Framing Climate Change Loss and Damage in UNFCCC Negotiations

How does an idea emerge and gain traction in the international arena when its underpinning principles are contested by powerful players? The adoption in 2013 of the Warsaw International Mechanism on Loss and Damage as part of the United Nations Framework Convention on Climate Change (UNFCCC) puzzled observers, because key state parties, such as the United States, had historically opposed the policy. This article examines the roles of frame contestation and ambiguity in accounting for the evolution and institutionalization of the “loss and damage” norm within the UNFCCC. The article applies frame analysis to the data from coverage of the negotiations and elite interviews. It reveals that two competing framings, one focused on liability and compensation and the other on risk and insurance, evolved into a single, overarching master frame. This more ambiguous framing allowed parties to attach different meanings to the policy that led to the resolution of differences among the parties and the embedding of the idea of loss and damage in international climate policy.

them (Moore 2012;Okereke 2008). Their negotiators have used a variety of tactics to avoid discussion and to scupper policy initiatives on the issue. In many ways, the contestation over loss and damage is part of a larger picture of enduring distrust between the developed and developing state parties in global environmental governance (Bernstein 2013).
In light of this history, one specific outcome of the 19th Conference of the Parties (COP 19) was particularly surprising. The establishment of the Warsaw International Mechanism on Loss and Damage Associated with Climate Change Impacts (WIM) in 2013 created an institutionalized policy space to address the adverse consequences of climate change (UNFCCC 2013). The establishment of the WIM was heralded as a major victory for developing nations that are particularly vulnerable to climate change impacts and their NGO allies. How was the agreement to establish this mechanism reached in the face of strong initial resistance from key developed states?
We argue that the answer lies in the ambiguity of the idea of loss and damage and the way in which that ambiguity was constructed. From 2008 onward, an overarching and ambiguous "loss and damage" frame began to replace two more specific historical framings-a "liability and compensation" frame and a "risk management and insurance" frame-in the discussions. The rhetoric of the older framings tended to stoke conflicts over culpability for greenhouse gas emissions and the appropriate realm in which climate-related harm should be addressed. The overarching master frame that replaced these two framings obscured these divisions, which allowed for a consensus to emerge in 2013. "Loss and damage" was an amorphous concept to which policy actors attached different meanings.
For the Group of 77 developing states (G77) and their NGO allies, the establishment of the WIM marked a profound institutional paradigm shift in the history of the UNFCCC. Driven by a global justice imperative, the negotiators for these states tended to argue that the idea of loss and damage is underpinned by an understanding of liability, and that compensation is one part of an appropriate policy response. They also asserted that, rather than focusing solely on the mitigation of emissions and adaptation to climate change effects, the UNFCCC is also the appropriate forum in which to consider ways of addressing inevitable damage. For them, addressing loss and damage is understood as something "beyond adaptation." For most industrialized states, the loss-and-damage agenda is much narrower. When they have addressed the idea of loss and damage, these actors have offered technocratic policy solutions and institutional prescriptions that minimize questions of responsibility and compensation. For them, if loss and damage is to be dealt with as part of the UNFCCC process at all, it should be dealt with as part of the adaptation framework rather than as a separate policy track.
This article builds on research that has explored why some frames gain traction and may result in policy change, while others do not (Hjerpe and Buhr 2014;Moore 2012). The loss-and-damage debate sheds light on the ways in which framing processes-how actors define a problem and its potential solutions-shape international environmental policy. Constructivist scholars have now established that norms matter in accounting for state behavior in international negotiations, but more research is needed on how the evolution of ideas over time-specifically, the role of frame contestation among different parties-shapes policy outcomes. We argue that, under certain circumstances, framing contestation over time may lead to the emergence of a master frame whose content is more ambiguous than the preceding framings. Following the literature on constructive ambiguity, we find that a more indistinct framing and the postponement of decisions about the specific content of a policy can increase the likelihood of managing and resolving differences during negotiations (Fischhendler 2008). However, we also find that successfully embedding a frame in the way an issue is discussed does not necessarily address material injustices.
This article proceeds in three sections. First, we elaborate on theoretical discussions concerning the role of framing in international environmental politics. We explain why an issue-framing perspective offers an important complement to existing approaches that seek to account for the outcomes of global environmental negotiations, and we outline our methodological approach. Second, we document the emergence and evolution of the loss-and-damage frame in climate change negotiations. Third, we show how the issue of loss and damage has been contested over time, and how this has shaped current understandings of the concept and influenced where the policy is housed in the international climate regime.

A Framing Approach: Theory and Methods
The research here relies on frame theory to explain the institutionalization of the idea of "loss and damage" in the international climate change regime. Building on Erving Goffman's (1974) concept of frames as "schemata of interpretation," Snow and Byrd argue that the framing perspective views actors not merely as promoters of existing ideas and meanings, but as "signifying agents actively engaged in producing and maintaining meaning for constituents, antagonists and bystanders" (Snow and Byrd 2007, 123). This matters for international policymaking because framing concerns are, according to Mitchell, "efforts to define a problem, its causes, and its potential solutions in ways that are calculated to gain support for the position of the actor doing the framing" (Mitchell 2010, 97). Framing processes shape which issues are seen as problems, which are discussed, and which are taken up for action. Benford and Snow (2000) identify several core framing tasks that agents practice, including diagnostic framing (identifying the problem and assigning blame) and prognostic framing (the articulation of a proposed solution to the problem). These processes allow agents to critique existing institutional arrangements and the patterns of distribution that they enshrine. Given the potential power of such processes, O'Neill and co-authors call for scholars to take problem-framing more seriously and to look at who does the framing at all stages of the policy process (O'Neill et al. 2004).
A frame analysis approach has two key benefits. First, it allows scholars to understand the often implicit, ideational drivers of policy change. Béland (2009) argues that to understand policy-making in international institutions, it is necessary to examine the framing strategies of key actors to understand how they convince other groups to support their policy alternatives. Failing to analyze the changing assumptions of actors as they affect the formulation and diffusion of new policy proposals makes it harder for scholars to understand the potential content and direction of policy change. A second advantage of a framing approach is that it recognizes the intersubjectivity of norms. This approach explicitly acknowledges that the meanings of norms are neither given nor understood similarly by all actors, but rather are subject to interpretation (Towns 2012). The process of resolving the conflicts over interpretation or of allowing parties to interpret a frame in their own way can help explain how an issue is placed on the agenda and ultimately recognized in international law.
To explore the influence of framing processes on policy outcomes, this article relies on what Beach and Pedersen (2013) refer to as "explaining outcome process-tracing." In this process-tracing approach, we craft an explanation of the outcome in question-the establishment of a loss-and-damage mechanism despite strong initial resistance from key states-by looking at ideational effects over time. Process-tracing the influence of ideas is a challenging endeavor, so we adopt a three-pronged methodology that has the benefit of looking beyond just the statements of a handful of elite actors who were "at the table" at key moments ( Jacobs 2015). This involves a content analysis, a historical mapping exercise, and a frame analysis. If frames mattered in the adoption of a decision on loss and damage, we should find their "fingerprints" in elite statements and behavior at critical junctures in the negotiations, but also in the broader patterns of continuity and change and in the interactions between actors over time (Jacobs 2015).
First, we undertook a qualitative content analysis through a systematic coding of two publications from 2003 to 2013: ECO, a newsletter published by the Climate Action Network, and Earth Negotiations Bulletin (ENB). These publications provide comprehensive daily coverage of the official UNFCCC negotiations. The former takes an NGO advocacy tone, and the latter offers a more neutral take on the proceedings, summarizing the day's discussions and the decisions adopted. Following Hjerpe and Buhr (2014), we believe that to study an evolving debate over decades calls for an open-ended approach that allows for the incorporation of changed meanings into the analysis. We relied on an iterative research approach and inductively generated codes from the text, compared our findings, and then returned to the text with a loose coding scheme. We identified three distinct framings: "insurance" and "risk transfer" constituted one category; mentions of "compensation," "liability," "damages," "restitution," "repayment of climate debt," and "tort" constituted a second; and mentions of "loss and damage" constituted a third. The context within which each term was used in the text was examined to ensure that it was of relevance to the subject and placed in the appropriate category. This broad picture of the changing way the issue was talked about provided an entry point into an analysis of the latent meanings behind these terms.
Second, we constructed a historical account of the emergence and evolution of the idea of loss and damage. We reviewed loss-and-damage policy documents, including articles in ECO and ENB, submissions by the state parties, NGO reports and UNFCCC reports, and summaries of meetings and decisions. This analysis is complemented by the data from thirteen semistructured elite interviews with negotiators, legal advisors, and NGO officers. These were conducted between June and August 2013 (before the adoption of the WIM) and in August and September 2015. We identified interviewees from the primary documents and through snowball sampling and offered them anonymity to encourage openness. Interviews were done in person and over the phone and lasted between 30 and 90 minutes.
Third, we conducted a frame analysis that involved identifying the diagnostic and prognostic frames around loss and damage. We coded the interview transcripts and key texts gathered as part of the historical analysis to unpack the meanings for different actors at particular points in time, and to understand the political and institutional implications of particular frames. By situating this frame analysis within a macro-historical perspective, we are able to offer insights into the subtle changes in interactions between actors over time and in the ways that the idea of loss and damage is understood and contested.

The Evolution of the Framing of Loss and Damage
The idea that the global community needed to address the material losses resulting from climate change emerged as early as 1991 during the negotiations to establish the UNFCCC. In a submission on behalf of the Alliance of Small Island States (AOSIS), Vanuatu proposed an international insurance pool to compensate small island developing states (SIDS) for the damages incurred as a result of rising sea levels ( Vanuatu 1991). One of the motivations behind tabling the proposal was to highlight the costs of climate impacts, to drive mitigation ambition (Actionaid et al. 2012). The Convention text included insurance as a possible solution for meeting the specific needs of developing countries (United Nations 1992, 8).
The idea of insurance was raised again at COP 7 in Marrakesh in 2001 (UNFCCC 2002). This was followed by two UNFCCC workshops in 2003 exploring insurance-related action. In 2004, at COP 10 in Buenos Aires, there were further calls for expert meetings on insurance and risk assessments of the impacts of climate change. At COP 11, in Montreal in 2005, Bangladesh, for the least developed countries (LDC) group, called for compensation for damages caused by climate change, but the issue failed to gain any traction. 1 Around this time the OPEC states began arguing that the adverse impacts of measures responding to the threat of climate change-known as "response measures"-are comparable to the adverse impacts of climate change. These states claimed that if there is a general effort to reduce the consumption of fossil fuels, oil-producing countries should be compensated for their loss in revenues. 2 In 2002 Saudi Arabia insisted on tying adaptation funding for the impacts of climate change to "compensation" for any losses in oil revenues that might occur due to mitigation measures. 3 In 2007/2008, the discussion of loss and damage took off. First, AOSIS (2008) proposed a mechanism to address loss and damage, consisting of three interdependent tracks: an insurance element, to address climate-related extreme weather events and risks to food security and livelihoods; a rehabilitation and compensation part, to address the progressive negative impacts of climate change such as sea-level rise; and a risk management component. The proposal also suggested that developed countries pay most of the insurance premiums (Actionaid 2010).
In 2008, nonstate actors also began advocating on the issue in a concerted manner. WWF published a catalytic report entitled Beyond Adaptation: The Legal Duty to Pay Compensation for Climate Change Damage. The authors called for a compensation mechanism under the UNFCCC and stressed that the mitigation and adaptation tracks were not adequate to address the inevitable consequences of climate change ( Verheyen and Roderick 2008). Germanwatch also published several discussion papers advocating an international insurance mechanism, but at that time argued that this could be part of the adaptation work stream (e.g., Harmeling 2008;Linnerooth-Bayer et al. 2008). Germanwatch arranged a joint workshop with Munich Climate Insurance Initiative (MCII), an interest group launched by the large, multinational private insurance group Munich Re, for UNFCCC actors on insurance solutions to climate change (MCII et al. 2008).
At this point some developed countries, uncomfortable with the direction of discussions on this issue, attempted to move the section on risk management into other policy streams, and eventually sought to cut off discussions about loss and damage to avoid debate of proposals related to compensation (Warner and Zakieldeen 2011).
However, the next several COPs led to increased attention and institutional contestation over the issue of loss and damage. In 2009, Tuvalu submitted a proposal for a compensation mechanism, but developed countries refused to engage. At COP 15 in Copenhagen later that year, developing countries, led by AOSIS, tabled a proposal for a loss-and-damage mechanism in the adaptation text that was intended to be less controversial in terms of attribution of culpability, but this was also removed before final negotiations, largely at the prompting of the US and EU (Moore 2012).
A divide along developed/developing party lines emerged. At the 2010 climate meetings in Bonn, the AOSIS proposal was endorsed by the Maldives, Bangladesh, Ghana, and China. It was rejected by New Zealand, who argued that it is not possible to attribute any specific extreme event to climate change. Canada and the EU argued against the need for new institutions within the UNFCCC. The US said it would consider several functions of the proposal related to insurance, but was against the proposal as a whole. Interview data suggest that the words "rehabilitation" and "compensation" in the proposal were strongly opposed by developed country parties. 4 Bangladesh and Pakistan argued that adaptation and loss and damage should not be linked (Actionaid 2010).
In the wake of general disappointment at the failure to reach a strong and binding agreement in Copenhagen, there was nonetheless some success for those promoting the loss-and-damage agenda in Cancún in 2010, with the establishment of a two-year work program on the issue. The program's three tracks were defined to be assessing the risks, investigating approaches to address loss and damage, and defining the role of the Convention in implementing these approaches (UNFCCC 2011).
With institutional recognition of the issue, NGOs began to debate the policy implications of the concept. In 2012, Germanwatch and four partner institutions, including the MCII, launched the Loss and Damage in Vulnerable Countries Initiative. Germanwatch pushed for the issue to be addressed as something beyond adaptation: the accompanying report refers to loss and damage as an "emerging third paradigm" in international climate policy ). Actionaid, in alliance with several other NGOs, also published a series of discussion papers between 2010 and 2013 calling for the establishment of an international mechanism, highlighting the distinction between slow-onset processes and extreme weather events and the importance of addressing noneconomic losses (Actionaid 2010;Actionaid et al. 2012;Actionaid et al. 2013). A series of briefing papers published by the Third World Network emphasized the limitations of risk reduction, risk retention, and risk transfer policies and the limits to adaptation (Stabinsky and Hoffmaister 2013). In terms of the role of financing, the authors write: "Clearly resources specific for loss and damage will need to be new and additional to existing resources" (Hoffmaister and Stabinsky 2012, 10).
In the lead-up to COP 18 in Doha, a number of parties from the LDC group participated in an extensive set of expert group meetings and regional workshops in conjunction with the work program on loss and damage, as well as in a series of MCII workshops. 5 They moved from a position of limited interest and engagement with the idea of loss and damage to pushing actively for an international mechanism beyond the adaptation and mitigation tracks. This can 4. Interview with legal advisor to the LDC group, London, August 28, 2015. 5. Interviews with a negotiator for the LDC group, July 19, 2013; a loss-and-damage policy coordinator for a large LDC state on July 29, 2013; and a senior policy professional in an international ENGO on July 31, 2013; all conducted in London.
be partially explained by an effort to broaden the relevance of the loss-and-damage frame beyond the idea of sea-level rise to include more focus on, for instance, climate-change-related desertification and the melting of glaciers, as well as extreme weather events. An increase in funding and support-for example, the establishment in 2009 of the pro bono Legal Response Initiative by WWF-UK and Oxfam-GB, which provides legal support to LDCs during the negotiations-also helps explain their growing engagement with the issue. 6 After a meeting organized by the MCII, the LDCs realized that loss and damage was very much an issue for them as well, that it is not just an AOSIS issue. From that point on, going into Doha, the LDC group were able to bolster the AOSIS position and have a common front against annex 1 [countries]. 7 Another interviewee noted that the work of civil society to make sure that actual survival and development is one and the same thing … helped to break some of the division that existed between AOSIS and LDCs. So it's helped build a powerful voice, by bringing a lot of developing countries together on one side. 8 In Doha, 47 NGOs sent an open letter urging parties to address the issue of loss and damage through the establishment of an international mechanism for compensation and rehabilitation under the UNFCCC. This letter echoed a call that many developing country parties had been making in the run up to the Doha COP. Eventually the issue became one of the themes discussed at the ministerial level, and the Doha text enshrined the principle of loss and damage for the first time. Paragraph 8 even includes a specific mention of "finance" that was hard fought for in Doha. However, the wording of the final text reflects the strong opposition by developed state parties, especially the US, to any measure that could suggest legal liability for climate change impacts. 9 As a result, the compromise text of the decision only obliged the parties at COP 19 to establish "institutional arrangements, such as an international mechanism … to address loss and damage associated with the impacts of climate change in developing countries that are particularly vulnerable to the adverse effects of climate change" (UNFCCC 2013). While the "such as" allowed for a wide degree of interpretive discretion, interview data suggest that there was a feeling among advocates that "now that they have agreed to it, the door is open, and now the question is, how do we widen [the opening of] that door?" 10 COP 19 in Warsaw in 2013 opened just days after Typhoon Haiyan had wreaked devastation in the Philippines, making negotiations on a loss-and-damage 6. Interview with legal advisor to the LDC group, London, August 28, 2015. 7. Interview with a negotiator for the LDC group, London, July 19, 2013. 8. Interview with campaigner for a large international ENGO, London, July 18, 2013. 9. The Australian, December 12, 2012. 10. Interview with a negotiator for the LDC group, London, July 19, 2013. mechanism a critical and high-profile issue. Due to Russia's blocking of the agenda and a standstill in discussions at the June 2013 Subsidiary Body for Implementation (SBI) meeting in Bonn, the negotiators had to work around the clock in Warsaw in an effort to draft text that balanced developing countries' call for compensation and developed countries' concerns about liability.
One key issue was where to house the mechanism in the institutional framework. Bolivia's negotiator on loss and damage, Juan Hoffmaister, representing the G77 and supported by AOSIS and the LDCs, called for loss and damage to be a stand-alone pillar. Hoffmaister reportedly changed tack from previous negotiating sessions by arriving with a draft text describing the mechanism (that had been negotiated by the G77 among themselves) for developed states to consider. 11 Fed up in part by obstructionist tactics by the Australian delegation, a group of G77 negotiators walked out of talks one morning at 4 a.m. This was followed by high-level bilateral talks that sought to resolve disagreements about whether work on loss and damage should be an institutional arrangement, a work program, or a task force, and whether it would get its own mechanism, be housed under one of the subsidiary bodies, or simply sit under the adaptation pillar. At the closing plenary, the G77 and China, AOSIS, and the delegate from the Philippines made it known that they considered loss and damage as meaning "beyond adaptation," whereas developed states, particularly the US and Australia, were pushing for the mechanism to be organized under the adaptation pillar. This opposition sparked an hourlong huddle, with US Special Envoy Todd Stern and Fiji's Sai Navoti discussing the use of the term "under" in the text while surrounded by about 50 other negotiators. 12 Consensus was ultimately reached by housing the mechanism in the adaptation pillar and by the introduction of language in the preamble on the limits of adaptation.
The WIM provides for three types of functions: enhancing knowledge and understanding of comprehensive risk management approaches; strengthening dialogue, coordination, coherence, and synergies among relevant stakeholders; and enhancing action and support so as to enable countries to take action to address loss and damage. Rather than requiring developed countries to provide developing countries with financial support, the decision "requests" this. Much remains to be decided on how the mechanism will function, how it will be financed, and what it actually requires states to do. 13 A review of the structure, mandate, and effectiveness of the mechanism at COP22 is scheduled for 2016, when the COP will make an "appropriate decision" on loss-and-damage governance.
Loss and damage continued to be a contentious issue at COP 21 in Paris in December 2015. In the lead-up to the conference, several options in the draft text highlighted the distance between different country groupings. One option, 11. Interview with an advisor on loss and damage to the LDCs and AOSIS, London, September 18, 2015. 12. Earth Negotiations Bulletin, Tuesday, November 26, 2013. 13. Interview, participant at initial meetings of the WIM interim ExCom, August 27, 2015. reflecting political concessions made by LDCs within the G77, included reference to loss and damage but made no mention of compensation or to the specific responsibilities of developed countries on the issue. A US proposal, supported by Canada, Australia, New Zealand, and others, was the removal of any reference to loss and damage from the negotiation text (Third World Network 2015). In the end, loss and damage is included in article 8 of the Paris Agreement (in a stand-alone article separate from adaptation), but the text of an accompanying decision also states that this recognition "does not involve or provide a basis for any liability or compensation."

Discussion
An analysis of the different frames presented in ECO and ENB between 2003 and 2013 (see Figure 1) gives an overall picture of how the loss-and-damage issue has been discussed over time. Before 2009, the issue tended to be discussed either in terms of an "insurance and risk transfer" frame or a "compensation and liability" frame. This changed dramatically after 2008, when a broader, more ambiguous "loss-and-damage" frame emerged, in part as a result of the Bali Action Plan language negotiated in 2007. Figure 1 shows a steep rise in references to "loss and damage" in ECO and ENB between 2008 and 2013. As the discussion below highlights, this new master frame can be interpreted by actors differently. One interviewee noted that the ambiguity around the idea of loss and damage was useful from an advocacy point of view: The reason loss and damage was easy was that nobody knows what it means yet. When you campaign for something that can mean anything to anybody, it's always easy. When you can simplify something, you can maximize the engagement. 14 Another interviewee, in reflecting on the position of vulnerable developing states in Warsaw, argued the following: Loss and damage is a euphemism for liability and compensation. Behind the scenes in Warsaw: we didn't actually want "liability" and "compensation," we were happy with "loss and damage"…. We are not giving up the right to demand it, but the compromise we made was to just discuss it and save the claiming compensation and liability for later. 15 A frame analysis shows that frame contestation has occurred in two broad areas: debates about the substance of loss-and-damage policy, and conflicts over its institutional implications.
First, there has been contestation over what the policy problem of loss and damage is (the diagnostic frames) and the best way of addressing the problem 14. Interview with campaigner for a large international ENGO, London, July 18, 2013. 15. Interview with advisor on loss and damage to the LDCs and AOSIS, London, September 18, 2015. (the prognostic frames) (Benford and Snow 2000;Boyd et al. 2016;Huq et al 2013). On one hand, some parties, such as the US, view the problem of loss and damage as mainly one of risk. The solution thus lies in the adoption of risk reduction strategies and the establishment of risk transfer or risk sharing mechanisms, such as insurance. On the other hand, some actors-for example, AOSIS-argue that the problem of loss and damage, in addition to considering risk, also raises questions of liability. According to this view, one part of the way to address loss and damage will involve the provision of compensation. It is important to note that those who advocate the latter perspective tend to see compensation and rehabilitation in a broader policy framework that also includes risk reduction and risk transfer tools. They argue that while risk reduction and insurance are appropriate for some elements of loss and damage, such as extreme weather events, they are not suitable for all elements. By focusing only on these tools, it is argued, negotiators are ignoring slow-onset impacts, noneconomic losses, and forgone development opportunities.
Second, these conflicting diagnostic and prognostic frames have different institutional implications. They raise questions about the appropriate policy venue for addressing loss and damage, and most recently have highlighted divisions about whether loss and damage should be seen as something separate and additional to adaptation efforts, or as part of the adaptation continuum.

The Risk-and-Insurance Frame
The first category of framing concerns climate-related risks and insurance. The diagnostic framing here sees the problem as one mainly of "risks," and the prognostic framing thus advocates insurance as the appropriate solution to the problem.
The idea of insurance as a way of addressing the adverse effects of climate change has been debated since the establishment of the UNFCCC. Generally, insurance has been understood by proponents as a potential mechanism to reduce uncertainty about the management of climate risks, and a way of transferring risk away from the most vulnerable nations or individuals (in the case of micro-insurance) to insurers. Insurance principles also provide incentives for the international community to move from reactive disaster management approaches to proactive prevention and risk sharing arrangements (MCII et al. 2008).
The 1991 AOSIS proposal to establish an international insurance pool to deal with the consequences of sea-level rise appears at first to sit squarely within this framing. In fact, the proposal that the revenue be drawn from developed country contributions based in part on their relative capacities to pay and in part on their relative responsibility for carbon emissions transcends both the "risk and insurance" frame and the "liability and compensation" frame (discussed below).
More recently, as the loss-and-damage agenda has gained profile, an increasing number of developed states have backed risk reduction and risk transfer approaches. For example, US submissions often place a great deal of emphasis on risk reduction and link this to the need for a loss-and-damage policy to be considered as part of a continuum with adaptation policy (USA 2011; USA 2012).
Other actors, however, have expressed hesitation about insurance solutions to loss and damage: Insurance could play an important role after a disaster has occurred but schemes must be designed to: Incorporate multiple hazards, specific to the context; Benefit and assist the most vulnerable people without burdening them financially; Be transparent and accountability [sic] and prioritise affordable protection over profit for the insurer. 16 This market-based approach to understanding loss and damage has a number of implications from a policy perspective. First, this diagnostic frame focuses much more on the idea of "risk" than on that of "harm." The focus on "risk" shifts attention away from questions of blame, causality, and accountability. It also minimizes any relationship between the "perpetrators" and "victims" of climate change. Second, the emphasis on insurance as a solution relies on the idea of uncertainty. However, not all of the impacts of climate change will be characterized by this uncertainty. As the predictive power of climate change science becomes more detailed and precise and as the number of slow-onset impacts increases, the less appropriate insurance will be as a solution. Permanent changes cannot be covered by market-based insurance, because there is no way of transferring the risk (Hoffmaister and Stabinsky 2012).

The Liability-and-Compensation Frame
In contrast to the relatively politically innocuous "insurance and risk-transfer frame" the "liability-and-compensation" frame focuses attention on questions of who is to blame for climate change and who will suffer most as a result. This framing has often underpinned proposals from developing country parties and has featured prominently in the arguments put forward by civil society actors. In this framing, funding and action by developed states, particularly those with the greatest responsibility for historic greenhouse gas emissions, is seen as a legal and moral requirement to recompense vulnerable states.
NGOs have pushed this framing of "loss and damage." For example, in their 2008 report (discussed above), WWF advocated a compensation-andliability regime. The report employs several framing strategies that address both the problem and solution to the issue. First, it outlines the scientific evidence for the inevitability of loss and damage and also identifies the problem as a legal one based on a review of international legal principles. It suggests that, unless action is taken, developed states will face an avalanche of costly lawsuits from states and individuals affected by the transboundary nature of carbon emissions.
This "liability-and-compensation" framing is also reflected in the interviews with NGO campaigners and in ECO. Typical statements include, for example: Climate change is a bit like a traffic accident. The US knows that they caused that accident, and they don't want to deal with the pieces so they're trying to negotiate their way out of it late at night. Climate change is a massive global social injustice, it's bigger than anything else, and it continues to be perpetrated by these [industrialized] states. 17 ECO also regularly includes discussions of climate justice in relation to the lossand-damage issue. For example, in Warsaw it argued: Tackling loss and damage is about climate justice. It is about protecting people, their livelihoods and, most importantly, their human rights and dignity. It is time for those who are mainly responsible for climate change to act here in Warsaw. 18 Similarly, in 2012 Actionaid and its partners called for the establishment of an international mechanism on compensation and rehabilitation. 19 While these morally charged assertions are now commonly espoused by certain groups of actors, they have had limited direct impacts on the negotiated text and seem 17. Interview with campaigner in a large international development NGO, London, July 16, 2013. 18. ECO, November 19, 2013. A year later, another discussion paper by the same group of NGOs called instead for an international mechanism to address loss and damage-a notable softening of language.
to be losing traction more generally: Figure 1 shows that mentions of "compensation" are decreasing. Any mention of liability or compensation has been rigorously resisted by most developed states. For example, at the Paris COP in 2015 the EU, the US, and other members of the umbrella group warned that they would permit the principle of loss and damage to exist in the Paris agreement only if the exclusion of compensation and liability was made explicit.
Recently, even some developing state negotiators have argued that this framing can downplay the need for support beyond financial compensation. For example, a 2014 blog by negotiators for G77 and China, AOSIS, the African Group, and the LDC group argued that: While liability and compensation form important elements of the loss and damage discussions, this perspective often trivializes the complexity of the issues and inaccurately reduces the issue to one of merely determining liability and seeking compensation…. Assuming that a compensation approach is established under the UNFCCC or other international forum, financial compensation may represent a normative solution to the peril of vulnerable countries, but it does not necessarily mean that the actual underlying needs are addressed. (Hoffmaister et al. 2014) This introduces further fragmentation into the "liability-and-compensation" frame by expanding the meaning of "compensation" beyond the purely financial. This "compensation plus" model addresses forgone development opportunities, noneconomic losses, the problem of eroding social safety nets, and the challenges of migration and climate refugees.
The liability-and-compensation frame is explicit in its implications for international policy. First, the frame focuses squarely on the actors responsible for climate change and on the states and individuals that are the victims and potential victims of the adverse effects of climate change. Second, in direct contrast with the insurance frame, the liability-and-compensation frame relies on the idea of "harm" rather than that of "risk." While the liability-and-compensation frame has received increasing traction among a number of states, with notable changes in position on the part of key developing states in recent years, it has nonetheless been explicitly rejected by developed states on a number of occasions. This helps explain its declining usage in the public realm in the negotiations and the shift to a more ambiguous master frame (Huq 2013).

Loss and Damage: Contestation over Institutional Implications
Over the history of negotiations, two key lines of contention have emerged between developing and developed countries on the institutional implications of the loss-and-damage agenda.
A first area of contention concerns disagreement about whether loss and damage should be addressed as part of the UNFCCC at all. States that have advanced a risk-and-insurance frame have also tended to challenge the idea of a global solution to the problem of loss and damage. One example of this is disagreements over whether loss and damage should be addressed at the country or at the global level. For example, in 2010 Spain (in a statement for the EU), preferred to have climate-related damage addressed at the national level. Similarly, in its 2011 submission to the two-year work program on loss and damage, the US stated that it had "serious concerns about the development of a global climate risk insurance facility … a global facility could inhibit a country-driven approach to adaptation" (USA 2011). In 2012 the US reiterated this point at the informal presessional meeting on loss and damage: We believe vulnerable countries should be able to decide to reduce risks and avert loss and damage. An international mechanism with insurance and compensation pillars could severely undermine countries' abilities to make those decisions at the national level, and reduce resources left for those kinds of measures. (USA 2012) This statement also highlights the US position on loss and damage in relation to adaptation: that loss and damage is still something that can largely be addressed through risk management rather than something that is inevitable.
This country-level prognostic framing of loss and damage has the advantage for these developed countries of minimizing questions of transboundary harm and attribution. Instead, the idea of responsibility in this framing concerns the relationship between the leaders of states vulnerable to loss and damage and their own populations. The "problem" concerns the ability of leaders to look after their own citizens, with the emphasis being placed on the idea that it is not too late to adapt and manage risk, and the "solution" therefore lies with national elites.
Contention has also resulted over whether loss and damage should be addressed within the UNFCCC process or in a different global regime. Some developed state parties have argued that loss and damage would be addressed better within the context of the Hyogo Framework for Action (and its successor the Sendai Framework), which seeks to build resilience to natural hazards and is housed in the UN Office for Disaster Risk Reduction (UNISDR). 20 The prognostic framing that loss and damage is a problem that should be a part of the Hyogo or Sendai Framework minimizes the fact that climate change is the cause (or amplifies the effects) of certain "natural" disasters. The Hyogo and Sendai frameworks address all types of extreme weather events, and thus skirt over questions of causality and responsibility. More generally, assertions that loss and damage should be addressed outside the Convention can also be understood as a strategy on the part of developed states to take advantage of the increasing international fragmentation of the climate regime (Zelli and van Asselt 2013). The UNFCCC offers a relatively robust international legal framework when compared to other global policy venues such as the UNISDR.
Developing country parties, on the other hand, have been steadfast in their support for a global mechanism to be housed within the UNFCCC. In their submission to the presessional informal meeting in 2012, Bolivia and other developing countries argued that "The UNFCCC is the relevant policy forum for discussing loss and damage…. Loss and damage from the adverse effects of climate change is directly related to the successful or unsuccessful fulfilment of the objective of the Convention" (Bolivia et al. 2012).
A second area of contention over institutional implications has emerged since 2009, when the idea of an international mechanism began to gain the backing of a large number of developing states. This concerns the relationship between the loss-and-damage issue and adaptation policy. Some actors argue that loss and damage constitutes a new paradigm in the climate change regime, whereas others have negotiated hard to ensure that loss and damage is understood as only one part of adaptation strategies.
The NGO community, country groups such as AOSIS and the LDCs, and some specific developing states such as Bolivia and the Philippines have been the strongest advocates for the idea of loss and damage as an issue "beyond adaptation." NGO interviewees point to two main reasons for the decision to push the idea of a "third paradigm" in the climate change negotiations: a sense that the negotiations were failing, and increasing scientific evidence that irreversible climate change was already happening. ECO has consistently made these arguments over time, since at least 2004: The possibility is emerging that life could soon become intolerable in some parts of some countries. The issue will then arise of compensation to these countries for the costs arising-such as from internal migration, feeding programmes, development foregone…. The science tells us that we need to prepare for scenarios where damages exceed adaptation capacity. 21 It took up a similar position five years later: "Leaders, you have to face this spectre, for it will not disappear. There will be losses which can't be addressed by adaptation." 22 Powerful state parties, such as the US, have successfully pushed a framing of loss and damage that sees it housed within the Cancún Adaptation Framework. Nonetheless, some compromise was achieved at the last minute, and a sentence in the preamble to the decision establishing the WIM recognizes the "beyond adaptation" frame by acknowledging "that loss and damage associated with the adverse effects of climate change includes, and in some cases involves more than, that which can be reduced by adaptation" (UNFCCC 2013). 21. ECO, December 1, 2004, 4. 22. ECO, December 18, 2009 Conclusions A framing approach is helpful for understanding how interactions between different actors ultimately shaped the policy problem of climate-related loss and damage and led to the surprising success of the adoption of the WIM in 2013. The consensus needed to achieve this success masks a long history of contention. Conflict over two proposed normative framings of adverse climate change impacts and the institutional implications of these framings characterized the nature of debate until 2008, when stakeholders began to use a more ambiguous overarching "loss-and-damage" frame, in line with the language of the 2007 Bali Action Plan. For some actors, notions of historic culpability and the global injustice of climate change lie at the heart of the idea of loss and damage. While these actors have been successful in finally placing the issue on the policy agenda, they have, over the history of the negotiations, been unsuccessful in achieving a specific, legally embedded understanding of loss and damage that includes acknowledgment of culpability or the possibility of compensation. Reluctance on the part of developed states to embed the idea of liability for climate change harms in international law has led to their emphasis on another frame-one that focuses on risk and insurance-that rhetorically minimizes questions of culpability and ignores certain types of events and consequences that fall outside of the "insurable risk" category. However, it is notable that the "risk-andinsurance" frame alone failed to achieve a consensus in the international arena. Instead, a broader, more ambiguous understanding of loss and damage-one that reflected last-minute compromises, allowed for varying interpretations by different parties, and postponed key decisions on issues such as finance-was adopted into international law in Warsaw.
With the negotiation of the Paris agreement in 2015, the pendulum swung again: away from the more ambiguous understanding of loss and damage to discussion over whether the issue should be referred to in the text at all, and the explicit exclusion of claims for compensation or assertions of liability. This increasing precision represents a new stage in the institutionalization of the idea of loss and damage, and future research could explore how the movement between ambiguous and specific framings of a concept affects the likelihood of an issue being placed on the agenda, institutionalized, and ultimately realized.