The Third Pillar of International Climate Change Law: Explaining ‘Loss and Damage’ after the Paris Agreement

With the 2015 Paris Agreement, ‘loss and damage’ (L&D) was introduced into the unfccc treaty framework as a new, third substantive area of climate change law. Both before and after its adoption, this new area has been subject to much contention— and this is reflected in a high degree of uncertainty surrounding its interpretation. This article examines the definition of L&D and the types of impact covered by the notion. It also examines the relationship of L&D with mitigation and adaptation, as well as the instruments that are covered by it. Finally, the article considers the controversial issue of who can invoke L&D—and against whom.


Introduction
When the unfccc was adopted in 1992, mitigation was established as the first pillar of international climate change law and adaptation as the second. In 2015, at cop 21, the Paris Agreement added a third pillar: 'loss and damage' .1 The introduction of a dedicated provision on L&D in the unfccc treaty framework was far from straightforward. On the contrary, it was one of the most contentious issues in the Paris Agreement negotiations. Whereas several developing countries were fervent supporters of introducing L&D into the Paris Agreement, many developed countries strongly opposed it. (I will use 'L&D' throughout in preference to 'loss and damage' because I want to advance the view that L&D is to be treated as a new term of art.) The clear division of positions in cop negotiations, which ultimately led to the adoption of the Paris Agreement's Article 8 on L&D, is the principal reason why this provision has all the hallmarks of a compromise.2 And it is not surprising that academics and practitioners alike grapple with the construction of the notion of L&D. In this article I will examine some of the notion's key aspects and put forward possible (but needless to say not definitive) answers to some of the most pressing questions regarding L&D. I will first provide a brief history of the concept (section 2). Next, I will consider the notion's definition (section 3); following which, I will discuss the types of impact covered by L&D (section 4). The relationship between L&D and adaptation and mitigation is considered in section 5. Article 8 provides a non-exhaustive list of instruments that may be considered to fall under L&D. These are briefly reviewed in section 6. A real challenge when it comes to construing Article 8 is that it seems not to provide a basis for liability or compensation. In section 7, I will consider the extent to which this precludes those suffering climate change L&D from seeking compensation.

L&D: the History So Far
The first discussions about the introduction of a scheme on L&D in the field of climate change were initiated by the Alliance of Small Island States during the negotiations that led to the adoption of the unfccc in 1992. aosis argued that it was necessary to link climate change mitigation to a compensatory scheme. The group proposed the establishment of a compensation fund (an insurance scheme) whereby: The financial burden of loss and damage suffered by the most vulnerable small island and low-lying developing countries (Group 1 countries) as a result of sea level rise shall be distributed in an equitable manner amongst the industrialized developed countries (Group 2 countries) by means of an Insurance Pool.3 While the scheme proposed by aosis was not included in the unfccc, the idea that the international community would need to consider compensation and insurance when addressing the effects of climate change took root. Countries' .8 Possibly on account of this paper, aosis tried to revive its proposal by submitting an expanded version of the original to the awg-lca. This socalled 'multi-window mechanism' would consist of three interdependent components, on insurance, rehabilitation/compensation, and risk management.9 In the lead-up to cop 15, in Copenhagen, in 2009, aosis kept pushing for the introduction of an L&D mechanism. In a declaration of 21 September 2009, the group asserted that the cop outcome should, among other things: Establish a mechanism to address loss and damage from climate change comprised of a disaster risk component, insurance, and compensation funds, to help sids manage the financial and economic risks arising from climate impacts; to assist in the rapid recovery and rehabilitation from climate related extreme weather events and to address unavoidable damage and loss associated with the adverse effects of climate change.10 cop 15 did not deliver on this, but at cop 16, in Cancun, in 2010, it was decided to: establish a work programme in order to consider, including through workshops and expert meetings, as appropriate, approaches to address loss and damage associated with climate change impacts in developing countries that are particularly vulnerable to the adverse effects of climate change.11 The programme drew on a wide variety of actors for input on issues relevant to furthering L&D work within the unfccc framework. In 2012, at cop 18, in Doha, it was decided that at the next cop, in Warsaw, the parties would seek to establish institutional arrangements aimed at addressing L&D.12 Duly established, the Warsaw International Mechanism for Loss and Damage (wim) states in its preamble: Acknowledging the contribution of adaptation and risk management strategies towards addressing loss and damage associated with climate change impacts, Also 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.13 And Article 1 provides: [The wim is] to address loss and damage associated with impacts of climate change, including extreme events and slow onset events, in developing countries that are particularly vulnerable to the adverse effects of climate change.
cop 19 established an Executive Committee to guide the operationalization of the wim in the course of an initial two-year work plan.14 The last substantive step so far was in 2015, when, as noted above, Article 8 of the Paris Agreement was agreed to.15 12 Subsidiary Body for Implementation, Approaches to address loss and damage associated with climate change impacts in developing countries that are particularly vulnerable to the adverse effects of climate change to enhance adaptive capacity, 37th session, Doha,

No Official Definition of L&D
There is a general consensus that the L&D mechanism has been established in the absence of formal definitions and through unclear language leaving very considerable leeway for interpretation.16 In tort law, L&D's two constituent terms, 'loss' and 'damage' , do not always carry the same meaning. A person may cause 'damage' to a machine, leading to a 'loss' of profit. It does not necessarily follow that a distinction between the two terms is equally warranted in the field of climate change law-even though some scholars have tried to draw such a distinction. Thus it has been argued that 'Loss refers to climate-related impacts for which restoration is not possible' , whereas 'Damage refers to negative impacts for which restoration is possible' .17 Undoubtedly, this distinction would be useful if there were a real need to distinguish between 'loss' and 'damage'-for example, if situations involving a 'loss' would need to be treated differently from cases of 'damage' . Some commentators have taken the view that the distinction is materially relevant. For example, Emma Lees has argued that the distinction 'is essential to the proper operation of any system which attempts to allocate responsibility for loss or damage' .18 However, applied to climate change issues, it is difficult to find convincing examples that support this view. Such a distinction would seem at odds with the fact that, in many cases of climate change impact, some kind of restoration may be possible even though full restoration may be difficult (or impossible), so that in practice the distinction will make the application of the notion of L&D unnecessarily complicated.
A more pragmatic and less complicated approach to the understanding of the term L&D is warranted. In legal English, the use of doublets is rather common. Well-known examples include 'null and void' , 'cease and desist' , and 'terms and conditions' . (There are even some triplets such as 'promise, agree, and covenant' .) A doublet is a legal term of art. Where the drafters of a legal text have used a doublet, a lawyer will construe each of the components separately only if it is clear that the drafters intended to signal different meanings for each of its components. Otherwise, where a doublet has been used over a sustained period of time, it will be quite certain that it must be construed as a single notion. By now, the notion of L&D has been used for several years without any obvious indication that the two constituent terms are to be construed differently. So I will proceed on the assumption that L&D is to be given a unified construction.19 The important question, obviously, is what this notion covers.

Types of Impact Covered by L&D
In the words of Article 8(1), the 'Parties recognize the importance of averting, minimizing and addressing loss and damage associated with the adverse effects of climate change' . The provision goes on to state that this includes both 'extreme weather events and slow onset events' . In other words, L&D covers a wide range of events, from sudden ones, like flooding, cyclones, and heat waves, to gradual ones, such as drought, sea-level rise, glacier melt, and desertification. L&D is generally considered to cover both irreparable (irreversible) impacts, such as loss of life or loss of an ecosystem,20 and impacts that can be repaired (reversed), such as damage to infrastructure.21 It covers both economic and non-economic loss, as set out in Article 8(4)(g). However, it is unclear what non-economic loss covers: cultural harm, harm to nature (including extinction threats), threats to the survival of certain states … or perhaps something different or something more.
In sum, it is in general possible to identify the types of impact covered by L&D, but in some areas-particularly non-economic loss-clarity is lacking. 19 See Sam Fankhauser, Simon Dietz, and Phillip Gradwell, 'Non-economic losses in the context of the unfccc work programme on loss and damage' , Policy

Relationship with Mitigation and Adaptation
L&D is normally understood to cover measures that address the impacts of climate change that are 'residual' to mitigation and adaptation. In the formula of Action Aid International, Care International, and wwf: 'Loss & damage = insufficient mitigation + inadequate adaptation' .22 Construing L&D in this way is sometimes referred to as the 'beyond adaptation approach' .23 However, 'beyond adaptation' is only one way of distinguishing between mitigation and adaptation, on the one hand, and L&D. In the literature we find two more distinctions. One focuses on the 'tolerable risk' . Here, adaptation is about keeping risks within the range of what is perceived as 'tolerable' , whereas L&D is a response to risks that cannot be kept within that range.24 A second one distinguishes impacts that are 'avoidable, unavoidable, or unavoided' . Thus, if it is impossible to adapt to an impact, so that it becomes unavoidable, it will fall in the L&D category. For impacts that are avoidable, it is necessary to distinguish between those that are avoided and those that are not. If it is possible to adapt to an avoidable impact so that it is avoided, this is a case of adaptation. However, if an avoidable impact is not avoided, it is unclear on this view whether it is to be categorized as (non) adaptation or L&D.25 Consider the case of a small-island state threatened by sea-level rise. The state decides to build protective dikes-a form of adaptation. Because it can only build dikes around some of its islands, it is forced to abandon some of them to the rising sea. This we will consider to fall into the L&D category. Note, however, that the question of whether or not to build a dike on a given island may not be merely a question of technical feasibility. It may be a political question about how to use the state's limited resources. In other words, it may be that the state must choose between either building dikes around the main islands and abandoning the rest, or building dikes around all of the islands and, because of the cost involved, cutting back on basic health services and schooling. In this situation, where it is a political decision not to (fully) adapt, it is not clear whether the abandonment of certain islands ('unavoided loss') should be categorized as (non) adaptation or L&D. Whereas the limits to adaptation in this example are economic, in practice they may be much more complex and involve socio-cultural and institutional (in addition to biophysical and technical) aspects, as illustrated by the case study of Otoara Ha'apio et al. of the Solomon Islands.26 The case illustrates how the distinction between adaptation and L&D can be blurred.27 I referred to the 'tolerable risk' distinction between adaptation and L&Dwhereby L&D is a response to risks that cannot be kept within the range of the tolerable. Building dikes around selected islands is a way of ensuring that risks remain tolerable, whereas abandoning some islands to the rising sea falls into the L&D category. However, the 'tolerable risk' approach is rather subjective, as it is based on the value judgments of those faced with the actual climate change impacts-meaning that it does not provide a clear and objective distinction between adaptation and L&D.
The different approaches to distinguishing between adaptation and L&D may not only lead to different categorizations of cases as either adaptation or L&D. They may also lead to different priorities within the field of L&D. As has been pointed out: while the [beyond adaptation] approach to distinguishing adaptation and L&D mainly focuses on the impacts and the measures they demand to differentiate responsibilities, the [tolerable risk] approach primarily derives the responsibilities to be differentiated from whether and to what extent capacity building is necessary.28 Because none of the proposals is fully workable or clear-cut, it might be better to focus on the diverging objectives of adaptation and L&D. Adaptation is a case where an adverse impact is prevented, whereas L&D applies where such impact is not prevented, irrespective of why. This basis for distinction would seem to accord with Doelle and Seck's view, who observe that in one definition of L&D 'The focus has been on harm caused by human-induced climate change itself' .29 In other words, adaptation is about avoiding harm from occurring, whereas L&D is about addressing the harm that occurs, whether or not it could have been reduced by adaptation. This approach also seems in line with Article 8(3) of the Paris Agreement, which merely refers to 'loss and damage associated with the adverse effects of climate change' .

Types of Instrument Covered by L&D
The L&D debate has circled around three different themes: arrangements to cover the costs of impacts and risk that are not prevented, including compensation schemes; arrangements aimed at creating awareness that neither human nor natural systems will be capable of handling climate change unless society adequately responds to it now; and arrangements aimed at enhancing adaptation and resilience through risk reduction and management. All three are reflected in Article 8, which requires parties to cooperate to avert, minimize, and address L&D. Article 8(4) provides the following non-exhaustive list of areas of cooperation and facilitation: (a) Early warning systems; (b) Emergency preparedness; (c) Slow onset events; (d) Events that may involve irreversible and permanent loss and damage; (e) Comprehensive risk assessment and management; (f) Risk insurance facilities, climate risk pooling and other insurance solutions; (g) Non-economic losses; (h) Resilience of communities, livelihoods and ecosystems. The above is a somewhat mixed bag, and it is unclear why, for example, 'resilience of communities…' is part of L&D and not adaptation.30 29 Doelle and Seck, supra note 27. 30 It has been pointed out that 'some of In practice much attention has been focused upon insurance solutions31 and on climate-induced displacement.32 However, these issues are only two among several relating to L&D, and, as a rule, risk insurance does not provide a workable solution with respect to, for example, non-economic loss.

7
Who can Invoke L&D-and against Whom?
The term L&D is drawn from the field of compensation, so it would seem only logical that a provision on L&D would allow the affected parties to raise a claim for compensation against those who caused the L&D. Nevertheless, an important limitation in the Paris Agreement is 'that Article 8 of the Agreement does not involve or provide a basis for any liability or compensation' .33 However, this does not mean that parties suffering L&D cannot claim compensation on other legal bases.34 Doelle and Seck have considered the possibility of claiming a remedy on the basis of harm as a result of unavoided L&D. They rightly point out that entities that may experience such harm 'include states, sub-national government actors, as well as a variety of non-state actors ranging from individuals to organizations and communities' ,35 and they show that the possibility of raising a claim for L&D depends on the specific legal context and whether it makes a legal remedy available. On potential defendants, they observe that: climate law 10 (2020) 211-223 Who will be an appropriate defendant will vary depending on the harm, the legal system called upon to provide a remedy, and the conduct that contributed to the harm. States, state actors, state owned enterprises, international organizations, and private actors are all potential defendants depending on the nature of the harm, the plaintiff, the legal system involved, and the remedies sought.36 In other words, in order to answer the question of who can invoke L&D, and against whom, it is necessary to survey domestic and international legal regimes containing different mixes of eligible claimants, respondents, remedies, and actionable wrongs.37 Whereas the above would seem to indicate that when it comes to the central question of compensation, the provision on L&D is all bark and no bite, I would argue that Article 8 within the unfccc framework may still play a (limited) role. It is above all about L&D suffered in developing countries as a consequence of actions on the part of developed countries.38 Prospective claimants will need to explore a patchwork of different legal regimes, but it will be those affected in developing countries who may be able to invoke rights, depending on the legal regimes available to them.39 Similarly, whether dutybearers exist will also depend upon the applicable legal regimes, but are very likely to be both public and private actors in developed countries.

Conclusion
We have seen that even though L&D was put on the negotiation table before the adoption of the unfccc in 1992, it took more than two decades for the notion to gain a dedicated foothold in the unfccc treaty framework. L&D thereby came to complement the pillars of mitigation and adaptation as a third pillar of international climate change law. The treaty regime does not give an official definition of L&D, and the formal L&D mechanisms have been established through unclear language leaving considerable leeway to their interpretation. In this context one may wonder 36 Ibid., 6. 37 Ibid., 10. 38 Cf. recital 8 of the Paris Agreement's preamble as well as the references to the wim in the Agreement's Article 8 and in unfccc Decision 1/CP.21, 'Adoption of the Paris Agreement' FCCC/CP/2015/10/ Add.1 (2015). The references to the wim make it natural to construe Article 8 as primarily addressing L&D in developing countries. 39 Thus, the right-holders may be individuals, communities, states, as well as others.
whether the two terms 'loss' and 'damage' must each be given a distinct meaning. In this article I have argued that the better view is that L&D is to be viewed as a single notion. L&D has a wide coverage, encompassing impacts stemming from sudden as well as slow-onset events, impacts that are irreparable (irreversible) as much as reparable (reversible), and losses that are economic in nature as well as (in some instances) non-economic. The delimitation of impacts covered by L&D remains unclear.
Both in theory and in practice it may be difficult to distinguish L&D from adaptation. In academic writing we find a variety of ways of making the distinction. However, none of them is convincing. The better approach is to focus, simply, on the different objectives of adaptation and L&D: adaptation applies where an adverse climate-change-induced impact is prevented, whereas L&D applies where such an impact is not prevented.
L&D is about several things at once: sharing the costs associated with climate change impacts and risk that are not prevented; drawing attention to the need to engage in fitting responses; and risk reduction and risk management through arrangements aimed at enhancing climate change adaptation as well as climate change resilience. In principle, Article 8 of the Paris Agreement provides a basis for instruments that may be used towards these objectives. However, even though the term L&D originates in the compensation field, Article 8 'does not involve or provide a basis for any liability or compensation' .
When it comes to who can invoke L&D against whom, the fact that Article 8 does not provide a basis for liability or compensation necessarily means that to a considerable extent the provision is without bite. It does not mean, however, that every possibility of liability or compensation with regard to L&D is precluded. What it means is that it is necessary to find a way to exploit domestic as well as international legal regimes, which will result in different combinations of eligible claimants, respondents, remedies, and actionable wrongs. The rights-holders under L&D must nevertheless be (those affected in) developing countries, whereas the duty-bearers will be found in developed countries.