Margulis is beyond the scope of this analysis. However, in retrospect, it seems unlikely that a change in the phenomena on the ground explain the change in focus. The Margulis book seems emblematic of an important ideological shift about how to frame the roots of the deforestation problem and how to combat the problem. The shift from the public sphere as the root of the problem to the private sector as the root of the problem has been interpreted as indicative of a shift in patterns of deforestation processes themselves (Rudel, 2007). Rudel (2007) explains the trend by suggesting that the shift can be explained by the heightened role of “local and national market forces” in deforestation “after 1990.” Rudel does not entertain the possibility that an epistemic shift might have helped to shape the shift he attributes to on-the-ground change. This comes despite the fact that Rudel himself points out the continuation of the instrumental role of the state in deforestation. The upshot of this change in the explanatory literature was that it would eventually form part of the intellectual basis for the normative literature arguing for the adoption of REDD to be synergistic with the Kyoto protocol (M. Santilli, et al., 2003; M Santilli et al., 2005; Stern et al., 2006; Eliasch, 2008; Laurance, 2008; Sajwaj, Harley, & Parker, 2008). I will discuss this literature at greater length in the upcoming section on the rise of REDD.

Deforestation Policies: 1985-2009

Overview This sections review three classes of forest policies and how they cohere with scholarly works framing the roots of the deforestation problem and sites of intervention. The first policy class reviewed is regimes rooted in the principle that nation-states constitute both the problem and solution to deforestation (public-public). These include the debt-for-nature movement and multilateral forest negotiations. When these public-public policies failed, forest certification arose. Forest certification is an example of private sector as the problem, private sector as the solution (private-private). Finally, the section turns to REDD. A hybrid of public-public, and private-private, REDD is based on the principle that the private sector is the problem and that public sphere is the solution (private-public).

Public-Public Almost simultaneous with the literature on state-caused deforestation came the rise of two somewhat related governance efforts that sought to intervene in the affairs of deforesting states to halt deforestation.47 The following two sections review these efforts—debt for nature swaps and the United Nations Forests Convention. The first of the efforts, debt for nature, was conceived of by the World Wildlife Fund in 1984 and first put into practice by a team including Thomas Lovejoy and others in Bolivia in 1987. According to Hansen (1989:77):

Debt-for-nature swaps involve the purchase of a developing country's debt at a discounted value in the secondary debt market and cancelling the debt in return for environment-related action on the part of the debtor nation

These swaps were premised on a number of alignments and assumptions. First many deforesting nations were indebted to multilateral entities bankrolled by developed nations (Hansen, 1989). The debt itself was shown to be promoting tropical deforestation (Kahn & McDonald, 1995). In this way, the debt for nature literature can be seen as closely coherent with the state-sponsored deforestation literature. Both specify the activities of nation-states as the root of the deforestation problem. The debt for nature theoretical premise was that tropical deforestation was being exacerbated by nations scrambling to secure foreign exchange by facilitating the expansion of export-oriented agriculture (Gullison & Losos, 1993). 47 Admittedly this simultaneity calls into question my hypothesis that the science is causing the policy. However, the point remains that the ideas in the science are crucial and to the extent that the scholarship serves to legitimize these ideas, it remains important to highlight the scholarly contribution.

As such, forgiving such debts on the condition that states curb deforestation had two synergistic mechanisms for success—a reduction in state demand for deforestation and the legal force of the accord. In sum, debt-for-nature proponents operated under the pretense that tropical forest governments were the root of the deforestation problem and the best hope to solve the problem.

Debt for Nature Debt for nature enjoyed a strong response in Washington D.C., but received a more mixed reaction in recipient nations (Walsh, 1987; Alagiri, 1991; Humphreys, 2008). In the United States, conservation advocates met with some success in getting the World Bank to take up the model. Far from a fringe cause, several versions of the measure were introduced into legislation.48 Responses by thinkers and decision-makers in the developing world were mixed by contrast. Some governments agreed to enter into debt for nature swaps (Walsh, 1987). A more common response was that of those who inaugurated a series of long-running debates about the appropriate compensation level in the swaps. Rarely couched in actuarial terms, these debates often referred to the fundamental injustice and inequality of the foregone development associated with debt for nature (Humphreys, 2008).49 A third group rejected the premise out of hand and argued that debt-for-nature would constitute an unjust and unpalatable compromise of national sovereignty and self-determination (Alagiri, 1991). Because of bureaucratic donor-side hurdles and because of the way the debate evoked issues of injustice in the global South, debt for nature, in the form it was initially envisioned, fizzled without much impact. Key premises of the theme that cohere closely with the state-sponsored deforestation literature are (i) tropical forest governments as the root of the deforestation problem and (ii) tropical forests governments as the best solution to deforestation.

48 Such efforts ended up facing bureaucratic headwinds because the United States government does not have control over the World Bank and as such, the U.S. Congress was an ill-suited venue for the legislation. 49 Humphreys sees this approach as emblematic of a broader range of activities associated with two organizing principles of the day- the New Economic Order (NIEO) and the Non-Aligned Movement (NAM). Both groups sought redistribution of resources from rich to poor countries. Humphreys (2008) argues that even though by the period, “[these concerns] had receded into the background of international diplomacy they had never quite disappeared (pg. 436) For more, see Humphreys, D. (2008). The politics of'Avoided Deforestation': historical context and contemporary issues. International Forestry Review, 10(3), 433-442.

A Global Forest Convention? Securing a global forest treaty was a target of a number of nations and NGOs at the Rio de Janeiro 1992 United Nations Conference on Environment and Development (UNCED). This coalition was largely compromised of developed nations and it was led by the United States. The forest negotiations took on a great deal of resonance at the UNCED in part because of Brazil’s prominent role as tropical forest custodian. However, the negotiations fell apart in a dispute centered on the appropriate price for tropical nations to sell away their sovereign rights to develop. Scholarly disputes persist about the extent to which the discord was more a function of inflexibility by tropical nations or by the United States (Davenport, 2005; Dimitrov, 2005). However, regardless of blame for the failure, it seems clear that the price of sovereignty debate in Rio constituted a close recapitulation of debt for nature politics and was derailed for similar reasons. Instead of a forests’ treaty, the UNCED produced a series of non-binding accords that have produced some advances in setting forestry management standards and developing frameworks to curb illegal logging, but have little or perhaps even a chilling effect on progress to fight tropical deforestation (Gulbrandsen, 2004; Dimitrov, 2005; McDermott, O'Carroll, & Wood, 2007). At the UNCED, the parties adopted the Forest Principles, a set of non-binding accords. According to Humphreys (2008), two controversial sticking points in these negotiations were text on “common, but differentiated responsibilities [for developed and developing nations]” and attendant considerations about compensation for avoided deforestation. Developed nations were particularly concerned about the inclusion of “opportunity costs foregone [by the governments of the developing nations themselves].” The Forest Principles ultimately resulted in a series of ineffective institutions. Soon after the UNCED parties formed an International Forum on Forests. This lasted three years, to be succeeded by the United Nations Forum on Forests (UNFF). Fifteen years after the Forest Principles, the UNFF “agreed to a non-legally binding instrument on all types of forests.” (Humphreys, 2008:437). Just six months later Parties to the UNFCCC adopted the Bali Action Plan including a call for:

policy approaches and positive incentives on issues related to relating to reducing emissions from deforestation and forest degradation [REDD] in developing countries (http://unfccc.int/files/meetings/cop_13/application/pdf/cp_redd.pdf)

Even as REDD continued to lurch forward, the UNFF continues to meet to discuss:

progress made on the implementation of the U.N. non-legally binding instrument on all types of forests and towards the achievement of the four global objectives on forests (http://www.un.org/esa/forests/documents-unff.html#10)

Whether by design (Dimitrov, 2005) or for other reasons, the stream of international forest governance begun at the UNCED remains moribund and ineffectual. The framing of state-led efforts to fight state-led deforestation never gained traction.

Private-Private

Forest Certification Overview Immediately on the heels of the Rio Earth Summit, non-governmental forest certification rapidly flowered with the founding of the Forest Stewardship Council (FSC) in Oaxaca, Mexico in 1993 (Bartley, 2003). In the years that followed, FSC gained popularity swiftly and a flood of other institutions of non-state market driven governance in the forest sector emerged (Cashore, 2002). In this section, I chart the rise of forest certification for two purposes. I compare its uptake and attributes with the multilateral forest governance regimes and I examine the extent to which it served as source code for REDD’s site of interventions and roots of the problem.

Forest Certification History The founding of FSC was a relief for a cadre of environmental NGOs frustrated in the wake of several failed or at least incomplete efforts to tack deforestation (Bartley, 2003; Gulbrandsen, 2004). In addition to the Earth Summit treaty failure, a number of government bans on tropical timber fell apart in the faces of challenges under the General Agreement on Tariffs an Trade (GATT) (Bartley, 2003). The idea for certification actually preceded the forest treaty optimism that marked the Earth Summit. During the height of campaigns against deforestation in the 1980’s, environmental groups had lobbied the International Tropical Timber Organization (ITTO) to oversee an international forest certification program (Bartley, 2003). However, this effort enjoyed little traction and attention strayed in the lead-up to Rio. The forest governance vacuum of the period undoubtedly aided the uptake of forest certification, but the rise of certification likely had several other prerequisites. Members of the coalition that formed to support forest certification efforts seem to

have been attracted by a mixture of moral, pragmatic, and cognitive priorities (Cashore, 2002).

Forest Certification Effects on Deforestation Policy & Politics Forest certification has changed the political landscape for forest governance and it has exchanged insights on governance with other forest governance efforts. For some NGOs, support for certification schemes and themes have come to serve as surrogate activities for efforts to prevent deforestation including boycott campaigns and efforts to reform certain government activities (Bartley, 2007). Schemes have also changed the organizational makeup of timber firms themselves, and the agenda of multi-laterals including the World Bank (Auld, Gulbrandsen, & McDermott, 2008). Some of these techniques are now brought to bear in the REDD regime (Kanowski, McDermott, & Cashore, 2011). Meanwhile, a number ideas originating in the UNFF have diffused into the forest certification realm (Gulbrandsen, 2004).

Private-Public: REDD Here I provide a narrative of the rise of REDD focused on how several elements LCS and practice have been foundational to the guiding principles of REDD. Numerous authors have traced and explained key facets of the rise of REDD. It would be redundant to retell the story in full. The abridged version of the foregoing literature is that REDD is an appealing regime due to:

• Cost-competitiveness with other GHG mitigation strategies, biodiversity conservation, ecosystem services provision, promotion of rural livelihoods, and a sense that GHG benefits could be immediate and buy time for a global climate deal to materialize (Boyd, 2010; Agrawal, Nepstad, & Chhatre, 2011; Kanowski, et al., 2011; Stephan, 2012)

Nevertheless, many of the same authors have raised concerns. These include risks of:

• Recentralization of forest governance, loss of indigenous rights, the promotion of low biodiversity plantations, forestalling other climate action, and leakage of deforestation, impermanence, and corruption (Corbera, Estrada, & Brown, 2010; Phelps, Webb, & Agrawal, 2010; Kanowski, et al., 2011; Stephan, 2012).

History The Coalition of Rainforest Nations (CfRN) made the first public use of the term REDD in 2005, but the intellectual basis traces to a presentation at the 2003 COP to the UNFCCC in Milan (M. Santilli, et al., 2003). According to Schlamadinger et al. (2005):

At a COP9 side event, Santilli et al. (2003a) presented a new proposal to include deforestation avoidance in tropical countries under the KP. The proposal labelled [sic] “compensated reduction” includes as its main element a voluntary, national deforestation stabilization and reduction target for non-Annex I countries such as Brazil or Indonesia. Its objective is to encourage conservation policies. If these policies prove successful by the end of the first commitment period, the respective carbon dioxide (CO2) reductions, once monitored and verified, can be sold to industrialized countries after the end of the first commitment period at the carbon market prices prevailing at that time (Santilli et al., 2003b).

In the author group were several LCS scholars. Daniel Nepstad, at the time co-author of several widely cited studies highlighting the role of commodity agriculture in deforestation in the Brazilian Amazon; Lisa Curran an expert on similar themes in Indonesia, and Steven Schwartzman, lead author of a widely cited paper on the limitations of protected areas to prevent deforestation.50 The Santilli et al. (2003) proposal has three salient elements. First, the authors argue that the private sector (i.e. landholders and commodity markets) are the root of the deforestation problem.51 Second, the authors suggest that a market mechanism could slow deforestation by making forest land use more lucrative for landholders than alternative land uses.52 Finally, the authors argue that tropical states are best suited to administrate and enforce these transactions and thereby serve as the primary intervening agents to mitigate the problem.53

50 Lead author, Santilli and author, Moutinho were at the time known as experts in Brazilian domestic policy making. Nobre is a Brazilian climate scientist well known for science and policy efforts concerning the landscape level effects of deforestation in the Brazilian Amazon. 51 The authors state that, “global integration of markets and demand for agricultural commodities appears to be driving substantial increases in deforestation rates.” (page 271) 52 Tropical country governments can reduce deforestation through adequate funding of programs designed to enforce environmental legislation, support for economic alternatives to extensive forest clearing (including carbon crediting), and building institutional capacity in remote forest regions, as recently suggested in part of the Brazilian Amazon (page 272) 53 “Tropical country governments can reduce deforestation [so long as they receive] adequate funding…”

Table 1-1 shows how the Santilli team creatively combined forest certification’s new framing of the forest governance problem (the private sector as the root of the problem) with an old vestige of the debt-for-nature discourse (tropical forest states as the best mitigation agents). As others have argued, the market mechanism element was crucial too (Stephan & Paterson, 2012). A market mechanism could conceivable with any cell in the 2X2 (for example, see Grainger, 1997 for a proposal to compensate governments for their opportunity costs). However, the logic and calculus of private sector opportunity costs is far more straightforward than public sector opportunity costs.

Root of the Problem (but especially analogous to

Turner’s “driving forces” i.e. those who must change their activities)

Public Private

Solutions/sites of

intervention (analogous to

Turner II’s “mitigating forces” i.e. actors with

the agency to prevent

deforestation)

Public Debt for Nature, UNFF (1)

REDD (3)

Private Commodity Chain Certification (4?)

Forest Certification (2)

Table 1-1. Schematic depicting framings of the problem and solutions to deforestation (1989-2012). The cells contain forest governance frameworks. The numbers chart the order of appearance of these frameworks. In the paper I argue that the REDD proposal contained a private sector target (cell 3) similar to that of Forest Certification while returning to the public sphere as an intervention site that was the focus of Debt for Nature/UNFF. In recent months, discussions of Roundtable REDD have emerged. Here the idea would be to use agricultural commodity chains as the site of intervention. This is in part a re-recognition of the role of governments as driving forces.

This matrix also helps me to structure discussion of the role of LCS in birthing REDD. I argue that the retreat from strong causal claims enabled LCS to rapidly construct and reconstruct the anatomy of the deforestation problem, in a normative quest for a policy-salient combination of sites and roots.

Uptake The pathway for adoption of this idea in the UN process led through Papua New Guinea (PNG) and Costa Rica by way of a team from the Earth Institute and the Graduate School of Business at Columbia University in New York. At Columbia the key actor was Kevin Conrad, the son of Papua New Guinea-based American missionaries, an alumnus of an executive MBA program at the Columbia Graduate School of Business, and family friend of former PNG Prime Minister Michael Somare. With the support of former professors including the Nobel Laureate Joseph Stiglitz and the noted environmental economist Geoffrey Heal, Conrad founded the

Coalition for Rainforest Nations (CfRN).54 In ensuing years, CfRN has come to represent the interests of dozens developing nations. Member governments control nearly half of the world’s tropical forests. In the lead-up to the bold proposal by Costa Rica and Papua New Guinea at the Montreal COP, CfRN and Columbia provided conceptual support, media savvy, and legitimacy. PNG Prime Minister Somare delivered a graduation address to the Columbia Graduate School of Business in May 2005, that Conrad researched and may have written. In the immediate lead-up up COP 11, Stiglitz helped to frame the proposal in interviews he gave to the Los Angeles Times and other media outlets. The upshot was a polished effort that comported with the state of the science and the zeitgeist of the multi-laterals. The fact that it was delivered by heads of state of tropical nations likely also heightened its popularity among developing nations jilted by previous standoffs over justice and reparations that had typified forgoing debates on tropical deforestation. The broadly warm reception that the proposal received has been widely described in the media and in the scholarly literature. The Costa Rica and Papua New Guinea proposal did face competition from other REDD proposals in the early years. Perhaps the most notable alternative was Brazil’s proposal for funding for compensated reduction to come as overseas aid instead of through linkage with compliance markets. While the differences between these two proposals are real and continue as a source of debate to the present, the Bali Action Plan adopted in 2007 is suitably vague on the subject. To date, either proposals could win out.55

54 The origins of CfRN date at least to May 2012. Then in a graduation speech to the Columbia Graduate School of Business, Prime Minister Somare called for a, “Coalition of Rainforest Nations.” 55 There have been dozens of REDD mechanisms proposed since 2005. For more see, Parker, C., Mitchell, A., Trivedi, M., & Mardas, N. (2008). The little REDD book: a guide to governmental and non-governmental proposals for reducing emissions from deforestation and degradation. Retrieved July 25, 2012, from www.theredddesk.org . For an analysis of the principles of justice embodied in the proposals see, Okereke, C., & Dooley, K. (2010). Principles of justice in proposals and policy approaches to avoided deforestation: Towards a post-Kyoto climate agreement. Global Environmental Change, 20(1), 82-95.

Figure 1-6. Notable events in the development of forest governance. This figures juxtaposes events in the development of forest governance described in this chapter with the level of coverage of forest matters in the New York Times. The results are based on a LexisNexis keyword search of the archives of the Times, 1976-2011. The search return all articles mentioning one or more of the following three words or phrases: (1) rain+forest (2) rainforest (3) deforestation.

The Cancún Decision Broadens REDD Norms This section examines and contextualizes several elements of the COP to the UNFCCC decision in Cancún that I argue have the potential to substantially broaden the guiding principles of REDD. I begin by describing the three elements of the Cancún decision that I see as potential premise-broadening elements. Next, I explore the actions negotiators undertook to support and shift these items. Following that I summarize reactions to the language. The section concludes with an analysis of language’s timing, importance, and significance.

Cancun in context Held in the November and December of 2010, the Cancún COP to the UNFCCC was dedicated to building the foundations of an agreement to replace the Kyoto Protocol. The tasks at hand were all supposed to have been completed one year earlier in Copenhagen, but the chaotic and frenetic nature of that COP precluded progress (Akanle et al., 2010). Copenhagen was also a source of procedural controversy as a hastily arranged, closed-door meeting of the world’s largest economies supplanted the conventional procedure in the final days of the gathering. In the end, the Accord struck among the major economies garnered a neutral response from the broader assembly (Akanle, et al., 2010). As Copenhagen closed, climate insiders worried over the stagnation of progress towards a new climate deal and also about cooption of the process. In Cancún, the host Mexico stressed procedure early in negotiations, but closed with heavy-handed tactics. Patricia Espinosa, the President of the COP, gaveled over Bolivia’s unitary objection at the conference’s close in order to ratify the Cancún Agreement (Vidal & Goldenberg, 2010). Stipulations on REDD comprise a major portion of the agreement. The REDD text focuses most heavily on regulatory dimensions. Monitoring, reporting, and verification protocols is the primary focus. Safeguards ran a close second initially, but at the 11th hour, much of the safeguards language was stripped from the text (Zwick, 2010). The bulk of the rest of the REDD text concerns the scope of the regime. At issue is how to handle subnational REDD efforts, how to deal with “enhancement of forest carbon stocks”, the role of the non-forest land uses in REDD, and “drivers of deforestation.”

“Drivers” and Change? Four related components of the Cancún decision have the potential to transform the REDD regime by grappling with “drivers” and thereby shifting the regime away from present model (private sector as the root of the problem, tropical forest governments as the site of intervention). Three of the four components come in Section C.56 The fourth component comes in Appendix II, a further explanation of portions of Section C. The first component is

56 Section C covers “Policy approaches and positive incentives on issues relating to reducing emissions from deforestation and forest degradation in developing countries; and the role of conservation, sustainable management of forests and enhancement of forest carbon stocks in developing countries”

paragraph 68. 68 is a hortatory paragraph, it contains no requirements, only exhortations. It reads:

Encourages all Parties to find effective ways to reduce the human pressure on forests that results in greenhouse gas emissions, including actions to address drivers of deforestation;

Paragraph 72 begins with the word “also” because it continues on from paragraphs 70 & 71, a long list of requests for developing countries parties in order to prepare for REDD policy. The notable element of paragraph 72 is again the phrase “drivers of deforestation.” In 72, the implication is that addressing “drivers of deforestation” is a requisite ingredient for effective “national [REDD] strategies or action plans.” The final element of note in the Cancún decision is Appendix II. Appendix II is a set of tasks generated by the Cancún decision on REDD for the Subsidiary Body for Scientific and Technological Advice (SBSTA) to take up going forward.57 SBSTA task (a) is the portion of Appendix II of note. It request that SBSTA:

57 Officially, Appendix II is them, “Work programme of the Subsidiary Body for Scientific and Technological Advice on policy approaches and positive incentives on issues relating to reducing emissions from deforestation and forest degradation in developing countries; and the role of

Here we find work plan tasks that may hint slightly at the intended meanings of the phrase “drivers of deforestation” in its appearances in Paragraphs 68 and 71. The appendix suggests that “drivers of deforestation” as things that can be “linked” with “land use, land-use change, and forestry activities.” This is admittedly vague, but notable because it is not necessarily consonant with the principles of private sector as the root of the problem and the public sphere as the site of intervention. Note that the first appearance of the phrase “drivers of deforestation” in official UN documents appeared nine months prior to Cancún. The notes from the February 2010 17th meeting on of the Ad Hoc Working Group on Long-Term Cooperative Action Under the Convention (AWGLCA) has this paragraph:

Requests developing country Parties when developing and implementing their national strategy or action plan, [or subnational strategies] to address, inter alia, drivers of deforestation (FCCC/AWGLCA/2009/17 5 February 2010: pg. 37)

Will Defining Drivers Transform REDD? To explore how the “drivers” shift may change guiding principles of the REDD regime, I conducted interviews with negotiators and advocates and studied official and unofficial text from period from December 2011 to April 2012, in the lead-up to the Bonn SBSTA 2012. The Bonn SBSTA was intended to be a venue for a discussion on the scientific dimensions of drivers. Interviews were largely open-ended, but always centered on how the subject understood the role of the drivers text in the broader context of REDD. I present the results in a conceptual

conservation, sustainable management of forests and enhancement of forest carbon stocks in developing countries”

diagram—a series of nodes depicting touchstone themes of the drivers debate and the links representing how experts understand linkages across the themes. The nine nodes represent the most common elements that the drivers evoked for the interview subjects.

REDD Out Ahead A phenomenon that I term REDD Out Ahead is a recurring theme for both REDD negotiators and REDD policy advocates, but for different reasons. For negotiators, REDD out ahead is indicative of a fear that of a lack of coordination between REDD and other elements of the climate schemes. Several of my interview subjects sees this as stemming in part from a lack of capacity for some parties to monitor all streams of the negotiation. A related concern is the unwillingness or inability to assign sufficiently competent staff to the REDD negotiations themselves. On the flip side, some advocates frame the rapid progress of REDD as a point of pride and evidence that the UNFCCC is a viable venue for climate policy. Another set of advocates and scholars, meanwhile argue that the rapid progress of REDD is indicative of the market orientation of REDD and attendant concerns about issues of environmental justice.

Agriculture Every one of my interview subjects spoke of drivers in a way that evoked agriculture. Despite that fact that agriculture is not at present mentioned in the REDD text, the interviewees suggest that the term “drivers of deforestation” is very nearly a proxy for agriculture. Some subjects see this is development as foundational for the future of REDD. Generally speaking, this camp argues that “stovepiping” of land use issues is to be avoided. The concern is that this is inefficient for monitoring, and that “you can’t get there from here”, meaning that reducing tropical deforestation appreciably cannot occur without addressing agricultural issues. However, an even larger portion of the party representatives I interviewed see agriculture as what one interviewee described as the “third rail” of international governance in general and in particular environmental governance. “I know we can’t get there from here without agriculture,” one subject told me, “but look, I am as big an advocate for a new treaty as they come and with agriculture in the mix it’ll never happen.” When pressed, the rationale is that the agriculture-deforestation link inherently involves grappling with international trade of agricultural commodities. One respondent described this as a, “$100 billion dollar issue.” Among, advocates, however, there is a strong push to get agriculture on the

REDD bandwagon. One advocate describes REDD as the last best chance to revive overseas development assistance for agriculture (B. Campbell, 2009).

Issue Creep Issue Creep is a theme closely related to REDD out ahead. Similarly, it appears as a procedural concern in my analysis as a concern of party representatives interviewed. It refers to a sort of negotiating philosophy that when a regime comes to cover too many issues, and therefore spawns too many negotiating streams that it can stall out before adoption. In addition, the proliferation of negotiating streams can test the capacity of some parties to participate in everything.

Trade Trade and trade laws emerged in a majority of the interviews with parties and in some grey literature related to REDD. For the parties, there is a sense that in order to address drivers it would be necessary to vigilant about the strictures of trade and the pitfalls of compromising the interests of traders. One interview subject instrumental in authoring the drivers text in the Cancún decision told me that it had been necessary to water down the drivers language with an eye towards the reaction of watchful trade representatives.

Polycentricity For several interviewees, the drivers theme heralds a broader trend in decentralizing the principles and content of REDD. These feelings closely cohere with how Boyd (2012) describes the emerging polycentricity of REDD. In the piece, Boyd describes a heterogeneous network of firms, governments, and advocates developing REDD institutions simultaneously. He uses the metaphor of DNA to describe how frameworks and approaches from any particular effort can be combined and recombined into other efforts without passing through the bureaucracy of the UN. One respondent noted that a good deal of finance under the banner REDD has already begun to address drivers, well before the concept has acquired substantive meaning in the UNFCCC. “It’s hard to know which is the cart and which is the horse sometimes,” observed the respondent.

Finance For party and advocacy interviewees alike, the drivers themes is closely associated with a gathering debate about the appropriate uses for REDD finance. Several respondents noted that agricultural supply chains might be better suited to REDD finance than traditional government targets and forest landholders. Among this

group, there was a sense that not only had existing lending been ineffective, it had been, as a consequence, heavily delayed by challenges finding appropriate borrowers and grantees. On the flip-side, some advocates who are currently beneficiaries of REDD finance schemes argued that drivers are a low priority distraction that could threaten a “successful” model for REDD. Yet another position questioned the wisdom of finance targeted for monitoring and evaluation of REDD at the expense of direct support for private sector activities. This subject argued that drivers might be helpful in rechanneling finance out of government hands and into the private sector. One respondent remarked that the insertion of drivers indicates a balkiness about the role of offsets from compliance markets to fund REDD activities. Drivers, this subject argued, indicates a growing credence in the “fund model” of REDD.

Demand-side and Supply Chain In my interviews, several respondents keyed in on the “all parties” language in Paragraph 68 as signal of a focus on policy targets beyond landholders themselves and instead along land use-intensive supply chains. Several respondents saw the opportunity and some the need for synergy between formal REDD activities and other complementary government and NGO initiatives. Potential linkages mentioned included the commodity roundtables (as a source of norms for agriculture, of stakeholders for REDD, and as an institution with which to implement supply chain interventions). Others expressed hope that the legal model to crack down on the international trade of illegal tropical timber could be adopted for agricultural products.58 However, several parties have expressed skepticism about the political and legal potential of such a measure. The European Commission, meanwhile, has called for further research on the matter. In sum, demand-side and supply chain approaches to REDD represent a very different set of guiding principles than REDD as usual. In a sense, such approaches constitute the fourth cell of the 2X2—the public sphere as the root of the problem, and private sector as the site of intervention.

Whole Landscape Whether glad, wary, or undecided on drivers in the text, many subjects suggest that the term is proxy for discussion about tropical land use, broadly writ. Several subjects use the phrase whole landscape to refer to this trend. One party explained

58 This subject referred to the Lacey Act in the U.S. and FLEGT in Europe as potential models for an agricultural commodity trade policy addressing deforestation.

that the insertion of the drivers language was an attempt to “revisit Marrakech” over issues of AFOLU.59 At issue for some subjects was how this might enable more effective REDD investments. Others seemed worried about how this might generate grinding debate over redistribution of slices of the “REDD pie.”60

Reference Levels Finally how drivers relate to reference levels was a less common, but highly charged theme. Reference levels are estimates of the rate of deforestation in the absence of REDD. They are important to parties because they will form part of the basis for determining performance payments.61 In turn, drivers are important to reference levels because drivers are factors associated deforestation. One subject speculated that the high stakes of reference levels could pull along work to define drivers, but only if drivers are included in the reference levels discussion.62 Another, disagreed, worrying that linking the charged atmosphere of the reference levels debates with the efforts to define drivers could derail progress on defining drivers. The concern was that reference levels might politicize drivers and thereby limit progress to define them rigorously.

59 By this, the party was referring to the Conference of the Parties to the UNFCCC in Marrakech in 2001where a tepid agreement was reached on the accounting of emissions from land use in Annex One countries. Marrakech resulted in the granting of the ability for Annex I countries to mitigate in using land use activities. However, non-Annex One countries were excluded at the time. 60 “REDD Pie” is negotiator slang for financial assistance. 61 Reductions between an observed deforestation rate under REDD and the reference level deforestation rate are the basis for payments. 62 One recent paper proposes to do just this. For more see Herold, M., Verchot, L., Angelsen, A., Maniatis, D., & Bauch, S. (2012). A step-wise framework for setting REDD+ forest reference emission levels and forest reference levels (pp. 8p). Bogor, Indonesia: Center for International Forestry Research (CIFOR).

Figure 1-7. A constellation of new themes “drivers” evoke for REDD experts.Based on quantitative coding of interviews with experts this figure displays: (1) the nine main themes that experts see as linked to drivers (2) the frequency with which experts referred to these themes (depicted by the diameter length of the circles) (3) the linkages experts saw between themes (calculated as the proportion of experts who seen theme pairs as linked and depicted as the distance between circle centers of the pairs).

Concluding Thoughts I have shown how a set of theoretical frameworks, developed in the early years of Land Change Science, have been foundational for establishing norms about how the REDD regime can and should operate. By defining deforestation as rooted in the activities of private sector and the failure of markets and potential suited to be controlled by the public sphere, these scholars have helped develop the foundational principles for REDD. And, I draw on text analysis and expert interviews to explain why implementing REDD sparked the inclusion of the term drivers of deforestation in text, and how this phrase, born out of Land Change

Science itself is sparking the need to revisit these theories of change with potentially profound impacts for REDD’s future prospects. Mine is hardly the first paper to argue that science in general and Land Change Science in particular have enabled and profoundly shaped REDD. The use of satellite-based remote sensing to “see” tropical deforestation is foundational to REDD (Boyd, 2010). Almost as soon as LCS scholars begin identifying these patterns, they began wonder how to explain them and also what to do to slow them. This led to a sort of normative theory-making with an emphasis on problem solving. In sum, LCS findings, theories, and norms are all foundational to the REDD regime. In light of the somewhat tired old discussion about norms and objectivity in the climate sciences, this might appear a familiar story. However, it is important to note, that the norms implicated here do not always straightforwardly map to policy aspirations by the scholars with the norms. Instead they emanate from epistemological choices and even choices about research foci that scholars may be making for entirely different reasons. These might be consonant with normative policy stances (i.e. the sense of urgency to fight global environmental crises), but they are far from examples of scholars manipulating policy. Instead, they pose an influence on policy mediated by chance and circ*mstance. So what agency might the theorizing of agency afford LCS? This depends on where REDD may lead and that depends on who one asks. Some see the initiative as growing more robust by the month, while others see it is a staggering under the loss of key incentives for support. Whatever the future, LCS has and may continue to provide foundational principles for REDD to develop. In a world without such science, REDD might still have occurred, but would likely not have had the premise of tasking governments to compensate landholders. Though that premise is a byproduct of a potentially critique-worthy set of assumptions, it seems to have had the crucial effect of increasing dramatically the robustness and perhaps capacity of institutions for forest governance. In this way, these ideas have helped to set up an experiment of the capacity for this constellation to adapt and assimilate, after its formation, the universe of Land Change Science in all its complexity and multi-causality. Perhaps within these institutions, the sophistication of concepts will “ratchet up” alongside stringency of enforcement. Indeed one interview subject suggested that, “the pendulum might be swinging back to tackling [state-led aspects of deforestation].”

Given the sheer number and variety of institutions and how their connectedness seems to reproduce this challenge everywhere, these experiments affords unparalleled opportunities to identify the resilience and robustness of institutions of governance as paradigms shift and grow. This is fertile ground for future scholarship on how scholarly ideas shape institutions of environmental governance.

Chapter Two: Global Trade Can Provide and Scatter GHG Benefits from Cattle Ranching Intensification Policies in Brazil63

Introduction

Policy Context Agricultural interventions to reduce deforestation are a focus of a growing number of policies, funding efforts, and research activities (Renewable Fuels Agency, 2008; Sajwaj, et al., 2008; Ecofys & Winrock International, 2009; Manzatto, Assad, Bacca, Zaroni, & Pereira, 2009; Nepstad et al., 2009; Tim Searchinger & Amaral, 2009; Angelsen, 2010; Embassy of Brazil, 2010; Gouvello, 2010; Lapola, et al., 2010; Amend, dos Santos, & Mattos, 2011; Boucher et al., 2011; Pinjuv, 2011). The hope is that higher agricultural yields can limit demand for agricultural land and thereby complement and enhance other deforestation policies in a process sometimes called land sparing (Balmford, Green, & Scharlemann, 2005; Green, Cornell, Scharlemann, & Balmford, 2005). Brazil has become a focal point for reducing agricultural drivers of deforestation through a mixture of political will, salience, and necessity. It is home to a rapidly growing and evolving agricultural sector, it is the largest source of emissions from deforestation in recent decades, and it retains the globe’s largest stock of tropical forest (Nepstad, et al., 2009). Emissions from deforestation and agriculture dominate national GHG emissions, necessitating that the Brazilian National Climate Mitigation Plan (PNMC) target land use interventions in order to achieve meaningful reductions (Cerri, Bernoux, Maia, Cerri, Costa Junior, Feigl, Fraz√£o, et al., 2010). Interventions to shift technologies and land uses in Brazilian cattle pasture systems are at the heart of the land use mitigation activities outlined in the National Climate Plan of Brazil (Federal Government of Brazil, 2008; Embassy of Brazil, 2010). Cattle pastures constitute one fourth of the land surface of Brazil or approximately 200 million hectares (Census Bureau of Brazil, 2010). Roughly 200 million cattle inhabit these pastures, but their distribution is highly uneven (Census Bureau of

63 This chapter is the product of a collaboration with colleagues Aline Mosnier, Hugo Valin, Petr Havlik, Michael Obserteiner (IIASA), Mario Herrero (ILRI), Erwin Schmid (Boku-Vienna), & Mike O’Hare (UC Berkeley)

Brazil, 2010). Animal numbers and productivity are concentrated on the one fourth of the area of pasture systems where semi-intensive technologies are employed (Census Bureau of Brazil, 2010). These semi-intensive systems are not explained strictly by biophysical productivity potential, but also by market access, and by choices of ranchers to adopt one or more semi-intensive management practices (Ricardo, 1817; J. von Thunen, 1966; Just & Zilberman, 1983). The technology portfolio includes improved grasses, management, breeding and feeding to boost outputs without greatly increasing production costs (Sampaio et al., 2001; Malafaia et al., 2004; Euclides et al., 2010). The existence of these cost competitive alternatives has fueled hope that substantial climate benefits could accrue from interventions that succeed in accelerating uptake of semi-intensive cattle pasture technologies across a larger portion the Brazilian pasture landscape.

The Literature Gap Previous studies suggest that accelerating adoption of higher output, semi-intensive Brazilian cattle pasture intensification technologies can reduce direct GHG emissions, slow deforestation rates, and secure GHG benefits from biofuels production (Gouvello, 2010; Lapola, et al., 2010; Thornton & Herrero, 2010). Adoption of one or more semi-intensification technologies can directly reduce emissions from cattle systems by limiting enteric methane produced per unit meat and by reducing nitrous oxide emissions (Thornton & Herrero, 2010). Direct reductions are notable, but they are an order of magnitude smaller than reductions estimated from avoided deforestation due to adoption of semi-intensification systems. Such reductions could deliver a substantial fraction of the GHG mitigation pledged in the PNMC, Brazil’s Climate Law (Gouvello, 2010). Previous research suggests that land sparing from pasture intensification could also readily sop up the additional demand for land associated with biofuels mandates (Tim Searchinger & Amaral, 2009; Lapola, et al., 2010). However, previous estimates of the land sparing potential of cattle pasture intensification have shortcomings in their depictions of production systems, land use change processes, policies, and GHG impacts. One problem is the non-mechanistic representation of agricultural intensification processes. Rather than delineate all changes to the material and financial flows through the cattle systems, some studies depict boosted output alone. This approach does not allow representation of the full scope of mechanisms by which intensification shifts land use and therefore may misrepresent outcomes resultant from intensification processes. In such studies, intensification balances the spatial

budget instead of depicting the land use change process using mechanisms consonant with economic principles and used at the cutting edge of land change science (Segerson, 2006; Seto et al., 2012). Second, policies are also often highly stylized, depicted through constraints without practical analogs (Cohn, et al., 2011). Some modeled policies are implemented globally, belying limits to the authority of international law (Wise et al., 2009; Reilly et al., 2012). Third, though greenhouse gas emissions are of global salience and land use change processes are increasingly mediated by global markets, previous studies have focused on land use and greenhouse impacts at scales ranging from local to national (A. Angelsen & D. Kaimowitz, 2001; A Angelsen & D Kaimowitz, 2001; Gouvello, 2010; Lapola, et al., 2010). Global impacts from national land sparing policies have not been an explicit focus.

Our Contributions This study investigates the global land use and GHG impacts of pasture intensification policies for achieving compliance with cattle ranching intensification targets specified in the Brazil National Climate Plan. We focus explicitly on three aspects of the problem that previous studies have not adequately addressed—a bottom-up, mechanistic representation of cattle production systems, policy interventions with practical analogs, and the effects of global trade. We employ the global economic land use simulation model GLOBIOM (Havlik et al., 2011; Schneider et al., 2011) adapted to Brazil in order to represent global land use and GHG impacts of changes to land policies and pasture technologies in Brazil. The model consists of (1) spatially explicit estimates of global crops, rangelands and timber production potential (2) spatially explicit product specific internal transportation costs in Brazil (3) an economic model which represents the competition for land between the forestry, agriculture, and livestock sectors, and (4) international trade for crops and livestock products. We introduce a semi-intensive pasture based cattle ranching alternative production type into the model. The production offers boosted yield per hectare and boosted costs of production per hectare. The main model outputs are crops, livestock and timber production, land use change, and GHG emissions from land use and agricultural sector, at a 50x50km resolution in Brazil for each 10 year-period over 2000-2030. Minimum levels of global food, timber and bioenergy demand are exogenously specified based on assumptions about population and economic growth.

We model two intensification policy mechanisms—a flat subsidy for all ranches adopting intensive alternatives and a flatly levied tax on all ranches that do not adopt intensive alternatives. Both of these interventions reduce the cost gap between higher cost semi-intensive alternative pasture systems and lower cost, business as usual pasture systems. We explore the extent to which international trade modulates the effects of these policies by comparing the model outcomes of freezing Brazilian external trade to the level observed in the baseline simulation or allowing external trade to adjust to the introduction of pasture intensification policies. The GHG impacts of the intensification policies are accounted as the difference between all GHG emissions over the modeled baseline scenario and all GHG emissions in each of the policy scenarios.

Figure 2-1. Political regions of Brazil. Simulation results are arrayed according to these regions. Note that the North region roughly corresponds to the Amazon Biome.

Results

Baseline Simulation

The Brazilian Cattle Sector In recent decades, the cattle sector of Brazil has expanded, shifted North and West, intensified, and become more export-oriented (Barona, Ramankutty, Hyman, & Coomes, 2010; M. S. Bowman et al., 2011; Pacheco & Poccard-Chapuis, 2012; N. Walker, Patel, & Kalif, Forthcoming). Brazil has the largest commercial cattle herd, the second largest share of international trade in cattle products, and widespread adoption of semi-intensive pasture technologies. Semi-intensive adoption is largely associated with good domestic market access and fertile soils—higher rates of output per area can be found in the South, Southeast, and parts of the Center West of the country.64 Meanwhile, the geographic center of cattle production has continually migrated North and West in Brazil as an integral part of the frontier settlement and deforestation process (Barona, et al., 2010; M. S. Bowman, et al., 2011; N. Walker, et al., Forthcoming). Internal demand for Brazilian cattle products is strong and growing. Per capita beef consumption is high and per capita consumption relative to GDP is higher. This despite the fact that prices are high relative to other countries of similar socio-economic status (M. D. Faminow, 1997; McAlpine, Etter, Fearnside, Seabrook, & Laurance, 2009; Millen, et al., 2011; Association of Brazilian Beef Exporters, 2012). Pasture still accounts for over 90 percent of animal nutrition in Brazil (Dias, 2007). Confinement facilities have emerged in the past decade, but they are cost competitive with pasture during only the height of the dry season and only in the core agricultural regions where land prices are higher than is typical (Dias, 2007). Age at slaughter has declined somewhat in the past two decades, but it still lags well behind producers in the U.S. and Europe (Association of Brazilian Beef Exporters, 2012). Growth in the sector has been uneven—fluctuations in currency, purchasing power, and export market access temper and enhance typical price-dependent managerial fluctuations in slaughter rate and age class ratios known as the cattle cycle (Jarvis, 1974).

64 Some recent research has begun to examine whether export market access may play a role in land dynamics, including influence on the intensity of production. For more see Faria, W., & Almeida, A. (2011). Agricultural Expansion, Openness to Trade and Deforestation at the Brazilian Amazon: A Spatial Econometric Analysis. Paper presented at the European Regional Science Association, Vienna, Austria, Bowman, M. (2012). Does Growth in Demand for Brazilian Beef Exports Drive New Deforestation?, February. New York: American Association of Geographers.

Figure 2-2. Distribution of cattle in Brazil by political region over time.

Our baseline simulation depicts the period 2000-2030 in a fashion broadly concordant with the aforementioned patterns. Beef production in Brazil is simulated to increase to 13.4 million tons of carcass weight equivalent by 2030. This increase constitutes a modest 50 percent rise over production in 2011. Most of this increase will be associated with exports. We simulate Brazilian beef exports will to grow to represent 40 percent of the production in 2030 compared to 8 percent in 2000 and 24 percent in 2007 as reported by the United Nations Food and Agriculture Organization (FAOSTAT, 2012). The number of head of cattle is simulated to increase in all five political regions of Brazil across the simulation, but differentially. The Center West is simulated to remain the top cattle-producing region but its share would decrease due to rapidly rising cattle production in the North region (Figure). Part of the shift comports is due to agricultural expansion in the Center West and Southeast regions. Simulated crop farmers are willing to pay more for land than ranchers in these high fertility, market proximate regions.

Brazilian Land Use and Land Cover Prior to 2004, growth and change in the Brazilian cattle sector closely paralleled conversion of savannah and tropical forest in the North and West of Brazil

(Nepstad, et al., 2009). Remote sensing studies report that cattle ranching occupies between 50 percent and 80 percent of cleared Amazon rainforest land at some point in the initial years following clearing (Federal Government of Brazil, 2004; M. Bustamante, Nobre, & Smeraldi, 2009; National Institute of Space Studies of Brazil (INPE) & Brazilian Agricultural Research Corporation (EMBRAPA), 2011). However, in key frontier regions, the last five years has seen growth in cattle production unmatched by concomitant land conversion (Macedo et al., 2012). This discontinuity has been interpreted as promising evidence of land sparing, but it also highlights the challenges of estimating the extent to which marginal growth in the cattle sector contributes to deforestation. We revisit this theme in our discussion session. Nevertheless, in some periods, cattle pasture and crops expand in a concerted fashion. New croplands arise closer to key infrastructure while new pasture arises close to the agricultural frontier (Barona, et al., 2010; E.Y. Arima, Richards, Walker, & Caldas, 2011; de Sa, Palmer, & di Falco, 2012). However, it is often the case that croplands arise through conversion of pasture. The net effect seems to be that crops are expanding more rapidly than pasture. Past trends in the balance of pasture, croplands and native vegetation in Brazil continue in our baseline simulation. Pasture and crops undergo modest gains and, for the most part, these come at the expense of native vegetation. We simulate less land abandonment than the rate that has been empirically observed in the Amazon biome (National Institute of Space Studies of Brazil (INPE) & Brazilian Agricultural Research Corporation (EMBRAPA), 2011).

Environmental Impacts: Land Cover Change & GHGs Land use change GHG effects from beef are uncertain, however, relative to other forms of food, beef is associated with substantially higher direct environmental impacts than other food products. Methane emissions, nitrous oxide emissions from soils, and nitrate water pollution are all major problems from cattle systems (Steinfeld et al., 2006; Weber & Matthews, 2008; Garnett, 2009). Methane emissions aside, the bulk of the GHG effects are associated with land use changes that the literature attributes to cattle productions systems (Steinfeld, Gerber, Wassenaar, Castel, Rosales, et al., 2006; Wassenaar et al., 2007; Cederberg, Persson, Neovius, Molander, & Clift, 2010; Mercedes Bustamante et al., 2012). However, these studies employ techniques for attributing land use change to cattle ranching that use correlational relationships between cattle pasture and deforestation as proxies for causal patterns. Emissions from land use change immediately preceding cattle ranching is the source of the vast bulk of GHG

emissions that these studies assign to cattle systems. This approach does not address how much of this land use change would have occurred in absence of the cattle production. Our approach sidesteps this attribution problem by simulating the marginal emissions effect of policies. We do not make claims about the emissions associated with particular areas of cattle production that we simulate.

Intensification Policy Scenarios Framework We modeled two intensification policy mechanisms—a flat subsidy for all ranches adopting intensive alternatives and a flatly levied tax on all ranches that do not adopt intensive alternatives. We assume that profits are equal to zero so that price equals marginal cost. PG is the price of grass fodder per ton, c is the total production cost per hectare, y is the pasture productivity in tons per hectare, across pixel i and time period j. The cost of production of pasture per hectare encompasses the land price. This is endogenously computed and varies across locations. Some elements of the management cost are uniform across Brazil, but inputs costs vary according to the isolation from input production facilities.

(1) !"!" = !!"/!!" (2) !"#(!, !")

We differentiate between two pasture management system types: the first type varies across i,j and is one of the twelve traditional sector types represented in the model (System 1). Here pasture management is limited. The second type is a semi-intensive alternative (System 2). Here, modest improvements in breeding, feeding, and pasture productivity lead output to double per hectare over business as usual. We assume that System 2 has a higher management cost than System 1. The necessary condition for adoption of System 2 is that its marginal cost per unit output is less than the marginal cost in System 1. In equations (3) and (4) we derive the necessary condition for pasture intensification. The cost of production in the semi-intensive alternative sector must be at least two times higher than the production cost in the business as usual sector for there to be no adoption.

(3) !!

!!> !!

(4) !!

!!> !

(5) 2!1 > !2

In our modeling framework, the adoption of new production technologies is determined by spatio-temporal flux in land prices and agricultural prices. These in turn, depend on the land productivity potential and spatially-explicit costs of transporting inputs to agricultural regions and costs of transporting agricultural goods to markets.

Scenario Types We introduce different tax levels which increase the cost per hectare in the traditional sector or different subsidy levels that lower the cost per hectare in the intensive sector. From (3) we deduce that the subsidy scenario is equivalent to the tax scenario if it is twice as high as the tax level (9).

(6) !!!!!!

= !!

(7) ! = !!

!− !1

(8) !!

!!!!= !

(9) ! = !2− 2!1

(10) ! = 2!

Five different tax levels between 10 and 50 USD per hectare per year are introduced and 5 different subsidy levels between 20 and 100 USD per hectare per year are also implemented. This corresponds to a reduction in the production cost gap between the two pasture systems between 15 percent and 80 percent of the initial cost gap.

Scenario Results

Tax Scenarios: Brazilian Cattle Sector, Land Use & GHG Effects

Taxes increase the cost of production of pasture System 1. This leads to higher beef and milk prices in Brazil (Figure). While the internal demand is only marginally affected (-3 percent), higher prices decrease the competitiveness of Brazilian beef on international markets and exports are reduced by 10 percent with a tax of 10 USD per ha (Figure). However beef prices do not vary much between a 20 USD tax and a 50 USD tax because the higher tax leads to a higher adoption rate of management System 2 and thus boosts output. The share of the pasture area under System 2 increases from 6 percent with a 20 USD tax to 16 percent with 30 USD tax and to 42 percent with a 50 USD tax. The area of pasture decreases by less than 3 percent for a tax level higher than 20 USD. This means that in these cases the higher output of cattle products under System 2 partially offsets the decrease in grassland area. Thus, a higher tax has little effect on beef production. This remains at 12.2 million tons per year for any tax level higher than 20 USD per ha. The tax does, however, affect the distribution of the cattle production across Brazil. Cattle production increases in the South and Southeast regions at the expense of the North, North East and Center West regions due to higher adoption rates of the System 2. Adoption of System 2 in the North requires a rate of greater than 30USD per ha per year and in the Center West for a tax level higher than 40 USD/ha per year. In the case of the North, low land prices are to blame for the limiting adoption. In the case of the Center West the problem is more the limited increase in output from boosted production.

Figure 2-3. Policy-induced adoption of semi-intensive alternative pasture management systems. The top chart displays subsidy induced adoption, the bottom chart displays tax-induced adoption. Subsidy and tax levels are expressed in USD per hectare per year.

The imposition of taxes increases slightly the share of cattle systems reliant on crops for supplemental feed. Hog production also rises. Thus the increases in cultivated area under the tax go primarily to increased soybean and corn production for domestic animal feed. However, the area of these crops increases modestly—just 1 percent to 10 percent depending on the tax level. These increases are small relative to the area of pasture reduction under the taxes. Less than 1 percent of pasture reduced becomes cropland. The remaining 90 percent of avoided pasture expansion mean reduced conversion of Amazon forest and cerrado savannah. The area is quite substantial—between 8 million and 33 million

hectares of avoided land use conversion results from the tax policies. Of these reductions, the vast bulk is a reduction in the conversion of forested lands.

Figure 2-4. Change in Pasture Area Caused by Intensification Policies, By Region 2010-2030.

The tax scenarios reduce the GHG emissions from deforestation in Brazil by 30 percent -66 percent. Substantially smaller declines in GHG emissions are associated with the combined declines of 5-10 percent in methane emissions from enteric fermentation and Nitrous Oxide emissions from cattle manure. These decreases stem both from a decline in animals numbers and due to switching cattle systems with shorter animal lifespans and thus better ratios of meat to enteric fermentation and nitrous oxide. In sum, compared to the baseline, the total GHG emissions from land use change, crops and livestock sectors are reduced by 30 – 70 percent depending on the tax scenario.

Tax Scenarios: Global Trade, Land Use, & GHG Effects Since Brazil is the 2nd largest beef exporter in the world, the reduction in Brazilian beef production by the tax scenarios spills over abroad. 80 percent of the drop in Brazilian beef production under the tax scenarios comes from reductions in beef exports. The remaining 20 percent is explained by reduced consumption in Brazil. Of decline in exports, half comes from reduced beef consumption under higher

beef prices and half results in increased production in other exporting and importing nations. The production increases especially in South East Asia and in the other Latin American countries (Figure). As a result, GHG emissions from land use change increase between 1.6 and 3.6 percent abroad and GHG emissions from livestock production increase between 0.3 and 0.4 percent abroad. However, the GHG emissions savings in Brazil more than offset the GHG emissions increase in the rest of the world so that the global GHG emissions from land use and agriculture are reduced between 2 and 7 percent with the implementation of the intensification tax scenarios.

Figure 2-5. Beef Production Displaced to Other Model Regions Under Tax Scenarios.

Subsidy Scenarios: Brazilian Cattle Sector, Land Use & GHG Effects Subsidies reduce the cost of production in System 2, the semi-intensive alternative cattle system. Just 2 percent of total pasture area sees adoption with a 20 USD/year subsidy. This reaches 17 percent of the total pasture area with a 60 USD/year subsidy and increases to 85 percent of the total area with a 100 USD/year subsidy. In 20 and 40 USD subsidy scenarios, pasture area declines, but cattle production remains similar to the no policy baseline. At the 60 USD subsidy level, beef price declines substantially (-5 percent). At 60 USD, beef production also increases by 6 percent (reaching a level of 14 million tons). The increase is split evenly between higher internal consumption (+ 3 percent) and higher exports (+10 percent). With a subsidy level of 100 USD/ha/yr, export increase 30 percent (Figure). As in the tax scenarios, the transition to more intensive pasture management is heavily concentrated in the South and the South East regions under smaller subsidy levels.

However, we observe that the transition occurs in a much higher share of the pasture in the North, North East and Center West regions in the high subsidy scenarios than in the high taxes scenarios (Figure).

Figure 2-6. Beef Production Displaced in Other Model Regions Under Subsidy Scenarios.

The total pasture area decreases with the introduction of the subsidies but it decreases less than in the tax scenarios (between 1 percent and 14 percent compared to the baseline). The reduction in pasture expansion is 7 million and 10 million ha lower than the tax scenarios of equivalent production cost gap reduction. The explanation is that cheaper beef leads to an increase in beef consumption and a decrease in the poultry meat and pork consumption. The total cropland area increases less than 1 percent across the different subsidy scenarios relative to the baseline simulation.

Figure 2-7. GHG Impacts of Tax and Subsidy Policies. Each column equals A minus B where A equals the sum of global emissions from crops, livestock, and land use change 2010-2030 under the policy scenario and B equals the sum of global emissions from crops, livestock, and land use change 2010-2030 in a baseline simulation.

The GHG emissions from deforestation are reduced by 12 percent with a 20 USD subsidy per ha and by 60 percent with a 100 USD subsidy in 2030. These correspond to a decline in land use change of 2 million hectares to 25 million hectares respectively. Of this change, the vast bulk is avoided deforestation in the Brazilian Amazon Biome. However, direct emissions from the ranching sector remain constant for subsidies below 60 USD/ha/yr and increase up to 16 percent. GHG savings in Brazil are between 2 and 3 times lower than the ones that are achieved with production cost gap equivalent tax policies.

Figure 2-8. Evolution of Beef and Milk Prices in Brazil Under Intensification Policies.

Subsidy Scenarios: Global Trade, Land Use, & GHG Effects The adjustments in domestic demand explain between 28-44 percent of the Brazilian domestic beef production increase to 2030. Beef demand abroad increases and Brazilian beef not only grows to satisfy the new demand, but it also displace other beef on the world market due to competitive prices. Production losses are scattered across Australia, Europe, Sub-Saharan Africa and South East Asia. GHG emissions from conversion of natural land decrease between 0.3 percent and 4.2 percent and GHG emissions from livestock production decrease between 0.1 and 0.4 percent in these nations where cattle production declines. Under the subsidy scenarios, GHG emissions are reduced both in Brazil and in the Rest of the World. Compared to other food products, demand for beef is relatively price elastic—when beef prices fall, the consumption of beef increases as consumers buy more beef instead of other protein sources. On average, a shift away from any other food product to beef would lead to increased global land in crops and pasture because

land requirements to produce beef are much higher than those for any other food product. However, because our subsidies explicitly target land efficient beef, the boosted land requirements associated with the shift from non-beef to beef is swamped by avoided deforestation associated with the replacement of land inefficient beef with land efficient beef.

Social Cost of GHG Mitigation

Figure 2-9. Global social cost of GHG mitigation from pasture intensification scenarios in Brazil (in USD/tCO2e). Where: (1) GHGs are computed as Ai minus B where A equals the sum of global emissions from crops, livestock, and land use change 2010-2030 under the policy scenario i and B equals the sum of global emissions from crops, livestock, and land use change 2010-2030 in a baseline simulation. (2) Social cost equals Ci – D where Ci equals(Consumer Surplus and Producer Surplus + State surplus) under policy scenario i and D equals ( Consumer Surplus and Producer Surplus + State surplus) in a baseline simulation. The state surplus is equal to the sum of the taxes (gives a positive surplus) or the sum of the subsidies (gives a negative surplus) in Brazil.

For the lower range of emissions savings levels65, the subsidies have lower social welfare costs than the taxes. An objective of 300 MtCO2e savings per year could be reached by a subsidy level of 100USD or a 50USD tax over the period 2010-2020. If the objective of the policy is to reach the highest CO2e savings, high subsidy levels seems the best strategy in the short run but they could become quite expensive in the long run without leading to additional CO2 savings. Between the period 2010-2020 and the period 2020-2030, we observe a shift of the marginal 65 lower than 300 MtCO2e per year

0 2 4 6 8 10 12 14 16

0 50 100 150 200 250 300 350 400 450

USD/tCO2e

Global GHG savings in MtCO2e/year

Tax2020 Tax2030 Subsidy2020 Subsidy2030

abatement curve to the right. This means that a certain level of CO2 emissions savings could be reached with a lower tax or a lower subsidy level. This suggests that the optimal policy should evolve over time. Taxes or subsidies should be high in the early portion of the modeled period, but they should progressively decrease and disappear. If a high subsidy persists, it could lead to a rebound effect by providing an incentive to increase pasture expansion.

Discussion

Feasibility and effectiveness of unilateral climate policies The GHG mitigation potential we simulate is similar in magnitude to mitigation found in previous land sparing studies, but under the tax scenarios it differs in mechanism and under the subsidy scenarios it differs in both geography and mechanism. These discrepancies are due primarily to our use of market-mediated mechanisms for land use change including international leakage. Our tax scenarios find international leakage largely concordant with previous studies. We find mitigation from unilateral taxation, but we also find that this effect is dampened by leakage associated with a concomitant rise in production activities offshore (Gan & McCarl, 2007). Here an accounting of the GHG effects focused exclusively within Brazil’s borders would overestimate mitigation. Nevertheless, the policy would also deliver global GHG mitigation. International leakage would be less than domestic mitigation. In this way, our tax scenarios tell a familiar story of the type sparked by research into the “common but differentiated responsibilities” approach of the Kyoto Protocol (Stavins, 1997; K. M. Chomitz, 2002; Aukland, Costa, & Brown, 2003; Sohngen & Brown, 2004; Babiker, 2005; Gan & McCarl, 2007). Unilateral climate policies may create competitive disadvantage in countries and dampening mitigation effects of these unilateral activities due to international leakage. The subsidy scenarios, meanwhile, deliver a sort of reverse leakage. By finding climate benefits where previous research has primarily found climate costs, our subsidy scenarios cover seldom theoretically traversed terrain and have potentially mixed prospects for political feasibility. The subsidies deliver larger GHG reductions than their tax scenarios counterparts. They induce reverse leakage where subsidized Brazilian low-carbon cattle products are replacing higher carbon cattle products that would have been produced offshore. The potential for industrial subsidies to reduce GHGs is a potentially politically alluring alternative

to the idea that GHG mitigation would inevitably be costly to firms and governments. Perhaps, these results could be viewed as promising for the potential of the Green Climate Fund and similar initiatives to deliver mitigation through industrial subsidy. However, because these subsidies we model intentionally induce boosted agricultural output, they unambiguously classify as the sort of agricultural subsidies least likely to be allowed in an Agreement on Agriculture in the World Trade Organization. Such subsidies could also spark compensatory or retaliatory measures. Such measures could attenuate GHG gains somewhat given that these gains are in part a function of the retiring of GHG intensive production systems in the rest of the world. Despite the recent emergence in the United Nations Framework Convention on Climate Change of hortatory language for nations to address international drivers of deforestation (United Nations Framework Convention on Climate Change, 2011), international leakage still lacks firm accounting or governance mechanisms. Such mechanisms would need to monitor trade flows to which the climate effects of the policy would be sensitive. Exports of subsidized Brazilian cattle products would only deliver climate benefits where they displace higher carbon cattle products. However, it is possible that such exports could replace lower carbon cattle products and cause emissions increases. Finally, the closest analogues—biofuels policies that penalize international land use change—remain methodologically unsettled, politically controversial, and are facing legal challenges in some instances (Farber, 2011; Khanna & Crago, 2012).

Technology Adoption Patterns are Central to Intensification Efficacy Our findings demonstrate the sensitivity of technology adoption to land productivity and market isolation, but there are numerous other factors that are likely to influence technology adoption patterns. Salient factors include energy prices, landholder heterogeneity, land tenure status, and non-agricultural motivations for agricultural land use. Energy prices will have a strong influence on the size of the production cost gap between lower input and higher input management systems (Chakravorty, Hubert, & Nostbakken, 2009). Factors such as educational level, age, access to inputs, access to credit, and farm size have all also been shown to mediate the technology adoption process in agricultural systems (Just & Zilberman, 1983; Jack, 2009). Land tenure status is a strong determinant of land user behavior. In Brazil, as in much of the tropics, land tenure remains a highly uncertain affair with context-contingent implications for agricultural technology adoption patterns and thus the efficacy of terrestrial climate mitigation efforts (Robinson, Holland, & Naughton-Treves, 2011).

Global economic models of land use do not explicitly represent non-agricultural motivations for agricultural land use. These factors could attenuate the efficacy of the tax and subsidy instruments of inducing adoption of intensive alternatives. Thus it would be wise to consider the adoption rates we model as an upper bound of the efficacy of the proposed policy. All across the agricultural landscape, factors such as risk aversion, cultural preferences, supply chain structure, and imperfect information can dampen adoption of intensive technologies. Even on the large farms of the U.S. Midwest, higher commodity prices do not always boost yields according to model estimates (Hertel, 2011). These confounding effects are most pronounced at extensive agricultural margins, where most Brazilian cattle ranching area can be found. Brazilian cattle ranchers do not just hold their ranches to produce cattle products. They also have non-cattle-product-producing motives for ownership that play important roles in management decisions (Jarvis, 1974; S. Hecht, 1993; Barreto, Arima, & Brito, 2006; M. S. Bowman, et al., 2011). Maintaining cattle on Brazilian pasture can help to secure tenure, obtain access to government subsidies, to hedge against inflation, and to speculate on increases in land values. Cattle ranching may also be explained by cultural reasons that are not utility maximizing (M. S. Bowman, et al., 2011). By underestimating these non-agricultural motivations, the model may overestimate the likelihood of technology adoption in response to subsidies and taxes (M. S. Bowman, et al., 2011). In such cases, alternative landholding activities may be more attractive options for landholders to allocate their land, labor, and capital than adoption of intensive alternatives. On farms where instruments are applied, attractive alternatives could mean a negative opportunity cost for intensification technologies that appear profitable relative to business as usual cattle ranching where the only output is cattle products. As a result, agents might prefer to allocate land, labor and capital to alternative activities besides intensification even if intensification is profitable. Where adoption occurs, leads to intensification, and as a result drives down prices, the land sparing effect at the extensive margin could also be dampened by non-agricultural motivations for cattle ranching. Non-agricultural activities could keep a greater proportion of ranching profitable under revenue losses from cattle product price declines. This could limit the land sparing effect.

Better Governance Needed for Effective Land Sparing If they are to be effective, instruments of the sort we model will require more research and data gathering on the structure and function of Brazilian cattle systems and continued improvement in land use governance institutions in Brazil and beyond. Specific tasks include furthering agricultural data collection and monitoring, improving land tenure institutions, and developing adaptive governance rooted in sound regulatory science. Improved land tenure institutions are foundational to advances in Brazilian land use governance including the implementation of effective intensification instruments. In recent years, Brazilian land tenure institutions have streamlined and the introduction of geo-spatial tools has greatly improved the uptake and transparency of the tenure regularization. Nevertheless, overall rates of participation in these fast-track programs are low, competing/conflicting tenure claims persist, and the tenure granting process itself may still in some instances induce increased land clearing. This would stymie the pre-requisite for adaptive governance in the early stages of such a policy. Brazil has relatively advanced agriculture statistics capacity, but several missing elements, if not remedied, will undermine efforts to monitor and evaluate the effectiveness of the policy instruments like the ones we model. The most glaring absence is the infrequent assessment of pasture quantity and quality by census or by remote sensing. At present, pasture area is assessed just once a decade. Without more frequent updates it is not possible to discern key parameters of the land use change process. Consequentially, it would not be possible to monitor and evaluate the effects of intensification instruments. Our focus on adaptive governance rooted in monitoring and evaluation stems from the aforementioned data gaps and also scientific uncertainty of Brazilian cattle systems. Efficacy is likely to require trial and error. Since the baseline farm characteristics and trends are typically unknown, it is crucial to also monitor systems of production in a control group of ranches. Doing this well would require a surge in data, enforcement, and regulatory science. One of the greatest and likely costliest challenges stems from the inherently unobservable nature of the GHG outcomes of intensification instruments. The magnitude and even the direction of the GHG impacts of intensification instruments will not be reliably indicated by superficial production system characteristics such as absolute or relative intensity. Instead, it will be necessary to mechanistically monitor production systems to model the multiple mechanisms by which cattle systems shape the global land use process. Such modeling is necessary to conduct because the global land use

process has no replicates and thus the effects of the policies cannot be meaningfully observed. Additionality66 has long been a controversial pre-requisite for land use climate mitigation strategies (Humphreys, 2008). The intensification technologies we model are already quite common across the Brazilian cattle sector and their use is already growing. The policies that we have devised are designed to accelerate the intensification process on laggard ranches simply to approximate the behavior of leading ranches. Therefore, the presence of the technologies on treated ranches would be insufficient evidence of a policy effect. To ensure policies are accelerating adoption, it will be necessary to use pilot and trial programs to better discern how the policies influence land users. To do so, it would be necessary to monitor both treated ranches and a control group of ranches. Such an approach would require an articulating and updating experimental design and a strong backbone of benchmarks. In the cattle sector, determining productivity benchmarks and monitoring intensification activities would require substantially improved agricultural earth observation and monitoring of the movement of cattle. The distinction between additional and non-additional intensification is important for informing and refining the intervention mechanism, but it should not necessarily be the approach for the policy itself. Penalizing early adopters of intensive technologies could slow the rate of endogenous adoption and thereby attenuate policy gains. Thus the trial and pilot periods should be used to explore the costs and benefits of need-blind payments to ranchers. The prospects of an inframarginal subsidy might break with convention of land climate mitigation, but it could boost policy effectiveness (O'Hare & Mundel, 1983).

Conclusions Policies to boost cattle ranching productivity in Brazil have the potential to provide a non-trivial level of GHG benefits. We show that they could mitigate as much as two thirds of Brazil’s terrestrial GHG emissions to 2030. To achieve these goals, however, such policies would likely require substantial improvements in systems of land tenure and in the rule of law along the agricultural frontier of Brazil. And even then, at their best, cattle ranching intensification policies in Brazil cannot achieve the emissions reductions delineated in the Brazil National Climate Action Plan. Therefore, such policies are best understood as a necessary, but insufficient strategy for GHG mitigation in Brazil. They are best complemented by a number of efforts

66 That is, that the impacts of the intervention must be able to be proven to be the result of the policy and not something that would have occurred anyway.

to target illegal deforestation directly. Such efforts have the added benefit of reducing the possibility of local leakage associated with the policies. Whereas a number of these other efforts can be described as command and control, both cattle ranching intensification approaches depicting in this research would be predicated on the behavior of hundreds of thousands if not millions of rural landholders across the country. Such an effort is audacious, but it also reflects the reality of the degree to which these landholders will influence future landcover and GHG outcomes in Brazil and beyond.

Methods

GLOBIOM Overview This analysis uses GLOBIOM to simulate the GHG and economic impacts of introducing intensification technologies and a variety of policy alternatives for encouraging intensification technologies in the cattle ranching sector of Brazil. We examine the effects of these technologies and policies on the Brazilian and ROW land use economies and terrestrial GHG emissions. To conduct our policy analysis, we employed a specially modified version of the 28-region version of the GLOBIOM. GLOBIOM is an economic partial equilibrium model of the global land use economy that depicts the competition for land between the forest, agriculture, and livestock sectors. Demand for food and wood is determined by exogenous population and GDP per capita projections and by projections of dietary patterns and trends (Alexandratos, 1999). Equilibrium prices are the result of a simulation to maximize producer and consumer and consumer surplus. The maximization problem is subject to resource, technological, and political constraints. Prices between regions vary according to the transport costs between regions and trade barriers. GLOBIOM has mostly constant elasticities, can be solved with linear programming, can be arrayed in either 11 or 27 regions, and can be run over a range of spatial resolutions in order reduce computational time. Production types are detailed, geographical and Leontief type (i.e. fixed input and output ratios). However, discrete changes in the technological characteristics of primary product production can occur because multiple production types can be specified in the model. In cattle systems the model is initialized to contain twelve different production types with varying inputs, outputs and feed, land efficiencies, and GHG emissions.

Demand Constraints Incremental demand for primary products elicits intensification and or extensification. Some of this extensification requires land use/cover change. In the model these changes have geographically explicit costs and these are included in the objective function.

Input Data GLOBIOM directly represents production from four major land cover types, cropland, grassland, managed forest and areas suitable for short rotation tree plantations, by implicit product supply functions based on Leontief production functions. The input data on the supply side of the model are structured around a detailed spatial resolution between 0.1 and 0.5 arcmin pixels (Skalsky, 2008). Global maps on soil types, climate, topography, land cover, crop area and management, and livestock production and management have been collected (GLC2000; You and Wood, 2006; Wint and Robinson, 2007). This information is used by the biophysical models, EPIC for crops (Izaurralde, Williams, McGill, Rosenberg, & Jakas, 2006) and G4M for forests (Kindermann, McCallum, Fritz, & Obersteiner, 2008) to assess the production potential for each pixel. Different livestock production systems for 5 different animal species have been designed based on ILRI/FAO nomenclature and populated with data using process-based models for ruminants, and using literature review and expert knowledge for the monogastrics (Sere & Steinfeld, 1996; Herrero, Thornton, Kruska, & Reid, 2008). Currently, 18 crops, 5 forestry products and 6 livestock products (4 types of meat, eggs and milk) are included in the model. Further description of GLOBIOM can be found in Havlik et al. (2011) and at the website globiom.org.

Brazil Modifications For this analysis, we modified the Brazil region to provide improved representation of agricultural logistics costs, pasture intensification pathways, and grassland productivity potential.

Logistics costs We developed the Brazilian Agricultural Transport Costs Model (BATCM), to incorporate into the Brazil region of GLOBIOM spatially explicit estimates of agricultural logistics costs. We incorporate these logistics costs into the production function of the model. This helps to align modeled land use with observed patterns of land use. Following from the approach used in “Travel Time to Major Cities,”

BATCM uses the Cost-Distance function in ArcMap Spatial Analyst to find the least cost path to transport inputs to rural properties and to transport agricultural products to market (Nelson, 2008). For each of the 18 agricultural crops and five animal products depicted in GLOBIOM we have developed maps of processing facilities and input producers. For each crop type we have also developed friction surfaces, maps that assigns a cost per ton kilometer traveled across the Brazilian logistics network and land surface. Cost-distance identifies the least cost path across the latter map to the infrastructure in the former map. The output is a series of maps depicting crop-specific input and output logistics costs such that Logistics Costs in USD per ton =L=f (Fi,j,k) where F is transport friction measured in USD per ton-kilometer over the vector i=the land cover type j=production systems k=input or outputs. Further detail of these activities is described in the supporting online material.

Pasture Intensification Pathways The intensification technology we introduce is improved pasture yield (in digestible biomass per hectare per year). At present we have characterized these yield improvements as a function of incremental fertilizer inputs. We selected a declining exponential curve to depict biomass response to fertilizer application. The fertilizer response curve is geographically explicit. It depends on a baseline yield surface that in turn depends on local climate and soils67. The fertilizer cost response curve is shaped differently though its shape is also geographically explicit. Further detail can be found in the supporting online materials.

Grassland Productivity We modified GLOBIOM to incorporate cattle pasture productivity base data consonants with agricultural statistical data and the scientific literature on pasture productivity estimates. First we reviewed the scientific and grey literatures on pasture productivity estimates for each of the six Brazilian biomes.68 We then identified a high and a low bound of metric tons of dry matter yielded annually in the grasslands of each biome.69 Next, we calculated a weighted average of total modeled grassland productivity at the biome level for each of the eleven pasture

67 Note that the asymptotic or maximum achievable yield is assumed to be 2.5*Y0. This is approximately consistent with yield maxima relative to baseline yields as is modeled in the Global Agroecological Zoning model. Ideally, this scalar would also depend on climatic and edaphic conditions. 68 The references at the end of this document comprise the literature we reviewed 69 Estimates were frequently given with just one significant figure and also frequently not supported by citations from scientific literature.

productivity models. Next we calculated the minimum deviation between modeled and expert biome averages. Then we summed the deviations from all biomes and selected the minimum deviation estimate for each biome. Further detail of these activities is described in the supporting online material.

Intensification Costs We introduce seven different intensification policy scenarios. Si, Ti, and NP. Where S is for subsidy; T is for tax, NP is for no policy, i=75 percent, 50 percent, and 25 percent. Under S_75 75 percent of unit (per hectare) grassland intensification cost is paid as subsidy to ranchers who convert to intensive pasture. The subsidy scenarios shock the model system by introducing payments to Brazilian cattle ranchers of up to 25 percent, 50 percent, and 75 percent of the average per hectare cost of converting conventional pasture systems to intensive, pasture-based cattle systems. The tax scenario levies a flat tax on all ranchers who do not adopt intensive practices. The tax rate per hectare is the equal to 50 percent of the average intensification cost per hectare. Intensification cost is =CA+F*(CT+CF), where CI=annualized intensification cost/ha, CA=area cost/ha, F=fertilizer/ha, CT=fertilizer transport cost per unit fertilizer, CF=cost of fertilizer purchase (fertilizer price)

GHGs For each scenario we account the cumulative GHG difference from the baseline scenario during the period 2010-2030. We compute global emissions effects and also the subset of these effects that occur within Brazil. The global GHG impact of the land use intensification mandate in each year is given for year i as Gi and total global emissions is:

! = !!

!"#"

!!!"""

Where: Gi= Gi,j - Gi,0 and Gi,j equals the global anthropogenic greenhouse gas emissions associated with scenario j in year i and Gi,0 equals global anthropogenic greenhouse gas emissions associated with the baseline scenario in year i. Additionally, Gij=Pij+Qij+Rij where Pij=total emissions from land use change,

Qij=total direct emissions from the agricultural sector, Rij=emissions from the production and transport of agricultural outputs.

Chapter Three: The viability of cattle ranching intensification in Brazil as a strategy to spare land and mitigate greenhouse gas emissions 70

Introduction The potential for interventions to reduce the pressure of agriculture on forests has become central to debates over the future of biofuels and has become an explicit focus of negotiations on Reducing Emissions for Deforestation and Forest Degradation (REDD). One proposal is land sparing, the concept of boosting outputs from agricultural lands and/or steering agricultural expansion onto low carbon content lands to make room for forests and other productive uses. This paper examines the viability of one intervention that intends land sparing—cattle ranching intensification programs (hereafter CRIPs) in Brazil. We define CRIPs as interventions for reducing GHGs by increasing the quantity of cattle product output per unit of pastureland. Momentum for Brazilian CRIPs may have originated with livestock researchers (Serrão & Toledo, 1990; E. Y. Arima & Uhl, 1997), but it now comes from a variety of sources including NGOs, governments and the scientific community (Renewable Fuels Agency, 2008; Sajwaj, et al., 2008; Ecofys & Winrock International, 2009; Manzatto, et al., 2009; Nepstad, et al., 2009; Tim Searchinger & Amaral, 2009; Angelsen, 2010; Embassy of Brazil, 2010; Gouvello, 2010; Lapola, et al., 2010; Amend, et al., 2011; Boucher, et al., 2011; Pinjuv, 2011). It is reflected in Brazil’s Nationally Appropriate Mitigation Actions (NAMAs), which, up until now are the best hint of how climate mitigation policies in Brazil might develop (Embassy (Embassy of Brazil, 2010).71

70 This chapter was prepared in collaboration with Maria Bowman, David Zilberman, and Kate O’Neill. An earlier version is published as a working paper by Climate Change, Agriculture, & Food Security (CCAFS). The citations for the working paper is: Cohn, A., Bowman, M., Zilberman, D., & O'Neill, K. (2011). The viability of cattle ranching intensification in Brazil as a strategy to spare land and mitigate greenhouse gas emissions. Copenhagen, Denmark: CCAFS. 71 Once resolved to mitigate, Brazil, like many Southern countries, has had little option but to address emissions from land use land use change and forestry (LULUCF) including agricultural drivers of deforestation. LULUCF emissions form the vast bulk of all anthropogenic GHG emissions from Brazil and addressing them may be harmonious with the momentum for UN-REDD, and with

As part of eleven billion tons of potential GHG mitigation between 2010 and 2030, Brazil’s NAMAs pledge roughly one billion tons of annual climate mitigation during the year 2020.72 The NAMAs commit just 10 percent of total mitigation to come directly from changes in cattle ranching practices. However, we estimate that roughly 90 percent of the mitigation proposed would come from reduced land use change and changes in output and production practices in agricultural systems that could hinge on increased cattle ranching productivity.73 As the report detailing the NAMAs describes it, “Increasing …intensification of livestock-raising can play an essential role in reducing the need for land… while releasing the land required for expansion of other activities” (Gouvello, 2010:28). The ranching focus of Brazil’s NAMAs may emerge from a typical, but questionably accurate model representation of cattle ranching systems in a World Bank report closely linked to NAMAs.74 To depict GHG-reducing CRIPs, the World Bank team

the multi-billion dollar Amazon Fund to address longstanding international pressure for Brazil to end deforestation. 72 The NAMAs letter details reduction potential in the year 2020 alone, but our analysis suggests that the pledge is a cross section of a mitigation path for the period 2010-2030 that corresponds with Brazil’s 2010 Climate Law. In subsequent paragraphs we describe this mitigation path as detailed in a World Bank report closely related to the NAMAs pledge - Gouvello, C. (2010). Brazil Low Carbon Country Case Study. Washington D.C.: World Bank: Sustainable Development Department of the Latin America and Caribbean Region. 73 The NAMAs letter does not explicitly state which mitigation actions CRIPs facilitate, but the NAMAs targets are nearly identical to targets published in Gouvello (2010), a report clearly stating the central role of land sparing for facilitating most GHG mitigation it proposes. The following passage is the best summary: “To avoid emissions from deforestation, ways would need to be found to reduce global demand for land, while maintaining the same level of products supply as in the reference scenario. In systemic terms, the mitigation of emissions through land-use change could be achieved by absorbing the expansion of these activities via the increased productivity of other ones. Brazil’s major [crop] agricultural activities already show high levels of productivity and consequently do not offer opportunities to increase productivity on the scale required to absorb these additional levels of demand for land. Beef-cattle farming shows much greater potential for increasing productivity per hectare, which can be applied to a much larger pasture area, since pastures occupy 207 million ha compared to 70 million ha for agricultural activities in 2030 in the reference scenario. Consequently, increasing the technological level and the intensification of livestock-raising can play an essential role in reducing the need for land for this activity, while releasing the land required for expansion of other activities.” 74 The Gouvello team’s findings for cattle ranching mitigation potential are consistent with other investigations of climate mitigation in Brazil. See for example: Eliasch, J. (2008). Climate change: Financing global forests: the Eliasch review: Earthscan/James & James. McKinsey & Co. (2009). Pathways to a Low Carbon Economy for Brazil. Retrieved August 24th, 2011, from

imposes a constraint on the total area of land available for use as pasture in Brazil and caps total Brazilian livestock and agricultural output at reference levels. By imposing the land constraint, the team “frees” land in the simulation for forests and crops. Together, the two constraints cause the simulation of substantially lower GHG emissions in the low carbon scenario than in the reference scenario. The simulation demonstrates the greenhouse gas mitigation potential of cattle ranching intensification in Brazil under these model constraints, but it is unclear how to design interventions to elicit analogous effects in practice. This paper builds and employs an analytic framework to identify essential elements of CRIPs effective at delivering GHG benefits. The paper begins with a section defining key concepts foundational to our analysis. Then we present an analytic framework suitable to analyze the GHG mitigation potential of CRIPs and other “land sparing” efforts. The framework draws on research from across a wide variety of fields including resource economics, lifecycle analysis, rural sociology, and land change science. We then present and examine a list of seven necessary elements for CRIPs to serve as GHG mitigation strategies. We close by recommending research and regulatory activities that should begin now to support the design of effective GHG mitigation from CRIPs. Table 3-1. Brazil’s Proposed NAMAs: Pledged Emissions Reductions for the year 2020

Mitigation potential (MT CO2)

Percent of total mitigation

Restoration of grazing land 83-104 9 percent-11

percent

Integrated crop livestock systems 18-22 2 percent

Total ranching targeted 101-126 10 percent-12

percent

Reduction in 564 54 percent-58

http://www.mckinsey.com/en/Client_Service/Sustainability/Latest_thinking/~/media/McKinsey/dotcom/client_service/Sustainability/cost%20curve%20PDFs/pathways_low_carbon_economy_brazil.ashx Ecofys & Winrock International. (2009). Mitigating Indirect Impacts of Biofuels Production: Case Studies & Methodologies. Retrieved August 24th, 2011, from http://webarchive.nationalarchives.gov.uk/20110407094507/http://www.renewablefuelsagency.gov.uk/sites/rfa/files/_documents/Avoiding_indirect_land-use_change_-_Ecofys_for_RFA.pdf

Amazon deforestation

percent

Reduction in cerrado deforestation 104 10 percent-11

percent

No-till farming 16-20 2 percent

Biological N2O fixation 16-20 2 percent

Biofuels use 48-60 5 percent-6 percent

Total ranching related 748-768 74 percent-77

percent

All Ranching Contingent 849-894 85 percent-88

percent

Energy Efficiency 12-15 1 percent

Hydroelectric power production 79-99 8 percent-10

percent

Other alternative energy 26-33 3 percent

Total non-ranching related 117-147 13 percent-15

percent

Grand total 966-1041 100 percent

Source: Embassy of Brazil, 2010; Gouvello, 2010

Definitions

Intensification Cattle ranching intensification is a subset of land use intensification and agricultural intensification. Neither land use intensification nor agricultural intensification are wholly subsets of the conventional notion of industrial intensification employed in production economics. Industrial intensification is the use of more of any production input relative to other inputs per a given quantity of output. By contrast, land use intensification generally refers to changes in agricultural production

practices that lead to more agricultural outputs per area of land input. This most commonly means boosting non-land inputs in ways that boost output. It could also mean an increase in factor productivity through the adoption of new, more efficient technologies. Alternatively, because land quality can vary, agricultural intensification can also mean boosting the quality of the land input by changing the land area farmed through acquisition, sale or swapping.75 The former form of land use intensification—changed production practices—is the version of intensification typically envisioned by proponents of CRIPs for land sparing. In the case of cattle ranching systems, intensification can be used to mean anything from a slight increase in intensity of extensive pasture systems to a switch to feedlots from open grazing. Note that in cases of production systems involving supplemental feed, the land used to produce the supplemental feed is not always accounted in the intensity metric.

Land Sparing The concept of land sparing is based on the theory that aggregate increases in agricultural yields over time can reduce the overall area of agriculture lands from what would have been needed without the increase in yields. These increases could occur through either the use of degraded, marginal, and abandoned lands or through increases in yields on lands currently in cultivation. The land sparing concept has been applied to scales ranging from local to global. The earliest discussion of land sparing centered on the potential for intensification to alleviate and prevent hunger, but the discussion has expanded to also address indirect environmental externalities including deforestation, biodiversity conservation, the provision of ecosystem services, (Green, et al., 2005; P. A. Matson & Vitousek, 2006; Fischer et al., 2008), and more recently GHG emissions (Rudel et al., 2009; Angelsen, 2010; Burney, Davis, & Lobell, 2010; Gouvello, 2010). 75 The role agricultural land quality in the expression of agricultural inputs and outputs is actually quite complex and variant. Some analyses consider the financial value of inputs and outputs to be the functional unit for accounting production; other studies use physical quantities and others use hybrid and/or inferred measures. See Hubacek, K., & Van Den Bergh, J. (2006). Changing concepts of land in economic theory: From single to multi-disciplinary approaches. Ecological Economics, 56(1), 5-27. Thus, an increase in the value of the land input could actually be associated with a decrease in the area of the land input under appreciating land prices or a shift towards higher quality land. Land quality can vary without the quantity of the land input varying if the quantity of the land input is expressed in area or some other metric independent of quality. When land price is the metric for land inputs then the quantity of land inputs would vary with both land quantity and land quality. Note also that changes to land quality can result from excessive or insufficient use of other non-land inputs.

Cattle Ranching Intensification Programs Few CRIPs exist yet, but we define them as any sort of intervention with the intent of reducing GHGs by increasing the intensity of cattle product output per unit pastureland. This can include early stage efforts to trial or pilot solutions. CRIPs comprise direct interventions in the cattle sector like credit, input taxes and subsidies, research & development, and output taxes and subsidies. They also comprise indirect interventions like agricultural infrastructure construction and planning, and even trade and fiscal policies. CRIPs could take many forms. For example, CRIPs might work by rewarding intensiveness or by penalizing for extensiveness. By definition CRIPs seek to alter the intensification trend. In period zero each ranch might be of average, above average, or below average productivity. In addition, each might be trending flat, intensifying, or extensifying over the period prior to the CRIPs. Note that productivity can be compared to national averages, and/or to regional averages. In this way, benchmark intensification could be absolute or relative to local biophysical and economic conditions.

Figure 3-1. Essential elements for cattle ranching intensification programs to mitigate GHGs.Interventions that accelerate the adoption of higher yield production practices are a necessary but insufficient condition to ensure desirable GHG mitigation. Boosted output must reduce producer prices such that the profitable area of cattle ranching shrinks. This then must lead to reduced deforestation and freed land for other productive uses. In addition, direct emissions increases must not offset the GHG benefits from land sparing. Finally, the benefits from the mitigation must exceed the loss to social welfare through intended and unintended consequences of the intervention.

The Essential Elements for CRIPs to Reduce GHGs For CRIPs to deliver GHG benefits, they must trigger a multi-step process more complex than some literature suggests. The Gouvello (2010, p. 27) report, for example, explains the CRIPs concept this way:

[CRIPs] reduce global demand for land, while maintaining the same level of products supply as in the reference scenario. In systemic terms, the mitigation of emissions through land-use change could be achieved by absorbing the expansion of these activities via the increased productivity of other ones.

This seems to suggest that the level of GHG mitigation is given by the following equation:

GHG mitigation = !"! !"#$ × [!"#$! − !"#$! × !"#$%! !"#$%! ]

We argue that the above equation does not necessarily hold because CRIPs-triggered yield increases do not ensure GHG benefits. CRIPs require a series of intermediate steps to ensure that the yield increases they cause lead to GHG mitigation. Figure 3-1 depicts what we consider the essential elements for one example CRIPs pathway to lead to reduced GHG emissions. Herein CRIPs interventions trigger increased higher intensity cattle product output.76 If land sparing is to reduce GHG emissions then it must consist of a reduction in land use area relative to business as usual land use area for a given level of production. This must come from increased productivity. By definition the “increased productivity” must be sparked by some intervention without which it would not have otherwise occurred. It cannot have been increased productivity that would have happened anyway. Next there is the matter of “absorbing the expansion”. This requires that a market conveys the lowered prices caused by the increased productivity production systems to the business as usual production systems. Thus there must be a shared market for the products of the intensive and extensive systems. Therein the increased productivity systems must outcompete the business as usual systems on cost and thus reduce overall prices. These price are intended to lead to a reduction in the demand for cattle ranching land by reducing the area of land over which ranching is profitable. This reduction, then “spares” land for other uses including the conservation of natural vegetation and crop agriculture. A subset of this crop agriculture could then be biofuels that under some circ*mstances could deliver further GHG benefits via displacement of fossil fuel. The shift from intensive to

76 By definition CRIPs work to increase the aggregate intensity of cattle ranching systems. This can occur via command and control or market intervention. If market interventions, CRIPs would entail increasing the competiveness of intensively produced cattle products relative to extensively produced cattle products. This could occur either by interventions to cheapen intensive production systems or interventions to make extensive systems more expensive. The former model has been much more frequently discussed and so it is the concept on which we focus our analysis.

extensive ranching should also contribute to reduction in the direct GHGs associated with the cattle sector. The direct and indirect GHG benefit together would then need to exceed GHG costs of these changes, including potential soil carbon flux from LUC caused and potential increases in direct agricultural and land use change emissions due to increased cattle product consumption under lower cattle product prices. Finally, these net CRIPs GHG benefits would need to be worth the implementation costs. In this section we explore the components and the plausibility of this CRIPs pathway by examining seven of its essential elements: (1) Intensive Ranching Technologies Must Be Able to Scale Up, (2) Inducing Adoption of Intensification Must be Straightforward, (3) Increasing Intensive Ranching Must Reduce Cattle Product Prices (4) Reducing Cattle Product Prices Must Reduce Pasture Area (5) Reducing Extensive Cattle Ranching Must Deliver GHG Benefits(6) CRIPs Must Deliver Net Environmental & Social Benefits (7) CRIPs Must Deliver Benefits Cost Effectively.

Intensive Ranching Technologies Must be Able to Scale Up A key premise for CRIPs is the existence across Brazil of pockets of “proven and mature” cattle production technologies that offer substantially lower direct emissions and higher land intensity and than typical production systems (Pinjuv, 2011). Improved diets, genetics, and management can reduce enteric fermentation per unit feed consumed and per unit weight gained (Thornton & Herrero, 2010). Together these combine to reduce methane emissions from enteric fermentation per unit cattle product. In semi-industrialized systems, waste management can also be a means to reduce nitrous oxide and methane emissions. Over the past decade, the average lifespan of Brazilian cattle, a key indicator of direct emissions, has dropped significantly (Association of Brazilian Beef Exporters, 2012). However, the aggregate decline is not due to uniform efficiency gains in the industry, but rather a vanguard of advanced regions (Gouvello, 2010). The hope is that other regions could now follow suit. Increases in land intensity, if associated with proportional decreases in deforestation, have the potential to be a much bigger source of mitigation from ranching than reductions in direct emissions (Cederberg, Persson, Neovius, Molander, & Clift, 2011). Widespread adoption of numerous land use intensification strategies can be found across Brazil. Several broad categories of

ranching practices influence the land productivity of cattle systems (1) intensive pasture management, (2) supplemental feeding, and (3) improved health and sanitation. Intensive pasture management relies on increased use of inputs, capital, and genetic resources during the establishment phase and to some extent during maintenance phases relative to more extensive management. The combination of land grading, liming, and seeding of grasses or grass/legume combinations such as varieties of Brachiaria sp. which are heartier and more digestible than native grasses, also requires more labor than traditional pasture management, which relies more heavily on pasture rotation and seasonal burning to control overgrazing, suppress weeds, and restore soil nutrients (A Angelsen & D Kaimowitz, 2001; Vosti, Carpentier, Witcover, & Valentim, 2001; Euclides, et al., 2010). Supplemental feeding and improved health and sanitation practices can also improve system productivity and contribute to the success of more land-intensive systems. In contrast to confining cattle to feedlots, supplemental feed can be used to promote weight gain and shorten the life-cycle, to supplement pasture forage during the dry season, or to increase cattle stocking densities (Sampaio, et al., 2001). In Brazil, animals are often fed grass, hay, or sugar cane grown on-site if forage becomes scarce, and some ranchers buy similar supplemental feeds. Supplemental feeding of mineral salts (as opposed to simple salt supplements) or salts in which bovines are known to be deficient can also improve animal health and growth rates (Malafaia, et al., 2004). Improved health and sanitation practices include the treatment of parasitic infections through periodic de-worming or topical insecticides, as well as vaccination campaigns against diseases such as foot-and-mouth disease, brucellosis, anthrax, and rabies. Critiques of this premise center on the limitations of productivities per unit area as a metric for the GHG performance of any given cattle ranch. We expect, for example that yields will vary according to climate, biophysical constraints and socio-economic factors relating to the farm and its location. Short-run anomalies could arise from climatic and economic variation and longer-run anomalies could arise from the ability of ranchers to improve or worsen their lands and also as the socio-economic circ*mstances of each ranch changes. Therefore questions emerge about whether intensities are best expressed in absolute terms or relative to some local and/or ranch specific benchmarks. Research is needed to develop CRIPs-relevant cattle production intensity performance metrics. Finally, as described in our conceptual model, markets mediate the relationship between interventions in cattle ranching intensification and GHG mitigation and this adds an additional layer of complexity. We address these issues in the following sections.

Inducing Adoption of Intensification Must be Straightforward A body of literature has emerged arguing that modest investments could accelerate the adoption of a suite of proven and reasonably widespread cattle ranching technologies (Cerri, Bernoux, Maia, Cerri, Costa Junior, Feigl, Fraz„o, et al., 2010; Euclides, et al., 2010; Thornton & Herrero, 2010). Even though the technologies are already preferred on some ranches, proponents argue that investments are needed to offset high logistics costs, to renovate degraded or low productivity soils, and to provide access to investment capital. The hope is to design low-cost, far-reaching CRIPs to facilitate more rapid diffusion of intensive technologies. One set of challenges of implementing such a program stem from data and science gaps about the relative merits, for ranchers, of conventional and intensive ranching technologies. Technical comparison of ranching technologies is excellent. By contrast, research on how adoption would affect rancher profits from ranch to ranch and place to place is severely limited. This gap poses two problems. First, it is possible that it could take more than modest interventions for intensive alternative technologies to out-compete conventional technologies on the remaining conventional ranches. Second, the reverse could be true. Ranchers might continue to adopt intensive ranching practices independent of interventions. In this second circ*mstance, CRIPs would not be necessary. The reality is likely a hybrid of the two with variations from ranch to ranch and region to region. Another set of challenge concerns the constraints to technology adoption that capital access would not solve. Land market imperfections, shortages of labor and expertise, externalities, lack of information, and poor risk mitigation strategies have all been shown to be strong determinants of agricultural production choices (Jack, 2009). Meanwhile, a literature in rural sociology and anthropology questions whether rural producers are properly theorized as utility maximizers. Pockets of satisficers and ranchers of habit undoubtedly add an additional layer of complexity to the design and implementation of effective adoption policies. Research quantifying the extent and implications of these market imperfections and cultural dimensions is sorely needed.

Theorizing Ranch-Level and Regional Determinants of Adoption Presuming utility maximization, ranchers will adopt intensive practices when their expected future profits less their conversion costs exceed expected future profits of business as usual (see Lubowski, 2003 for a model example). The attractiveness of any intensification technology will vary across space, time and ranch characteristics. Variable input and output prices and the ratio of the cost of inputs

to the value of outputs will affect producers’ choices about what input mix to use in current and future periods, strategic decisions about when to slaughter, and the returns to ranching. The expectation about future input and output prices when combined with any risk or uncertainty about prices will also impact decision-making; stable, high output prices and several years of profit may make a rancher more likely to invest in land or capital to dedicate to ranching, or shift toward more input-intensive types of production. Input and output prices vary at both the local and regional levels and are mediated by supply, demand, infrastructure, physical geography, and regulation. Macro-level, local, and regional characteristics such as labor, land, and credit market conditions, current and expected transport costs, land quality, risk perception, and land use policies influence where and when producers adopt intensive technologies.

Labor & Land Labor and land markets are important determinants of profitability. Incomplete labor markets or shortage of labor supply (particularly in remote regions) may make producers more likely to choose relatively more land- or capital-intensive types of production. Competition for land among various uses, such as soybean production, cattle ranching, and sugar cane production, may drive up the land price—and the expectation that competition between land uses may happen at some point in the future and cause land values to appreciate may cause producers to ranch large tracts of land extensively in hopes that the land will be profitable in some other land use in the future (S. Hecht, 1985; Margulis, 2004; E. Arima, Barreto, & Brito, 2005; Cattaneo, 2008; R. Walker et al., 2009).

Credit Credit availability at reasonable interest rates is essential for ranchers to adopt many new technologies, including more productive grass varieties and other types of pasture productivity improvements, or to buy the capital necessary to manage land more intensively (e.g. tractors, fences, etc.). The transition from more land-extensive to more land-intensive forms of ranching requires some combination of increased input usage in the form of fertilizer, lime, grass seed, supplemental feed, mineral salt etc.; upfront investments in machinery; infrastructure and pasture reformation; and increased labor costs. While the returns over the long run to more intensive practices may make their adoption rational, many ranchers and particularly small producers may struggle to obtain credit or access to the necessary financial capital to purchase inputs or machinery. Conversely, cheap or subsidized agricultural credit may encourage ranchers to make larger-than-optimal outlays for land or capital and could result in either greater-than-optimal extent of extensive

ranching or intensive ranching, depending upon the confluence of other market characteristics (Margulis, 2004).

Transport Costs Transportation costs underlie ranching profitability because they affect both the cost a rancher pays to get beef and/or dairy products to market (or to pay someone to pick up animals and transport them to slaughterhouses) and the prices of inputs. As such, regions with high transport costs such as remote regions of the Amazon are inherently less profitable for many forms of agriculture, including ranching. This means that reductions in transportation costs, such as the construction of new roads or the paving of existing roads that allow commercial trucks to transport products to and from market even during the height of the rainy season will have important implications for the profitability of different production types and for the adoption of more intensive technologies which are more input-dependent (E. Arima, et al., 2005; A Pfaff et al., 2007; R. Walker, et al., 2009; Angelsen, 2010). The prior shape and extent of transportation infrastructure can strongly influence how a change in infrastructure influences the competiveness of different land uses (A Pfaff, et al., 2007), and changes in transportation costs play an important role in determining land and labor market conditions. Changes in transportation costs can result in out- or in-migration, and affect the wage rate/opportunity cost of labor (A Pfaff, et al., 2007). Exogenous increases in transportation infrastructure/decreases in transportation costs will be internalized to land values in the areas affected by the change, thereby causing land rent to accrue to producers in the region.

Property Rights Property rights and or land tenure policies or enforcement might contribute to either under- or over- investment in cattle ranching intensification relative to a case of tenure security (S. Hecht, 1993; M. Faminow, 1997; Angelsen, 1999; Araujo, Bonjean, Combes, Combes Motel, & Reis, 2009). One key determinant of the effect of tenure on intensity is whether tenure is exogenous (independent of any actions of the landowner) or endogenous (produced as a result of some combination of landowner actions), and if endogenous, how landowner behaviors will contribute to tenure security. Underinvestment in intensification may occur when ranchers are reluctant to adopt risky technologies or make investments on their property as a result of some risk that they will lose their land and associated investments in the future. On the other hand, overinvestment may occur in the case where land tenure is endogenous. In the case of endogenous property rights, investment in intensive technologies or

property improvements gives the landowner a stronger claim to the land, either by demonstrating productive use, or by deterring land grabbers. The types and quantities of investments landowners make may increase the probability of retaining the land over the long run. The use of cattle to establish both de facto and de jure properties rights over recently cleared land is an example of a tenure endogenous production technology that may be a major driver of extensive ranching on the Brazilian Amazon frontier (S. Hecht, 1985; Binswanger, 1991; Margulis, 2004). Wherever clearing one’s land is not only essential to obtaining/retaining tenure, but is also insufficient without the use of cattle to maintain the claim, we would expect more extensive land-use relative to a situation where tenure is secure. In such a system, the optimal density of cattle on the landscape for tenure establishment might be lower or higher than the density to maximize production77. Where questions exist about the performance of an intensification technology, the probability of adoption for the marginal rancher may depend on the frequency of previous adoption. A new technology may be more appealing for a rancher if the rancher’s neighbor has already used it and can recommend how much to apply, and can demonstrate increased profitability through its use. A notable implication is that, all else equal, prior adoption may reduce the marginal cost of adoption by lowering risk alone.

Farm Size Farm size may have several implications for the propensity to adopt intensive technologies. First, we might expect that the optimal farm size for production systems that are more labor-intensive when compared to other systems (e.g. systems that involve dairy) is smaller. Because labor requirements for these systems will vary directly with herd size, even pasture-based dairy systems are likely to be smaller and more intensive than systems that are more focused on beef production. On the other hand, large producers might be more likely to adopt intensive ranching practices if there are increasing returns to scale associated with using particular types of capital or with pasture reformation or feed storage (A. Angelsen & D. Kaimowitz, 2001; Somwaru & Valdes, 2004). Large farms located in regions where the opportunity cost of land is low, however, may be less likely to intensify,

77 For land sparing considerations, land use intensity should be measured as a function of the productive output of a cattle herd and thus the lower the slaughter rate, the lower the intensity of production. If land use intensity is measured as a function of the herd itself lower slaughter rates can increase herd density. The need to use cattle to establish/maintain property rights could discourage slaughter.

particularly if establishment of property rights still occurs through clearing and occupation/productive use (Binswanger, 1991; Margulis, 2004). Finally, it seems likely that access to credit is tied to existing farm size. If it is true that producers with larger farms and/or significant capital accumulation can more easily gain access to credit, we might expect that large and successful ranchers will be most likely to adopt more intensive technologies before these technologies diffuse to smaller and potentially more risk-averse farmers that are less able obtain agricultural credit. Understanding of technical adoption patterns across space and ranch characteristics can be improved via one or both of two data collection strategies. The first strategy would combine wall-to-wall land use data with frequently updated, farm-level agricultural census data. The second strategy would combine experimentally derived data on the efficacy of interventions to accelerate adoption with modeling of how these could be expected to change the salient outcomes for GHGs.

Increasing Intensive Ranching Must Reduce Cattle Product Prices Interventions that cause increased production of intensively produced cattle products must reduce the cattle product producer prices germane to the extensive cattle industry of Brazil. For this to be true, there must be a linkage between markets for the intensive and extensive cattle products. Until quite recently, large isolated markets persisted in the North and Center West of Brazil. Now infrastructure improvements have greatly increased market interconnectivity across Brazil (M. Faminow, 1997; Association of Brazilian Beef Exporters, 2012). Some barriers to market integration do still remain. Some areas of Brazil are still not able to participate in export-oriented markets. They are inhibited for a variety of reasons. For example, they may lack certification demonstrating that they are free of hoof and mouth disease or they may lack the ability to demonstrate that the full supply chain complies with forest laws. As Brazil’s beef exports rise, we should assess the market for Brazilian beef as part of a global beef market of increasing size. Given this scale, the extent of price effects in Brazil from an increase of intensive beef in Brazil could be dampened.78 The formation of livestock prices under an intensification intervention could depend on a number of indirect effects driven by associated inputs prices such as both fertilizer and land. Such indirect price effects could both magnify and dampen the price effects of increased output for intensive

78 As a result, changes in Brazilian cattle ranching could affect ranching systems in other nations too. These effects might deliver climate benefits that would be difficult to incorporate in some climate policy frameworks.

cattle systems. Research to quantify the price effects of increased intensive output should be a part of CRIPs pilot projects.

Reducing Cattle Product Prices Must Reduce Pasture Area A decline in germane cattle product prices could reduce the area of Brazil over which extensive ranching is profitable. The mechanism would be that the increase in supply could reduce cattle product prices enough that for some meaningful fraction of extensive cattle ranches, revenues would no longer exceed costs (D Kaimowitz & Angelsen, 2008). Over time the reduction could take the form of both a reduction in the expansion of extensive cattle ranching and the abandonment of existing extensive areas. This premise could hold true in sufficiently small markets where a small intervention could have a meaningful price effect or in large markets where a sufficiently large intervention is undertaken. The activities of land users at the Brazilian forest frontier cannot be explained by agricultural profit maximization alone. First timber and non-timber forest products can be important sources of household income. Second, not all frontier actors are profit maximizers; some may seek to maximize utility in ways such that they do not maximize profit. If it were leisure, higher production prices would mean greater profitability per unit area and a decline in the area required to satisfy the leisure objective (Angelsen & Kaimowitz, 2001). Of even greater salience, however, is ample evidence suggesting that Brazilian ranchers hold land not only to produce cattle products, but also to secure tenure, obtain access to government subsidies, and to speculate on increases in land values (Kaimowitz & Angelsen, 2008). Each of these considerations could dampen the extent that cattle product price declines lead to declining area of extensive ranching. A further consideration is the extent to which demand for cattle products varies as prices vary. If a price decline leads to an increase in demand, this too would dampen the land-sparing phenomenon. Of note is not only the own price elasticity of demand, but also cross price elasticities of consumption substitutes. Supporting CRIPs means assuming that demand for cattle products is sufficiently inelastic such that when yields increase, areas will decrease enough to spare land. Inelastic demand may apply to staples and to food in aggregate. In this way, higher efficiency food production may escape Jevons’ Paradox—i.e. the circ*mstance where a technology makes a production process more input efficient, reduces output prices, and thereby increases aggregate demand by the sector for the input (Alcott, 2005). Circ*mstances of elastic demand worsen the problem, all else equal. Figure JJJ provides a diagram of Jevons’ Paradox.

Figure 3-2. Jevons’ paradox illustrated. Jevons’ paradox highlights circ*mstances where efficiency gains in sourcing the raw material basis of a product lead that same industry to consume more of that same material overall. Jevons observed that James Watt’s energy efficient coal boiler led to lower energy prices, increased consumption of energy, and increased the aggregate quantity of coal used for coal boilers. The paradox is not that lower energy prices increased energy consumption, it is that higher efficiency boilers increased coal use. In the case of cattle systems, the analogous concern would be that more land efficient cattle production systems might sufficiently lower the costs of cattle products to induce more land use in cattle systems than if the cattle systems were less land efficient. Thus for land sparing, ceteris parabis, the magnitude of land efficiency gains must trump the magnitude of increased consumption that results. Such a circ*mstance is more likely to occur under scenario (A), an example of price effects of improved technology under price inelastic demand. Under scenario (B), where demand for the final product is price elastic, Jevons’ Paradox is more likely to occur. (C) shows the differential effect of (A) and (B) on the consumption by the sector of the raw material. Note that Q0 could be less than QA, greater than QB and anywhere in between. Thus (C) shows the increase in input use due to price effects compared to perfectly inelastic demand. However, these price effects may not lead to increased input use in absolute terms. Demand for beef might be quite price elastic (Andreyeva, Long, & Brownell, 2010). Also, the central tendency of an increase in the share of food from beef would be

substantial increases in GHG emissions from food consumption.79 The best solution to the problem would be two pronged: research the relationship between cattle product prices and ranching intensity, and in the meanwhile design corrective taxes such that CRIPs keep consumer prices for beef at equal or higher levels.

Reducing Extensive Cattle Ranching Must Deliver GHG Benefits The fifth premise is that reduced extensive cattle ranching can have three effects relative to business as usual- reduce direct emissions, avoided deforestation and land spared for other productive uses.

Reduced Direct Emissions If a unit of beef is produced using the semi-intensive technologies available in Brazil it is most likely that these emissions will be measurably less than emissions from typical extensive production. This would be due to primarily to reduced enteric fermentation (Thornton & Herrero, 2010). If land is in fact spared, questions do arise, however, about the soil carbon that could be lost in the conversion from pasture to cropland. The quantity of carbon sequestered in the managed pasturelands of South America may be quite high and may be at risk of loss if converted to croplands (Fisher et al., 1994; Fargione, Hill, Tilman, Polasky, & Hawthorne, 2008).

Avoided Deforestation CRIPs proponents have argued that extensive cattle ranching causes the bulk of deforestation in the Brazilian Amazon and, as result, reduced extensive ranching would substantially reduce deforestation (Gouvello, 2010; Martha, et al., 2012). Ranching has been identified as strongly correlated with recently deforested lands in Brazil and this has been widely and liberally interpreted to suggest that ranching is in fact is the cause of this deforestation (Federal Government of Brazil, 2004;

79 Further research is needed on the contribution to land use change of an addition or subtraction of a marginal unit of livestock products. Such analysis depends on the extent to which land were functionally equivalent to produce beef and other feed products. To the extent that land is suitable for use as either rangeland or cropland beef and other ruminant meat would be by far most climate intensive food. Meat of monogastrics and other livestock products would much better than beef, but somewhat worse than vegetable products. For more, see Weber, C., & Matthews, H. (2008). Food-miles and the relative climate impacts of food choices in the United States. Environ. Sci. Technol, 42(10), 3508-3513. To the extent that beef is produced on land unsuitable for other livestock and agricultural production, perhaps this hierarchy changes. Further research is needed on the contributions to the suitability of land attributes such as yield potential, governance, and levels of market access.

David Kaimowitz, Mertens, Wunder, & Pacheco, 2004; Wassenaar, et al., 2007; Cederberg, et al., 2011; Mercedes Bustamante, et al., 2012). In its action plan for prevention of deforestation of the Amazon (2004:10), the Government of Brazil asserts that, “Cattle ranching is responsible for almost 80 percent of the total deforestation in the legal Amazon.”80 In response to the rapid growth of the industry and allegedly associated deforestation and GHG emissions in Brazil, a number of government and non-government initiatives have sought to curb ranching expansion. This includes the Cattle Agreement for traceability to prevent sourcing from recently deforested ranches, the Brazilian government’s embargoed municipalities list, and efforts by several Brazilian public prosecutors to limit credit for cattle sector actors that don’t follow forest laws (N. Walker, Bramble, & Patel, 2010). The theory is that CRIPs too can curb ranching expansion. Perhaps the best-known study on the environmental impacts of livestock also presumes that correlation of pasture and recently deforested land indicates that the pasture has caused the deforestation. Steinfeld et al. (2006)base their estimate of aggregate LUC emissions caused by livestock on Wassenaar et al. (2007). In Wassenaar, the authors clearly state that they have simulated the likely future location of cattle ranching based on biophysical characteristics of current cattle ranching. They then overlay this on a simulation of deforestation conducting using similar criteria. As Wassenaar et al. (2007) explain, the relationship that they simulate is correlative, not causal. Cederberg et al. (2011) employ another approach that also considers correlation a proxy for causation. They “present a model to distribute emissions from land use change over [cattle] products…” produced in the Brazilian Amazon. The Cederberg team uses a land use change model developed by Fearnside (1996) and presented in Ramankutty et al. (2007) to estimate (i) total emissions from deforestation in the Brazilian Amazon and (ii) the proportion of deforested land in Brazilian Amazon eventually occupied by pasture as the, “final land use stabilized after ~50 years i.e. the equilibrium state”. They use (ii) to estimate the proportion of (i) to attribute to cattle production. This is supposition since the Markov land

80 This number most likely comes from an analysis at the census tract level to estimate the proportion of recently deforested land occupied by cattle pasture. See Chomitz, K., & Thomas, T. (2001). Geographic Patterns of Land Use and Land Intensity in the Brazilian Amazon. Washington DC: The World Bank.. A forthcoming study finds that ranching occupies 62 percent of the area deforested in the Brazilian portion of the Amazon Biome over the period 2007-2010. See National Institute of Space Studies of Brazil (INPE), & Brazilian Agricultural Research Corporation (EMBRAPA). (2011). Terraclass: Study of Land Use and Land Cover in the Amazon Biome (Executive Summary). Sao Jose dos Campos, Sao Paulo Brazil: National Institute of Space Studies of Brazil (INPE).

transition model employed imputes no causality. Since there is no clear causal link between ranching activity and the deforestation, Cederberg et al. present three scenarios for determining which, “beef should carry the burden of the emissions.” The scenarios are beef from recently deforested land in the Amazon, all beef from the Amazon, and all beef from Brazil. They could easily present all global beef, all beef consumed by consumers who eat more than a certain quantity of beef, or all beef made into hamburgers. Their approach is normative and arbitrary. Cattle ranching is a cause and perhaps it is even the largest cause of Brazilian deforestation. Some of this deforestation includes some or all of the forest lost at the location of the cattle ranching itself. Interventions that reduce the area of ranching in Brazil could directly and/or indirectly reduce the amount of deforestation in Brazil. It is not necessarily true, however, that avoided deforestation would occur where the ranching is lost or in proportion to the area of ranching lost. The management practices of each ranch, it agronomic resource endowments, its isolation from markets and its market and regulatory context will affect the net influence of the ranch on agricultural extent and forest cover. For this reason, it is crucial to base policy in a strong understanding of the relationship between forests and extensive livestock. One consequence of not doing this could be to overestimate the GHG benefits from reducing extensive ranching extent. This is because if extensive ranching in general and in particular demand for extensively ranched cattle products is not causing all the deforestation it is occupying, reducing extent of extensive ranching might not create a corresponding decrease in deforestation. Since some the mechanisms by which agriculture affects deforestation are global (i.e. commodity markets), spatio-temporal variation will never fully capture the effects of agriculture on forests. Therefore, there will always be a role for land use change models to supplement observation on the deforestation effects of agricultural systems.

Increased Productive Uses The third portion of this premise is that by occupying a very large area in Brazil, cattle pasture is displacing more productive land uses. In 2006, at last count, cattle ranching occupied roughly 200 million hectares of Brazil, more than one fifth of the land surface of Brazil (Census Bureau of Brazil, 2010). By contrast, all of crop agriculture in Brazil combined occupies just 62 million hectares (Census Bureau of Brazil, 2011).

Ricardo (1891) and Von Thunen (J. H. Von Thunen & Hall, 1966), theorize that the quality of land and its isolation from markets can influence it value and its use. With a potential for value to now be placed on land use changes emissions and/or emissions reductions from reforestation, a third consideration for CRIPs would thus be the GHG value of each piece of land according to its role in the globe’s carbon stock and/or flow. Such a value could correspond to the GHG benefits of maintaining an ecosystem and/or the greenhouse gas costs of converting it (Anderson-Teixeira et al., 2012). The relative significance of these three distinguishing characteristics, however, has not been rigorously examined and it is not a typical component of discussions on the fate of Brazil’s 200 million hectares of cattle pastures. For CRIPs, research is needed to develop a new functional unit for land that takes into account land quality, isolation from markets, and GHG value.

CRIPs Must Deliver Net Environmental & Social Benefits A fundamental premise for promoting CRIPs is that unintended environmental and social costs of the program do not erase the benefits gained from GHG mitigation (at a given estimate of the social cost of carbon).

Non-Climate Environmental Effects of CRIPs The environmental benefits of agricultural intensification are of debated size because it can be difficult to compare local loss of environmental quality with global land sparing. Serious environmental impacts are associated with a number of forms of agricultural intensification and in particular with industrial livestock production (Naylor et al., 2005). Water use and pollution are especially pressing and costly problems and their effects are magnified in regions with poor environmental governance. With the exception of the United States and Europe, the cattle sector has not seen the same levels of industrialization as other livestock sectors. Intensification need not be associated with more local environmental costs, however. Opportunities for environmentally friendly intensification exist, suggesting that under some circ*mstances local environmental quality vs. global land sparing is a false choice (P. Matson, Parton, Power, & Swift, 1997; Tscharntke, Klein, Kruess, Steffan-Dewenter, & Thies, 2005).

Social Effects of CRIPs At the country level, some argue that agricultural intensification may be a necessary but not sufficient condition for development. Yet to more specifically examine when and how agriculture contributes to development, more empirical

research is necessary, and/or modeling that better incorporates non-linear systems, threshold effects and other complex aspects of the problem (Lee & Barrett, 2001). Perhaps such work could help to identify priority intensification opportunities. When we consider the distributional implications of CRIPs, it is important to consider several key factors: the existing policy landscape, the dynamics of the adoption process associated with more intensive technologies, the costs, benefits, and externalities associated with ranching and other competing land uses, and the relative cost-effectiveness of ranching intensification as a GHG mitigation strategy. Whether future intensification or policies designed to encourage intensification result in welfare gains or losses depends on the broader land use policy landscape (Zilberman, Schmitz, Casterline, Lichtenberg, & Siebert, 1991). For example, productivity improvements (and the resulting shift in supply) in ranching could be social-welfare-improving if the government subsidizes producers via credit programs, but it depends on the opportunity cost of the subsidy and the economy-wide effects of the tax system. Conversely, there may be a net social welfare loss (through price pass through to consumers) in the absence of government interventions, but this could be offset by the opportunity for government spending on other more socially beneficial things. Adoption of intensification technologies may require up-front capital investments and potentially exhibit economies of scale so wealthier producers might more quickly adopt intensification technologies. This trend, however, might result in a reduced cost of implementation of the technologies for intensified production and some learning-by-doing for smaller or less-wealthy producers. Smaller producers might still adopt, but lag behind larger producers. This would result in potential welfare improvements for different types of producers at different points in the adoption process. If larger producers adopt first, there is also the potential for concentration of landholdings, however, as larger producers accumulate capital more quickly and take advantage of any economies of scale that might not be available to smaller landholders.

CRIPs Must Deliver Benefits Cost Effectively Perhaps the most important element of CRIPs is that they could deliver large-scale GHG abatement at less than the cost of implementation and less than the costs of many mitigation alternatives. In its NAMAs, Brazil has pledged for the years 2010-2030, over 7 billion tons in CRIPs-centered GHG mitigation activities worth roughly $140 billion81, but at an abatement cost estimated to be closer to $100

81 At August 2011 EU ETS carbon prices retrieved from pointcarbon.com

billion in private costs (Gouvello, 2010). Yet for CRIPs to be well designed and executed, additional costs will arise in order to close crucial data and science gaps required for good land use governance in Brazil. If CRIPs were to deliver 7 billion tons of CO2e of mitigation, the scale of the potential risks and opportunities for CRIPs would nonetheless warrant a substantial investment by buyers and/or sellers to verify the carbon value that CRIPs could generate. Here we make a rough estimate of costs associated primarily with closing key data gaps. We also list a number of other science and data gaps of uncertain costs. Better monitoring of the area, productive capacity and use of pasturelands could boost the viability of CRIPs. In the past Brazil has accounted pasture area only once a decade as part of the national agricultural census. If area were accounted with greater frequency- perhaps annually—it would substantially improve understanding of the drivers of pasture use changes. The agricultural census is expensive to conduct. The entirety of the last agricultural census cost roughly $250 million (Oliveira, Bolliger, & Florido, 2006). At that price, annual estimates of pasture area would be less than 10 percent of the purported mitigation benefits of CRIPs. Much cheaper alternatives may exist however. In Fall 2011, The Brazilian Space Agency (INPE) was set to release data from the first iteration of TerraClass, a collaboration between INPE and The Brazilian Agricultural Research Corporation to classify land use occupying the roughly 70 million hectares of recently deforested land in the Legal Amazon. Endowed with roughly $400 million by the Brazilian government and a consortium of Northern nations, TerraClass is intended to map land cover including pasture each year. TerraClass is able to leverage PRODES, a $1 million per year system to monitor deforestation across the 320 million hectares of the Legal Amazon (Bitencourt, 2011; Brazilian Ministry of Science & Technology, 2011; G1 News Brazil, 2011). Even if we assume that extending TerraClass to the entire country would require extending PRODES-style satellite monitoring to whole country, monitoring of pasture area during the NAMAs period could cost as little as $1 billion total. Moreover, since the systems leverages remote sensing, perhaps it would be reasonable to also develop assessments on the productivity of pasturelands. Knowing productivity potentials could be helpful to identify land-conserving management practices and establish performance metrics for CRIPs. An excellent complement to improved monitoring of area and quality of pastureland would be to monitor the location and movements of the national cattle herd. Traceable radio frequency identifier chips are already implanted in some cattle in Brazil to prevent sourcing from illegally deforested lands. Traceability

could also greatly aid measurements of cattle ranching productivity. Baseline levels of productivity are, in turn, critical to estimate the extent of intensification a CRIP causes. Early experience from the cattle industry indicates that traceability can be expensive. While the hardware costs $4/animal life, in a personal communication to us, one industry source estimates total private costs of approximately $30/animal life. Since animal lives are roughly three years this would mean costs of roughly $10/animal/year. For 200 million animals over 20 years this would mean $40 billion in private costs of tracing cattle. Private traceability initiatives are prone to fraud and so government enforcement would be needed. This too could be substantial. Cattle ranches cover one quarter of Brazil’s land area and enforcement in these areas has been lax. If it has been zero enforcement up until now and if enforcement varies with area, this could increase the budget for environmental enforcement in the country by 25 percent or roughly $500 million annually (Senate of Brazil, 2010). Thus in all, cattle traceability could cost $50 billion dollars over the course of mitigation period. This is an untenably high figure as it comprises a 50 percent increases in CRIPs implementation costs and would mean that CRIPs costs would exceed CRIPs benefits. Its cost would need to be reduced by a half in order to enter the feasible range. Streamlining seems a reasonable expectation. If instead of a census a random sampling approach were employed, cost savings could be dramatic. The GHG impacts of CRIPs cannot be directly measured because they intend to affect global land cover and there is therefore no way to control for other influences on land use change. However, better modeling of the GHG benefits could be accomplished through further data, and science. We specify a number of research and agricultural statistics priorities in this section and at the end of each of the first six premise sections. Publicly-available, spatially-explicit agricultural input and output price data, scientific data on the propensity of adoption of intensive practices and data on the influence of environmental regulations on production practices would all be essential to parameterize a model of CRIPs impacts on GHGs. A great deal of this data collection would be best accomplished by designing the early stages of CRIPs as randomized control trials. This would be the most straightforward way to improve the science of agricultural technology adoption and ultimately the potential to manage GHG emissions from the Brazilian cattle sector.

Conclusions To reduce its climate emissions, Brazil, like many Southern countries must reduce emissions from land use. These dominate all anthropogenic emissions. Given its

large area, cattle ranching will surely play a role in future of land use and hence climate policy. For this reason and the many premises we outline, CRIPs have become central to discussions on strategies to address agricultural drivers of deforestation and develop climate-friendly biofuels policies. As we have demonstrated, however, a number of these premises could lead to unintended climate costs from some CRIPs approaches. For now, it is best to proceed cautiously with CRIPs, to begin by collecting data on cattle systems and to develop trials to build scientific understanding of rancher technology adoption. Even with better science and data, the viability of CRIPs for climate mitigation will be contingent on factors beyond the cattle sector like forest governance, and beyond Brazil’s borders like global demand for cattle products. In the meanwhile using CRIPs to close data and science gaps can pay dividends not only as preparation for CRIPs, but also by enabling better management of cattle ranching for broader social and environmental benefits.

Concluding Thoughts Rio de Janeiro, Brazil (6-25-12) – “Learn to measure so that you can manage!”

Slide from University of California, Berkeley, Civil Engineering 268E, January 22, 2008, slide 5 of 16, Professor Arpad Horvath.

“If you can’t measure it you can’t manage it.”™

Registered as a trademark by the Kinex AHA Corporation, March 200082 “Count what is countable, measure what is measureable, and what is not measureable make measureable.”

Galileo Galilei

Overview I conclude by explaining how my dissertation helps to show a number of concrete ways to build on the green accounting model of earth systems governance. I use the case of land use governance to demonstrate the need for environmental social science that complements accountings of physical processes. Social science is needed because social factors including policy design, policy adoption, cognition, cultural dimensions, and price effects can all shift the relationships between production activities and environmental outcomes. Social elements are foundational to the design of effective policies. The study of these phenomena will require the cooperation of policymakers, the use of mixed methods, and it cannot be done in silos. The effort will require quantitative research in communication with qualitative research, model-to-model comparisons, and numerous other efforts to validate and translate. Finally, scholars and policy makers alike will need to be reflexive in order to surmount the challenges of this ambitious agenda. With all of

82 This is according to a thread on the website Google Answers - http://answers.google.com/answers/threadview?id=139473 .

the above, efforts to end deforestation can inform earth systems governance that supports people and our planet.

Green Accounting and Earth Systems Governance Dan Suarez, a student in my program and one of my hostel roommates has been enormously busy here in Rio following around an Earth Summit protagonist as part of collaborative ethnographic research on the meeting. He is assigned to Pavan Sukhdev, perhaps the most active participant at the Earth Summit. Sukhdev has been on a mad dash around the Rio events, speaking at expensive invitation-only business fairs, open-to-the public lecture halls, and at the International Society of Ecological Economics Conference that I’ve been attending. Trained as an economist and formerly a banker, Sukhdev is the lead author of four interlinked books, published by the United Nations Environment Program called The Economics of Ecosystems and Biodiversity (TEEB) (Sukhdev, 2009). The TEEB reports are voluminous, meticulous, multi-faceted, evangelical, and, like Sukhdev’s madcap schedule of Rio appearances, packaged to reach multiple audiences. A volume for local policymakers is distinct from one for national policy makers. Editions for citizens, scientists, and business leaders also exist. Framing aside, all the volumes share a common theme—Sukhdev’s ethos that green accounting is the backbone of the green economy.83 Before he wrote the TEEB volumes, Sukhedev founded and directed the Green Indian States Trust, an ambitious green accounting project in India. Green accounting has no fixed or formal definition, but the type that Sukhdev practices works to develop a matrix of all sectors of an economy populated with the environmental costs and benefits that each production practice requires as inputs and constitutes as outputs. These effects vary according to the price of things, but a core premise is that at any given set of prices, production activities will have environmental costs and benefits that are intrinsic. The green accounting ethos is summarized by the trademarked quote, ubiquitous in the management world and featured at the beginning of this section, “If you can’t

83 To be fair, TEEB involves more than just accounting the environmental impacts of the economy. It also seeks to, “draw attention to the tangible benefits of biodiversity, and to highlight the growing costs of biodiversity loss and ecosystem degradation.” For more, see page 277 of Sukhdev, P. (2009). Costing the earth. Nature, 462(7271), 277-277.

measure it, you can’t manage it.”84 The quote is itself modest in its claims about whether measurement of a set of things inevitably leads to management of those same things. A spinoff phrase85 widely used across peer-reviewed literature, “measuring and managing” seems to more blithely imply that measurement may catalyze green outcomes. Other elements such as managerial will and know-how at designing interventions are surely also essential but throughout the literature, there is a consistent focus on the importance of measurement for management.

Case as Critique: Green Accounting and Deforestation Policy REDD policy circles have had a focus on green accounting in the form of satellite monitoring of tropical deforestation. Long before Brazil was fighting deforestation effectively, Brazil was heralded as being on the cutting edge of deforestation policy primarily because of the Project for Monitoring of the Brazilian Amazon by Satellite (PRODES), a system for classifying deforestation in satellite images (Câmara, Valeriano, & Soares, 2004). PRODES is renowned for its design, transparency, and ease of use.86 Expanding PRODES-style systems across the tropics has ,for good reason, been a major emphasis in REDD readiness funding. Merely monitoring the rate of deforestation is far from a green accounting of the land use change process. A recent work by the Center for International Forest Research (CIFOR) calls more nuanced satellite monitoring and better interpretative capacity (Herold, et al., 2012). The rationale is that satellite monitoring of deforestation on its own is merely a preliminary step for green land accounts. 84 The aforementioned quote and its variants are apparently quite popular in business management and engineering. I first heard a version of the phrase in my first engineering course, a graduate level Civil Engineering course in lifecycle analysis. Early in the semester, in a lecture on the motivation for green accounting, the professor, Arpad Horvath, who pioneered the link between green and financial accounting matrices, displayed a slide with the bullet point, ““Learn to measure so that you can manage!” 85 This exact phrase appears in over 20,000 scholarly works that appear in search results on Google Scholar. 86 The Brazilian national space agency, INPE is a highly regarded, highly capacious agency, PRODES is user friendly, transparent and vetted, and Brazil pursuing South South memoranda of understanding to go further. This week Brazil announced a partnership to train African nations in fighting deforestation. Some see such partnership as a crucial element of the fight to slow pan-tropical deforestation. For more see Butler, R. (2012). 10 African countries to develop satellite-based deforestation tracking systems with help of Brazil. Retrieved July 27th, 2012, from http://news.mongabay.com/2012/0730-african-mrv.html

Herold et al. (2012) argue that it is important to monitor not just the transition from forest to non-forest, but so-called wall-to-wall land use—all changes in all land cover types. This would require discerning shifts between difficult to distinguish landcover types; for example, pasture and savannah. Such a mapping must entail a series of map classifications and not just a snapshot. With a frequently updated wall-to-wall land cover map, a country could then develop a matrix of transition probabilities, i.e. the likelihood that any land use is to become any other land use (see for example P. M. Fearnside & Guimaraes, 1996; Ramankutty, et al., 2007). While scholars and policymakers are positioned to use a transition matrix like a green accounting matrix, the two are actually quite different., a transition matrix represents likelihoods, while a green accounting matrix represents responsibilities; the difference is that of correlative and causal relationships. In a typical green accounting matrix, the implication is that the environmental costs and benefits exist as a consequence of the production activity with which they are associated. A matrix of land use transition probabilities, by contrast, represents just spatio-temporal proximity. It does not measure the causes of deforestation and thus, this data alone it is insufficient to know what the causes of deforestation are. That said, the land transition matrix may be instrumental for eventually understanding the causes and consequences of land use patterns. And, indeed, it is this assessment of what the authors call “drivers” that constitutes the next step after satellite data in the best-case version of green landscape accounting described in Herold et al. (2012). Here, however, the logic breaks down, at least to the extent that existing government activities speed or slow deforestation and do not vary sufficiently so that their effects can be discerned by variations in the land use matrix. The green landscape accounting is undoubtedly important to aid in examining the causes of deforestation, but it may do little to allow the examination of state-led causes of deforestation. How exactly can the deforestation footprints of products and practices be intrinsic if the rate of deforestation is contingent on policies and regulations? This is the same shortcoming I describe in my discussion in Chapter Three of the contingencies associated with estimating carbon footprints of beef.87

87 In a related piece with colleagues from Berkeley, we demonstrate that for products derived principally from land use activities, the “green” is nearly always strongly contingent on regulations. For more, see Lemoine, D., Plevin, R., Cohn, A., Jones, A., Brandt, A., Vergara, S., et al. (2010). The Climate Impacts of Bioenergy Systems Depend on Market and Regulatory Policy Contexts. Environmental Science & Technology, 44(19), 7347-7350.

Given this, the notion of a green accounting of the environmental burden of land use activities is problematic. But, there are solutions.

Deforestation Policy Analysis to Inform Deforestation Policy One solution is authoritarian whole landscape zoning and control. Such a system would eliminate the need to understand the processes that explain land use dynamics because all changes would be by mandate. Such a system would also obviate the need to monitor deforestation at all. Despite some convincing arguments that forest governance may re-centralize under REDD, few imagine a scenario where complete centralization would be a beneficial development (Phelps, et al., 2010).88

88 In an online article about the future of economics, Ali Wyne asks “eight of the world’s top young economists” to, “identify the biggest unanswered questions in economics and predict what breakthroughs will define it a decade or two hence.” Twenty seven year old, University of Chicago Economics Professor, Glenn Weyl, muses about the relationship between information and government referencing a 1945 article by one of the Chicago School’s most famous:

In his famous 1945 article, “The Use of Knowledge in Society,” F. A. Hayek argued that despite their inequity and inefficiency, free markets were necessary in order to allow the incorporation of information held by dispersed individuals into social decisions. No central planner could hope to collect and process all the information necessary for social decisions; only markets allowed and provided the incentives for disaggregated information processing. Yet, increasingly, information technology is leading individuals to delegate their most “private” decisions to automated processing systems. Choices of movies, one of the last realms of taste one would have guessed could be delegated to centralized expertise, are increasingly shaped by services like Netflix’s recommender system. While these information systems are mostly nongovernmental, they are sufficiently centralized that it is increasingly hard to see how dispersed information poses the challenge it once did to centralized planning. Information technology thus fundamentally challenges the standard foundations of the market economy. For many years to come, economists will increasingly have to struggle with this challenge. Some will harness the power of the data and computational power provided by information technology to provide increasingly precise and accurate prescriptions for economic planning. Others, who value the libertarian tradition that has often been associated with economics, will be forced to articulate other arguments, perhaps based on privacy, that are not susceptible to erosion by the increasing power of centralized computation.

Based on the my argument above, the role of government in the deforestation would undermine authoritarian land planning based on “centralized computation.” Either land planning is entirely authoritarian or it is something else.

Another, more realistic solution is to undertake policy analyses that empirically examine or simulate the deforestation effects of policy interventions. This means analyzing the activities of firms, and consumers, and also the activities of governments. Such analyses may draw on land transition matrices, as well as other green accounting tools. However, ultimately, analyzing the efficacy of policies is the best ways to make effective policies. What is somewhat novel here is the idea that systemized environmental social science data and theories constitute a complement and occasionally a substitute for the green accounting backbone of earth systems governance.89 The research itself is not necessarily novel, but doing it well enough to allow comparisons across space, time, and research groups would pose a non-trivial challenge. This would undoubtedly require close collaboration between scholars. As various parts of this dissertation show, policy design, policy adoption, cognition, cultural dimensions, and price effects can all shift the relationships between production activities and environmental outcomes. Therefore, all of these factors are crucial elements for informing deforestation policy. It is crucial to work to systematize these sorts of data to link environmental and social dimensions of land use. Scholars are likely need some help in this regard. Policymakers will need to work closely with scholars to use policies as experimental, data generating processes. Ideally, they would work together on experiments where policymakers will randomly assign policies to allow scholars to study the effects of interventions to prevent deforestation. Along with partners from the academe, industry, and government, I will be running one such policy experiment as part of a fellowship with Energy Biosciences Institute, at the University of California, Berkeley. In it, we will employ a

89 I say somewhat novel because a number of examples of collaborative data collecting efforts already exists and are starting to be employed to inform policy. The Climate Change Agriculture and Food Security of the Consultative Group on Agricultural Research has benchmark cites for collecting social and environmental data on farmer adaptation to climate change. For more, see http://ccafs.cgiar.org/where-we-work. International Forestry Resources and Institutions maintain a combined social and environmental database of variables describing forest communities around the world. For more, see www.umich.edu/~ifri/ and Persha, L., Agrawal, A., & Chhatre, A. (2011). Social and ecological synergy: Local rulemaking, forest livelihoods, and biodiversity conservation. Science, 331(6024), 1606.

randomized control experimental design to examine the effectiveness of incentives and information in influencing cattle ranchers to take up semi-intensive alternative production practices. Experiments are critical, but they should be just one of a pluralistic palette of methodological approaches to deforestation policy analysis. Qualitative, historical, ethnographic, and modeling methods are all also of crucial importance to examine germane social processes.

Validate and Communicate Ultimately, one of the greatest strengths of methodological pluralism is interdisciplinary interaction and the creative tensions that this fosters. The social science of preventing deforestation will require epistemologies to meet and clash, models to be compared with other models, observation to be compared with models, and in general, even more triangulation, communication and validation than we now see. I am currently working with the National Center for Atmospheric Research (NCAR) Integrated Assessment Modeling (IAMs) group to develop a project for comparing backward-looking simulations of market-driven agricultural extensification and related deforestation with empirical observations of deforestation over the same period. The goal is to examine how much deforestation can be explained by markets alone. Our aim is to inform the debate about the role of prices and policies in causing deforestation and advance methods going forward (Andersen & Reis, 1997; Hargrave & Kis-Katos, 2011; Assunção, et al., 2012). However, I see the work of validation as having much broader goals. One exciting and completely distinct example is research comparing the pursuits and proclivities of political ecology and land change science (B. Turner II & Robbins, 2008). Here the focus is on comparing two very different disciplinary approaches to the study of land use change.

Reflexivity The aforementioned review piece is notable not only for cross-episteme communication, but also the reflection that it has sparked in the field. Jacqueline Vadjunec & Christian Brannstrom are preparing a book project, based on the article, that will be written by and primarily for political ecologists (Vadjunec &

Brannstrom, 2012). They will provide analysis on political ecology’s encounter with LCS. This sort of reflexivity about norms, methods, theories, and performance metrics of environmental social science is of crucial importance. Not only do these things shape our science, but as I show in Chapter One, they also have the potential to shape policies. As Galileo reminds us, we as scholars can and should choose what we make measureable. Reflexivity can take many forms that extend beyond the sort found in the qualitative social sciences. For example, a former professor of mine, Holmes Hummel, built a formal model to demonstrate the importance of assumptions that are embedded deep within integrated assessment modeling of climate change mitigation (Hummel, 2006). She looked at the sensitivity of model outcomes to these assumptions about people, markets and the environment, and astutely described how these sorts of assumptions could easily leach into policy. This type of honest and probing analysis, presented publicly and accessibly, is also crucial for advancing the environmental social sciences. With all of the above, environmental social to examine and support deforestation policies can be foundational to more effective earth systems governance.

References Agrawal, A., Nepstad, D., & Chhatre, A. (2011). Reducing emissions from

deforestation and forest degradation. Annual Review of Environment and Resources, 36(1).

Akanle, T., Appleton, A., Kulovesi, K., Recio, E., Schulz, A., Spence, C., et al. (2010). Summary of the Cancun Climate Change Conference: 29 November – 11 December 2010. Earth Negotiations Bulletin, from http://www.iisd.ca/vol12/enb12498e.html

Alagiri, P. (1991). Give Us Sovereignty or Give Us Debt: Debtor Countries' Perspective on Debt-for-Nature Swaps. Am. UL Rev., 41, 485.

Alcott, B. (2005). Jevons' paradox. Ecological Economics, 54(1), 9-21. Alexandratos, N. (1999). World food and agriculture: outlook for the medium and

longer term. Proceedings of the National Academy of Sciences, 96(11), 5908.

Allen, J. C., & Barnes, D. F. (1985). The causes of deforestation in developing countries. Annals of the Association of American Geographers, 75(2), 163-184.

Amend, M., dos Santos, A. S., & Mattos, L. (2011). Subsídios para a pecuaria ea conservacao da floresta: estimativas para o município de Humaita, Amazonas. from http://conservation-strategy.org/sites/default/files/field-file/Subsidio_-_Final_Report.pdf

Andersen, L. E., & Reis, E. J. (1997). Deforestation, development, and government policy in the Brazilian Amazon: an econometric analysis (Vol. 513): Ipea.

Anderson-Teixeira, K. J., Snyder, P. K., Twine, T. E., Cuadra, S. V., Costa, M. H., & DeLucia, E. H. (2012). Climate-regulation services of natural and agricultural ecoregions of the Americas. Nature Climate Change.

Andreyeva, T., Long, M. W., & Brownell, K. D. (2010). The impact of food prices on consumption: a systematic review of research on the price elasticity of demand for food. American Journal of Public Health, 100(2), 216.

Angelsen, A. (1999). Agricultural expansion and deforestation: modelling the impact of population, market forces and property rights. Journal of development economics, 58(1), 185-218.

Angelsen, A. (2010). Policies for reduced deforestation and their impact on agricultural production. Proceedings of the National Academy of Sciences, 107(46), 19639-19644

Angelsen, A., & Kaimowitz, D. (2001). Rethinking the Causes of Deforestation: Lessons from Economic Models. The World Bank Research Observer, 14(1), 73-98.

Angelsen, A., & Kaimowitz, D. (Eds.). (2001). Agricultural Technologies and Deforestation. New York: CABI International.

Araujo, C., Bonjean, C. A., Combes, J. L., Combes Motel, P., & Reis, E. J. (2009). Property rights and deforestation in the Brazilian Amazon. Ecological Economics, 68(8-9), 2461-2468.

Arima, E., Barreto, P., & Brito, M. (2005). Pecuária na Amazônia: tendêndias e implicações para a conservação ambiental. Belém: Imazon.

Arima, E. Y., Richards, P., Walker, R., & Caldas, M. M. (2011). Statistical confirmation of indirect land use change in the Brazilian Amazon. Environmental Research Letters, 6, 024010.

Arima, E. Y., & Uhl, C. (1997). Ranching in the Brazilian Amazon in a national context: Economics, policy, and practice. Society & Natural Resources, 10(5), 433-451.

Association of Brazilian Beef Exporters. (2012). Beef Cattle Data Sheet: 1994-2011 Retrieved July 28, 2012, from http://www.abiec.com.br/download/balanco.pdf

Assunção, J., e Gandour, C. C., & Rocha, R. (2012). Deforestation Slowdown in the Legal Amazon: Prices or Policies? Rio de Janeiro, Brazil: Climate Policy Initiative.

Aukland, L., Costa, P. M., & Brown, S. (2003). A conceptual framework and its application for addressing leakage: the case of avoided deforestation. Climate Policy, 3(2), 123-136.

Auld, G., Gulbrandsen, L. H., & McDermott, C. L. (2008). Certification schemes and the impacts on forests and forestry. Annual Review of Environment and Resources, 33, 187-211.

Babiker, M. H. (2005). Climate change policy, market structure, and carbon leakage. Journal of International Economics, 65(2), 421-445.

Balmford, A., Green, R., & Scharlemann, J. (2005). Sparing land for nature: exploring the potential impact of changes in agricultural yield on the area needed for crop production. Global Change Biology, 11(10), 1594-1605.

Barona, E., Ramankutty, N., Hyman, G., & Coomes, O. T. (2010). The role of pasture and soybean in deforestation of the Brazilian Amazon. Environmental Research Letters, 5, 024002.

Barreto, P., Arima, E., & Brito, M. (2006). Cattle ranching and challenges for environmental conservation in the Amazon. IMAZON, BelÈm, Brazil.

100

Bartley, T. (2003). Certifying forests and factories: States, social movements, and the rise of private regulation in the apparel and forest products fields. Politics & Society, 31(3), 433-464.

Bartley, T. (2007). How foundations shape social movements: The construction of an organizational field and the rise of forest certification. Social Problems, 54(3), 229-255.

Binswanger, H. (1991). Brazilian policies that encourage deforestation in the Amazon. World Development, 19(7), 821-829.

Bitencourt, R. (2011, 9/2). Ranching Surpasses Agriculture for Deforestation in the Amazon, say study (Portuguese). Valor Econômico. Retrieved from http://www.valor.com.br/brasil/997012/pecuaria-supera-agricultura-no-desmatamento-da-amazonia-diz-estudo

Boucher, D., Elias, P., Lininger, K., May-Tobin, C., Roquemore, S., & Saxon, E. (2011). The Root of the Problem: What's Driving Tropical Deforestation Today? Washington D.C.: Union of Concerned Scientists.

Bowman, M. (2012). Does Growth in Demand for Brazilian Beef Exports Drive New Deforestation?, February. New York: American Association of Geographers.

Bowman, M. S., Soares-Filho, B. S., Merry, F. D., Nepstad, D. C., Rodrigues, H., & Almeida, O. T. (2011). Persistence of cattle ranching in the Brazilian Amazon: A spatial analysis of the rationale for beef production. Land Use Policy.

Boyd, W. (2010). Ways of seeing in environmental law: how deforestation became an object of climate governance.

Brazilian Ministry of Science & Technology. (2011). 2011 Budget. Retrieved September 5th, 2011, from http://www.mct.gov.br/index.php/content/view/322842.html

Burney, J., Davis, S., & Lobell, D. (2010). Greenhouse gas mitigation by agricultural intensification. Proceedings of the National Academy of Sciences, 107(26), 12052.

Bustamante, M., Nobre, C., & Smeraldi, R. (2009). Estimating Recent Greenhouse Gas Emissions from Cattle Raising in Brazil. São Paulo, Brazil: Friends of the Earth, Brazilian Amazon.

Bustamante, M., Nobre, C., Smeraldi, R., Aguiar, A., Barioni, L., Ferreira, L., et al. (2012). Estimating greenhouse gas emissions from cattle raising in Brazil. Climatic Change, 1-19.

Butler, R. (2012). 10 African countries to develop satellite-based deforestation tracking systems with help of Brazil. Retrieved July 27th, 2012, from http://news.mongabay.com/2012/0730-african-mrv.html

101

Câmara, G., Valeriano, D., & Soares, J. (2004). Metodologia para o Cálculo da Taxa Anual de Desmatamento na Amazônia Legal. Retrieved 2/26, 2009, from http://www.obt.inpe.br/prodes/metodologia.pdf

Campbell, B. (2009). Beyond Copenhagen: REDD+, agriculture, adaptation strategies and poverty.

Campbell, J. E., Lobell, D. B., & Field, C. B. (2009). Greater Transportation Energy and GHG Offsets from Bioelectricity Than Ethanol. Science, 324(5930), 1055-1057.

Cashore, B. (2002). Legitimacy and the privatization of environmental governance: How non-state market-driven (NSMD) governance systems gain rule-making authority. Governance, 15(4), 503-529.

Cattaneo, A. (2008). Regional comparative advantage, location of agriculture, and deforestation in Brazil. Journal of Sustainable Forestry, 27, 25-42.

Cederberg, C., Persson, U. M., Neovius, K., Molander, S., & Clift, R. (2010). Including Carbon Emissions from Deforestation in the Carbon Footprint of Brazilian Beef. Environmental Science & Technology, null-null.

Cederberg, C., Persson, U. M., Neovius, K., Molander, S., & Clift, R. (2011). Including Carbon Emissions from Deforestation in the Carbon Footprint of Brazilian Beef. Environmental Science & Technology, 45(5), 1773-1779.

Census Bureau of Brazil. (2010). 2006 Agricultural Census. Retrieved 3/27, 2010, from http://www.sidra.ibge.gov.br/bda/tabela/protabl.asp?c=854&z=t&o=23&i=P

Census Bureau of Brazil. (2011). Levantamento Sistemático da Produção Agrícola Fevereiro 2011. Retrieved 3/27, 2011, from http://www.ibge.gov.br/home/estatistica/indicadores/agropecuaria/lspa/lspa_201102_1.shtm

Cerri, C. C., Bernoux, M., Maia, S. M. F., Cerri, C. E. P., Costa Junior, C., Feigl, B. J., et al. (2010). Greenhouse gas mitigation options in Brazil for land-use change, livestock and agriculture. Scientia Agricola, 67(1), 102-116.

Chakravorty, U., Hubert, M. H., & Nostbakken, L. (2009). Fuel versus food. Annu. Rev. Resour. Econ., 1(1), 645-663.

Chomitz, K., & Thomas, T. (2001). Geographic Patterns of Land Use and Land Intensity in the Brazilian Amazon. Washington DC: The World Bank.

Chomitz, K. M. (2002). Baseline, leakage and measurement issues: how do forestry and energy projects compare? Climate Policy, 2(1), 35-49.

102

Cohn, A. (2008). How the measurements matter: The politics of lifecycle assessments of biofuels. Paper presented at the International Studies Association.

Cohn, A. (2009). Full Terrestrial Climate Accounting Affects the Greenhouse Gas Intensity of Steel Produced in Brazil.

Cohn, A., Bowman, M., Zilberman, D., & O'Neill, K. (2011). The viability of cattle ranching intensification in Brazil as a strategy to spare land and mitigate greenhouse gas emissions. Copenhagen, Denmark: CCAFS.

Corbera, E., Estrada, M., & Brown, K. (2010). Reducing greenhouse gas emissions from deforestation and forest degradation in developing countries: revisiting the assumptions. Climatic Change, 100(3), 355-388.

Dauvergne, P. (1993). The politics of deforestation in Indonesia. Pacific Affairs, 497-518.

Davenport, D. S. (2005). An alternative explanation for the failure of the UNCED forest negotiations. Global Environmental Politics, 5(1), 105-130.

de Sa, S. A., Palmer, C., & di Falco, S. (2012). Dynamics of indirect land-use change: empirical evidence from Brazil. London: Centre for Climate Change Economics and Policy.

Deacon, R. T. (1995). Assessing the relationship between government policy and deforestation. Journal of Environmental Economics and Management, 28(1), 1-18.

Dias, F. (2007). Feedlot and Beef Business in Brazil. Paper presented at the Journée Bovine Provinciale.

Dimitrov, R. S. (2005). Hostage to norms: states, institutions and global forest politics. Global Environmental Politics, 5(4), 1-24.

Downie, D. L., Krueger, J., & Selin, H. (2004). Global Policy for Hazardous Chemicals. Global Environmental Policy: Institutions, Law and Policy, 125-145.

Ecofys & Winrock International. (2009). Mitigating Indirect Impacts of Biofuels Production: Case Studies & Methodologies. Retrieved August 24th, 2011, from http://webarchive.nationalarchives.gov.uk/20110407094507/http://www.renewablefuelsagency.gov.uk/sites/rfa/files/_documents/Avoiding_indirect_land-use_change_-_Ecofys_for_RFA.pdf

Eliasch, J. (2008). Climate change: Financing global forests: the Eliasch review: Earthscan/James & James.

Embassy of Brazil. (2010). Brazil's Nationally Appropriate Mitigation Activites. from http://unfccc.int/files/meetings/cop_15/copenhagen_accord/application/pdf/brazilcphaccord_app2.pdf

103

Euclides, V., do Valle, C., Macedo, M., Almeida, R., Montagner, D., & Barbosa, R. (2010). Brazilian scientific progress in pasture research during the first decade of XXI century. Revista Brasileira de Zootecnia, 39, 151-168.

Evans, P. (1997). The eclipse of the state. World Politics, 50(1), 62-87. Fairhead, J., & Leach, M. (1996). Misreading the African Landscape: Society and

Ecology in Forest-Savanna Mosaic. Cambridge (UK): Cambridge Universtiy Press.

Fairhead, J., & Leach, M. (1998). Reframing deforestation: global analyses and local realities with studies in West Africa: Psychology Press.

Faminow, M. (1997). Spatial economics of local demand for cattle products in Amazon development. Agriculture, Ecosystems and Environment, 62(1), 1-11.

Faminow, M. D. (1997). Spatial economics of local demand for cattle products in Amazon development. Agriculture, Ecosystems and Environment, 62, 1-11.

FAOSTAT. (2012). Agriculture Organization of the United Nations. Statistical Database.

Farber, D. A. (2011). Indirect Land Use Change, Uncertainty, and Biofuels Policy. U. ILL. L. REV., 2011, 381.

Fargione, J., Hill, J., Tilman, D., Polasky, S., & Hawthorne, P. (2008). Land Clearing and the Biofuel Carbon Debt. Science, 319(29 February 2008), 1235-1237.

Faria, W., & Almeida, A. (2011). Agricultural Expansion, Openness to Trade and Deforestation at the Brazilian Amazon: A Spatial Econometric Analysis. Paper presented at the European Regional Science Association, Vienna, Austria.

Farrell, A. E., Berkeley, U. C., Sperling, D., Davis, U. C., Arons, S. M., Brandt, A. R., et al. (2007). A Low-Carbon Fuel Standard for California: University of California.

Fearnside, P., & Guimarães, W. (1996). Carbon uptake by secondary forests in Brazilian Amazonia. Forest Ecology and Management, 80(1-3), 35-46.

Fearnside, P. M. (2001). Soybean cultivation as a threat to the environment in Brazil. Environmental Conservation, 28(1), 23-38.

Fearnside, P. M., & Guimaraes, W. M. (1996). Carbon uptake by secondary forests in Brazilian Amazonia. Forest Ecology and Management, 80(1-3), 35-46.

Federal Government of Brazil. (2004). Action Plan for the Prevention of Deforestation in the Amazon (Portuguese). Relatôrio da Presidéncia da República, Casa Civil, Retrieved October, 12, 2009, from http://www.planalto.gov.br/casacivil/desmat.pdf

Federal Government of Brazil. (2008). National Action Plan For Climate Change (PNMC). Retrieved from

104

http://www.mma.gov.br/estruturas/smcq_climaticas/_arquivos/plano_nacional_mudanca_clima.pdf.

Fischer, J., Brosi, B., Daily, G., Ehrlich, P., Goldman, R., Goldstein, J., et al. (2008). Should agricultural policies encourage land sparing or wildlife-friendly farming? Frontiers in Ecology and the Environment, 6(7), 380-385.

Fisher, M., Rao, I., Ayarza, M., Lascano, C., Sanz, J., Thomas, R., et al. (1994). Carbon storage by introduced deep-rooted grasses in the South American savannas.

Forests for People. (2011). Fact Sheet 2011. Retrieved 7/5, 2012, from http://www.un.org/esa/forests/pdf/session_documents/unff9/Fact_Sheet_ForestsandPeople.pdf

G1 News Brazil. (2011). Inpe and Embrapa Reveal the Occupation of Deforested Areas in the Amazon. Retrieved 9/5, 2011, from http://www.anoticiamt.com.br/ver_noticia.asp?cod=120151&canal=brasil

Gan, J., & McCarl, B. A. (2007). Measuring transnational leakage of forest conservation. Ecological Economics, 64(2), 423-432.

Garnett, T. (2009). Livestock-related greenhouse gas emissions: impacts and options for policy makers. Environmental Science and Policy, 12(4), 491-503.

Geist, H., & Lambin, E. (2001). What drives tropical deforestation? A meta-analysis of proximate and underlying causes of deforestation based on subnational case study evidence. LUCC Report Series, 4, 116.

Geist, H. J., & Lambin, E. F. (2002). Proximate causes and underlying driving forces of tropical deforestation. Bioscience, 52(2), 143-150.

Goldsmith, P., & Hirsch, R. (2006). The Brazilian Soybean Complex. Choices: The magazine of food, farm and resource issues, 21(2), 97-105.

Gouvello, C. (2010). Brazil Low Carbon Country Case Study. Washington D.C.: World Bank: Sustainable Development Department of the Latin America and Caribbean Region.

Grainger, A. (1997). Compensating for opportunity costs in forest‚Äêbased global climate change mitigation. Critical reviews in environmental science and technology, 27(S1), 163-176.

Grainger, A. (2010). Uncertainty in the construction of global knowledge of tropical forests. Progress in Physical Geography, 34(6), 811-844.

Green, R., Cornell, S., Scharlemann, J., & Balmford, A. (2005). Farming and the Fate of Wild Nature. Science, 307(5709), 550-555.

Gulbrandsen, L. H. (2004). Overlapping public and private governance: Can forest certification fill the gaps in the global forest regime? Global Environmental Politics, 4(2), 75-99.

105

Gullison, R. E., & Losos, E. C. (1993). The role of foreign debt in deforestation in Latin America. Conservation Biology, 7(1), 140-147.

Haas, P. M. (1990). Saving the Mediterranean: the politics of international environmental cooperation. New York, New York: Columbia University Press.

Hansen, S. (1989). Debt for nature swaps‚ An overview and discussion of key issues. Ecological Economics, 1(1), 77-93.

Hargrave, J., & Kis-Katos, K. (2011). Economic Causes of Deforestation in the Brazilian Amazon.

Hasenclever, A., Mayer, P., & Rittberger, V. (1997). Theories of international regimes. New York, New York: Cambridge University Press.

Havlik, P., Schneider, U. A., Schmid, E., Boettcher, H., Fritz, S., Skalsky, R., et al. (2011). Global land-use implications of first and second generation biofuel targets. Energy Policy, 39(10), 5690-5702.

Hecht, S. (1985). Environment, development and politics: Capital accumulation and the livestock sector in Eastern Amazonia. World Development, 13(6), 663-684.

Hecht, S. (1993). The logic of livestock and deforestation in Amazonia. BioScience, 43(10), 697-695.

Hecht, S. B. (1982). Cattle ranching development in the Eastern Amazon: evaluation of a development policy. University of California, Berkeley.

Herold, M., Verchot, L., Angelsen, A., Maniatis, D., & Bauch, S. (2012). A step-wise framework for setting REDD+ forest reference emission levels and forest reference levels (pp. 8p). Bogor, Indonesia: Center for International Forestry Research (CIFOR).

Herrero, M., Thornton, P., Kruska, R., & Reid, R. (2008). Systems dynamics and the spatial distribution of methane emissions from African domestic ruminants to 2030. Agriculture, Ecosystems & Environment, 126(1), 122-137.

Hertel, T. W. (2011). The global supply and demand for agricultural land in 2050: A perfect storm in the making? American Journal of Agricultural Economics, 93(2), 259-275.

Hochman, E., & Zilberman, D. (1986). Optimal strategies of development processes of frontier environments* 1. The Science of the Total Environment, 55, 111-119.

Hubacek, K., & Van Den Bergh, J. (2006). Changing concepts of land in economic theory: From single to multi-disciplinary approaches. Ecological Economics, 56(1), 5-27.

Hummel, H. (2006). Interpreting Energy Technology & Policy Implications of Climate Stabilization Scenarios. Stanford, Stanford, California.

106

Humphreys, D. (2008). The politics of'Avoided Deforestation': historical context and contemporary issues. International Forestry Review, 10(3), 433-442.

Izaurralde, R., Williams, J. R., McGill, W. B., Rosenberg, N. J., & Jakas, M. (2006). Simulating soil C dynamics with EPIC: Model description and testing against long-term data. Ecological Modelling, 192(3), 362-384.

Jack, K. (2009). Barriers to the adoption of agricultural technologies in developing countries. Agricultural Technology Adoption Initiative. JPAL (MIT), CEGA (Berkeley), White paper, Funding round, 1.

Jarvis, L. S. (1974). Cattle as capital goods and ranchers as portfolio managers: an application to the Argentine cattle sector. The Journal of Political Economy, 82(3), 489-520.

Jasanoff, S. (2001). Image and imagination: the formation of global environmental consciousness. Changing the atmosphere: Expert knowledge and environmental governance, 309-337.

Just, R. E., & Zilberman, D. (1983). Stochastic structure, farm size and technology adoption in developing agriculture. Oxford Economic Papers, 35(2), 307-328.

Kahn, J. R., & McDonald, J. A. (1995). Third-world debt and tropical deforestation. Ecological Economics, 12(2), 107-123.

Kaimowitz, D., & Angelsen, A. (2008). Will livestock intensification help save Latin America's Tropical Forest? Journal Title: Journal of Sustainable Forestry, 27(1-2), 6-24.

Kaimowitz, D., Mertens, B., Wunder, S., & Pacheco, P. (2004). Hamburger connection fuels Amazon destruction. Technical report. Retrieved 3/21/06, 2006, from http://www.cifor.cgiar.org/publications/pdf

Kanowski, P. J., McDermott, C. L., & Cashore, B. W. (2011). Implementing REDD+: lessons from analysis of forest governance. Environmental Science & Policy, 14(2), 111-117.

Khanna, M., & Crago, C. L. (2012). Measuring Indirect Land Use Change with Biofuels: Implications for Policy. Annual Review of Resource Economics, 4(1), null.

Kindermann, G. E., McCallum, I., Fritz, S., & Obersteiner, M. (2008). A global forest growing stock, biomass and carbon map based on FAO statistics. Silva Fennica, 42(3), 387.

King, V. T. (1993). Politik pembangunan: The political economy of rainforest exploitation and development in Sarawak, East Malaysia. Global Ecology and Biogeography Letters, 235-244.

Krasner, S. D. (1983). International regimes. Ithaca, New York: Cornell Univ Press. Lapola, D. M., Schaldach, R., Alcamo, J., Bondeau, A., Koch, J., Koelking, C., et al.

(2010). Indirect land-use changes can overcome carbon savings from

107

biofuels in Brazil. Proceedings of the National Academy of Sciences, 107(8), 3388.

Laurance, W. F. (2008). Can carbon trading save vanishing forests. Bioscience, 58(4), 286-287.

Lee, D. R., & Barrett, C. B. (2001). Tradeoffs or synergies?: agricultural intensification, economic development, and the environment. New York, NY: CABI Publishing.

Lemoine, D., Plevin, R., Cohn, A., Jones, A., Brandt, A., Vergara, S., et al. (2010). The Climate Impacts of Bioenergy Systems Depend on Market and Regulatory Policy Contexts. Environmental Science & Technology, 44(19), 7347-7350.

Lubowski, R. (2003). Determinants of land-use transitions in the United States: econometric estimation of a Markov model. Economic Research Service of the USDA.

Lutz, E., Vedova, M., Martinez, H., San Roman, L., Velasquez, R., Alvarado, A., et al. (1993). Interdisciplinary fact-finding on current deforestation in Costa Rica: World Bank, Environment Department.

Macedo, M. N., DeFries, R. S., Morton, D. C., Stickler, C. M., Galford, G. L., & Shimabukuro, Y. E. (2012). Decoupling of deforestation and soy production in the southern Amazon during the late 2000s. Proceedings of the National Academy of Sciences, 109(4), 1341-1346.

Mahar, D. J. (1989). Government policies and deforestation in Brazil's Amazon region: International Bank for Reconstruction and Development, Washington, DC (USA).

Malafaia, P., Peixoto, P. V., Gonçalves, J. C. S., Moreira, A. L., Costa, D., & Correa, W. S. (2004). Weight Gain and Costs of Beef Cattle Fed Two Types of Mineral Supplements (Portuguese). Pesq. Vet. Bras, 24(3), 46-50.

Manzatto, C., Assad, E., Bacca, J., Zaroni, M., & Pereira, S. (2009). Agroecological Zoning of Sugarcane: To expand production, preserve life and guarantee the future (Portuguese). Documento 110 Embrapa Solos.

Margulis, S. (2004). Causes of deforestation of the Brazilian Amazon. Washington D.C.: World Bank Publications.

Martha, G. B., Alves, E., & Contini, E. (2012). Land-saving approaches and beef production growth in Brazil. Agricultural Systems.

Matson, P., Parton, W., Power, A., & Swift, M. (1997). Agricultural intensification and ecosystem properties. Science, 277(5325), 504.

Matson, P. A., & Vitousek, P. M. (2006). Agricultural intensification: will land spared from farming be land spared for nature? Conservation Biology, 20(3), 709-710.

Mayr, E. (1961). Cause and effect in biology. Science, 134(3489), 1501-1506.

108

McAlpine, C., Etter, A., Fearnside, P., Seabrook, L., & Laurance, W. (2009). Increasing world consumption of beef as a driver of regional and global change: A call for policy action based on evidence from Queensland (Australia), Colombia and Brazil. Global Environmental Change, 19(1), 21-33.

McDermott, C. L., O'Carroll, A., & Wood, P. (2007). International Forest Policy: The instruments, agreements and processes that shape it. New York, New York: Department of Economic and Social Affairs: United Nations Forum on Forests Secretariat.

McKinsey & Co. (2009). Pathways to a Low Carbon Economy for Brazil. Retrieved August 24th, 2011, from http://www.mckinsey.com/en/Client_Service/Sustainability/Latest_thinking/~/media/McKinsey/dotcom/client_service/Sustainability/cost curve PDFs/pathways_low_carbon_economy_brazil.ashx

Millen, D. D., Pacheco, R. D. L., Meyer, P. M., Rodrigues, P. H. M., & Arrigoni, M. D. B. (2011). Current outlook and future perspectives of beef production in Brazil. Animal Frontiers, 1(2), 46-52.

Murray, M. P. (2006). Avoiding invalid instruments and coping with weak instruments. The journal of economic perspectives, 20(4), 111-132.

National Institute of Space Studies of Brazil (INPE), & Brazilian Agricultural Research Corporation (EMBRAPA). (2011). Terraclass: Study of Land Use and Land Cover in the Amazon Biome (Executive Summary). Sao Jose dos Campos, Sao Paulo Brazil: National Institute of Space Studies of Brazil (INPE).

Naylor, R., Steinfeld, H., Falcon, W., Galloway, J., Smil, V., Bradford, E., et al. (2005). AGRICULTURE: Losing the Links Between Livestock and Land. Science, 310(5754), 1621-1622.

Nelson, A. (2008). Travel time to major cities: A global map of Accessibility. . from http://gem.jrc.ec.europa.eu/

Nepstad, D., Soares-Filho, B., Merry, F., Lima, A., Moutinho, P., Carter, J., et al. (2009). The End of Deforestaton in the Brazilian Amazon. Science, 326(5928), 1350-1351.

O'Hare, M., & Mundel, D. S. (1983). When to Pay for Sunk Benefits. In R. J. Zeckhauser & D. Leebaert (Eds.), What role for government?: lessons from policy research (pp. 255-261). Durham North Carolina: Duke Univ Press.

O'Neill, K. (2009). The environment and international relations. New York, New York: Cambridge University Press.

Ojima, D., Galvin, K., & Turner, B. (1994). The global impact of land-use change. Bioscience, 44(5), 300-304.

109

Okereke, C., & Dooley, K. (2010). Principles of justice in proposals and policy approaches to avoided deforestation: Towards a post-Kyoto climate agreement. Global Environmental Change, 20(1), 82-95.

Oliveira, O., Bolliger, F., & Florido, A. (2006). Brazil Agricultural Census 2006: Innovations and Impacts

. Paper presented at the Fourth International Conference on Agricultural Statisticss, Beijing, China.

Oxford English Dictionary. (2012a). "deforest, v.". from http://oed.com/view/Entry/153562?rskey=aaE2mv&result=1

Oxford English Dictionary. (2012b). "proximate, adj.". from http://oed.com/view/Entry/153562?rskey=aaE2mv&result=1

Pacheco, P., & Poccard-Chapuis, R. (2012). The Complex Evolution of Cattle Ranching Development Amid Market Integration and Policy Shifts in the Brazilian Amazon. Annals of the Association of American Geographers.

Parker, C., Mitchell, A., Trivedi, M., & Mardas, N. (2008). The little REDD book: a guide to governmental and non-governmental proposals for reducing emissions from deforestation and degradation. Retrieved July 25, 2012, from http://www.theredddesk.org

Persha, L., Agrawal, A., & Chhatre, A. (2011). Social and ecological synergy: Local rulemaking, forest livelihoods, and biodiversity conservation. Science, 331(6024), 1606.

Pfaff, A. (1999). What Drives Deforestation in the Brazilian Amazon? Evidence from Satellite and Socioeconomic Data. Journal of Environmental Economics and Management, 37(1), 26-43.

Pfaff, A., Robalino, J., Walker, R., Aldrich, S., Caldas, M., Reis, E., et al. (2007). ROAD INVESTMENTS, SPATIAL SPILLOVERS, AND DEFORESTATION IN THE BRAZILIAN AMAZON*. Journal of Regional Science, 47(1), 109-123.

Pfaff, A. S. P. (1996). What Drives Deforestation in the Brazilian Amazon? : MIT Joint Program on the Science and Policy of Global Change.

Phelps, J., Webb, E. L., & Agrawal, A. (2010). Does REDD+ threaten to recentralize forest governance? Science, 328(5976), 312-313.

Pinjuv, G. (2011). Gigaton Analysis of the Livestock Industry: The Case For the Adoption of a Moderate Intensification Model. Retrieved 9/7, 2011, from http://www.carbonwarroom.com/livestock-register

Rajan, S. R. (2006). Modernizing nature: forestry and imperial eco-development 1800-1950: Oxford University Press, USA.

Ramankutty, N., Gibbs, H., Achard, F., DeFries, R., Foley, J., & Houghton, R. (2007). Challenges to estimating carbon emissions from tropical deforestation. Global Change Biology, 13(1), 51-66.

110

Reilly, J., Melillo, J., Cai, Y., Kicklighter, D., Gurgel, A., Paltsev, S., et al. (2012). Using Land To Mitigate Climate Change: Hitting the Target, Recognizing the Trade-offs. Environmental Science & Technology.

Renewable Fuels Agency. (2008). The Gallagher Review of the Indirect Effects of Biofuels. London.

Repetto, R., & Gillis, M. (1988). Public policy and the misuse of forest resources. Ricardo, D. (1817). Principles of political economy. Cambridge: Cambridge. Ricardo, D. (1891). Principles of political economy and taxation. London, UK: G.

Bell and Sons. Richards, P. D., Myers, R. J., Swinton, S. M., & Walker, R. T. (2012). Exchange

rates, soybean supply response, and deforestation in South America. Global Environmental Change, 22(2), 454-462.

Robinson, B. E., Holland, M. B., & Naughton-Treves, L. (2011). Does secure land tenure save forests? A review of the relationship between land tenure and tropical deforestation.

Rudel, T. K. (1983). Roads, speculators, and colonization in the Ecuadorian Amazon. Human Ecology, 11(4), 385-403.

Rudel, T. K. (2007). Changing agents of deforestation: from state-initiated to enterprise driven processes, 1970-2000. Land Use Policy, 24(1), 35-41.

Rudel, T. K., Schneider, L., Uriarte, M., Turner, B., DeFries, R., Lawrence, D., et al. (2009). Agricultural intensification and changes in cultivated areas, 1970-2005. Proceedings of the National Academy of Sciences, 106(49), 20675.

Sajwaj, T., Harley, M., & Parker, C. (2008). ELIASCH REVIEW: FOREST MANAGEMENT IMPACTS ON ECOSYSTEM SERVICES: AEA, Didcot. New work commissioned for the Eliasch Review.

Sampaio, A. A. M., de Brito, R. M., de Aguiar, L. L. M., Junior, P. R., da Cruz, G. M., Alencar, M. M., et al. (2001). Comparison of Sistems of Evaluation of Diets in the Intensive Beef Production Model: Supplemental Feeding During the Dry Season (Portuguese) Revista Brasileira de Zootecnia, 30(4), 1287-1292.

Santilli, M., Moutinho, P., Schwartzman, S., Nepstad, D., Curran, L., & Nobre, C. (2003). Tropical Deforestation and the Kyoto Protocol: a new proposal, December (Vol. Oral). Milan: 9th Conference of the Parties.

Santilli, M., Moutinho, P., Schwartzman, S., Nepstad, D., Curran, L., & Nobre, C. (2005). Tropical Deforestation and the Kyoto Protocol. Climatic Change, 71(3), 267-276.

Schama, S. (1996). Landscape and memory (Vol. 101): Vintage Books USA. Schlamadinger, B., Ciccarese, L., Dutschke, M., Fearnside, P. M., Brown, S., &

Murdiyarso, D. (2005). Should we include avoidance of deforestation in the international response to climate change? In D. Murdiyarso & H. Herewati

111

(Eds.), Carbon forestry: who will benefit (pp. 26-41). Bogor, Indonesia: Center for International Forestry Research.

Schlamadinger, B., Johns, T., Ciccarese, L., Braun, M., Sato, A., Senyaz, A., et al. (2007). Options for including land use in a climate agreement post-2012: improving the Kyoto Protocol approach. Environmental Science & Policy, 10(4), 295-305.

Schneider, U. A., Havl√≠k, P., Schmid, E., Valin, H., Mosnier, A., Obersteiner, M., et al. (2011). Impacts of population growth, economic development, and technical change on global food production and consumption. Agricultural Systems, 104(2), 204-215.

Searchinger, T. (2007). Factoring Land Use Change into the Greenhouse Gas Effects of Biofuels. Paper presented at the The Intersection of Energy and Agriculture: Implications of Biofuels and the Search for a Fuel of the Future, UC Berkeley.

Searchinger, T., & Amaral, L. (2009). Assuring Environmental Benefits from Sugarcane Ethanol in Brazil-One Approach (unpublished). German Marshall Fund.

Searchinger, T., Heimlich, R., Houghton, R. A., Dong, F., Elobeid, A., Fabiosa, J., et al. (2008). Use of US Croplands for Biofuels Increases Greenhouse Gases Through Emissions from Land-Use Change. Science, 319(5867), 1238.

Segerson, K. (2006). Theoretical Background. In K. P. Bell, K. J. Boyle & J. Rubin (Eds.), Economics of rural land-use change. Burlington, VT: Ashgate Pub Co.

Seixas, L. (2012). June 11th. Coffee with the President Sere, C., & Steinfeld, H. (1996). World livestock production systems: current status,

issues and trends. Rome: FAO. Serrão, E., & Toledo, J. (1990). The search for sustainability in Amazonian pastures.

In A. Andersen (Ed.), Alternatives to deforestation: steps toward sustainable use of the Amazon rain forest. (pp. 195-214). New York: Columbia University Press.

Seto, K. C., Reenberg, A., Boone, C. G., Fragkias, M., Haase, D., Langanke, T., et al. (2012). Urban land teleconnections and sustainability. Proceedings of the National Academy of Sciences, 109(20), 7687-7692.

Skalsky, R. (2008). Geo-bene global database for bio-physical modeling v.1.0. Concepts,methodologies and data. Laxenburg, Austria: International Institute for Applied Systems Analysis.

Sohngen, B., & Brown, S. (2004). Measuring leakage from carbon projects in open economies: a stop timber harvesting project in Bolivia as a case study. Canadian Journal of Forest Research, 34(4), 829-839.

112

Somwaru, A., & Valdes, C. (2004). Brazil's beef production and its efficiency: A comparative study of scale economies, The World Bank, Washington, D.C., USA.

The Soy Moratorium Will Be Renewed to Combat Deforestation in the Amazon (Portuguese). (2011, 5/25). Globo Rural Online. Retrieved from http://revistagloborural.globo.com/Revista/Common/0,,EMI236052-18095,00-MORATORIA+DA+SOJA+SERA+REPACTUADA+PARA+ENFRENTAR+O+DESMATAMENTO+NA+AMAZONIA.html

Stabile, M. C. C., Nepstad, D., & Azevedo, A. (2012). Brazil's "Low Carbon Agriculture Program": Barriers to Implementation. Brasilia, Brazil: Amazon Environmetnal Research Institute.

Stavins, R. N. (1997). Policy instruments for climate change: how can national governments address a global problem. U. Chi. Legal F., 293.

Steinfeld, H., Gerber, P., Wassenaar, T., Castel, V., & de Haan, C. (2006). Livestock's long shadow: environmental issues and options: FAO.

Steinfeld, H., Gerber, P., Wassenaar, T., Castel, V., Rosales, M., & De Haan, C. (2006). Livestock's long shadow. Rome: UN Food and Agriculture Organization.

Stephan, B. (2012). Bringing discourse to the market: the commodification of avoided deforestation. Environmental Politics, 21(4), 621-639.

Stephan, B., & Paterson, M. (2012). The politics of carbon markets: an introduction. Environmental Politics, 21(4), 545-562.

Stern, N. H., Peters, S., Bakhshi, V., Bowen, A., Cameron, C., Catovsky, S., et al. (2006). Stern Review: The economics of climate change. London: Her Majesty's Treasury.

Sukhdev, P. (2009). Costing the earth. Nature, 462(7271), 277-277. Thornton, P. K., & Herrero, M. (2010). Potential for reduced methane and carbon

dioxide emissions from livestock and pasture management in the tropics. Proceedings of the National Academy of Sciences, 107(46), 19667.

Tscharntke, T., Klein, A. M., Kruess, A., Steffan-Dewenter, I., & Thies, C. (2005). Landscape perspectives on agricultural intensification and biodiversity‚Äìecosystem service management. Ecology Letters, 8(8), 857-874.

Turner II, B., & Robbins, P. (2008). Land-change science and political ecology: Similarities, differences, and implications for sustainability science. Annual Review of Environment and Resources, 33, 295-316.

Turner II, B. L. (1989). Human Causes of Global Change. In R. S. DeFries & T. F. Malone (Eds.), Global change and our common future: papers from a forum. Washington: National Academies Press.

113

Turner II, B. L., Kasperson, R. E., Meyer, W. B., Dow, K. M., Golding, D., Kasperson, J. X., et al. (1990). Two types of global environmental change:: Definitional and spatial-scale issues in their human dimensions. Global Environmental Change, 1(1), 14-22.

UNfairplay. (2011). Leveling the Playing Field: A Report to the UNFCCC on Negotiating Capacity & Access to Information.

United Nations Framework Convention on Climate Change. (2011). Cancun Decision | Section C | Paragraph 68. Bonn, Germany.

Energy Independence and Security Act of 2007, 42 U.S.C. 17001 et seq. (2007). Vadjunec, J., & Brannstrom, C. (2012). Synergies and Divergences between Land

Change Science and Political Ecology, February. New York, New York: American Association of Geographers.

Vidal, J., & Goldenberg, S. (2010). Cancun Climate Change Summit: A deal is reached. Guardian UK. Retrieved from http://www.guardian.co.uk/environment/2010/dec/11/cancun-climate-change-summit-deal

von Thunen, J. (1966). Von Thunen's' Isolated State': An English Edition: Pergamon. Von Thunen, J. H., & Hall, P. G. (1966). Isolated state: an English edition of Der

isolierte Staat: Pergamon Press. Vosti, S., Carpentier, C., Witcover, J., & Valentim, J. (2001). Systems in the Western

Brazilian Amazon. Agricultural technologies and tropical deforestation, 127. Walker, N., Bramble, B., & Patel, S. (2010). From Major Driver of Deforestation

and Greenhouse Gas Emissions to Forest Guardians? New Developments in Brazil’s Amazon Cattle Industry.

Retrieved August 24, 2011, from http://www.nwf.org/Global-Warming/Policy-Solutions/~/media/4878226C49BF48EB9EB54C1B7C616327.ashx

Walker, N., Patel, S., & Kalif, K. (Forthcoming). From Amazon Pasture to the High Street: Deforestation and the Brazilian Cattle Product Supply Chain. Tropical Conservation Science.

Walker, R., Browder, J., Arima, E., Simmons, C., Pereira, R., Caldas, M., et al. (2009). Ranching and the new global range: Amazonia in the 21st century. Geoforum, 40(5), 732-745.

Walsh, J. (1987). Bolivia swaps debt for conservation. Science, 237(4815), 596-597.

Wassenaar, T., Gerber, P., Verburg, P., Rosales, M., Ibrahim, M., & Steinfeld, H. (2007). Projecting land use changes in the Neotropics: The geography of pasture expansion into forest. Global Environmental Change, 17(1), 86-104.

Weber, C., & Matthews, H. (2008). Food-miles and the relative climate impacts of food choices in the United States. Environ. Sci. Technol, 42(10), 3508-3513.

114

White, A., & Martin, A. (2002). Who owns the world's forests? Forest tenure and public forests in transition: Washington, DC, USA.

Wise, M., Calvin, K., Thomson, A., Clarke, L., Bond-Lamberty, B., Sands, R., et al. (2009). Implications of Limiting CO2 Concentrations for Land Use and Energy: Supporting Online Material. Science, 324(5931), 1183.

Zaks, D., Barford, C., Ramankutty, N., & Foley, J. (2009). Producer and consumer responsibility for greenhouse gas emissions from agricultural production perspective from the Brazilian Amazon. Environmental Research Letters, 4, 1-12.

Zilberman, D., Schmitz, A., Casterline, G., Lichtenberg, E., & Siebert, J. (1991). The economics of pesticide use and regulation. Science, 253(5019), 518.

Zwick, S. (2010). Insiders say REDD+ Work Plan was Edited at Last Minute. Retrieved 3/20, 2012, from http://www.ecosystemmarketplace.com/pages/dynamic/article.page.php?page_id=7873&section=news_arti

Examining and Supporting Agricultural Interventions to Reduce Deforestation … · 2018. 10. 10. · from deforestation and forest degradation in developing countries.” The second - [PDF Document] (2024)

References