The treaty management organization established under the UNFCCC and the Paris Agreement: an international actor in its own right?

A peculiar treaty management organization operates under the UNFCCC and the Paris Agreement that does not fit established categories of international law and political science. Unlike traditional international organizations (IOs), it lacks the formal status of an IO, comprises only a limited secretariat with predominantly servicing functions, and is not even denominated as an organization. We argue that it has nevertheless become an international actor, mainly due to activities of its Conference of the Parties and several specialized organizational schemes. Theoretically, we develop an analytical framework that shows how even heavily member-dominated IOs can become international actors and what this means for global environmental governance. IOs gain the capability to influence international politics by their own action if authorized to make decisions with external effects. They gain autonomy if organizational rules and procedures shape organizational decision-making and create specific organizational rationales. Empirically, we demonstrate that the organizational component of the UNFCCC and the Paris Agreement has acquired both considerable action capability and autonomy. It is authorized to flesh out the skeletal provisions of the constituent treaties through far-reaching COP decisions and to decide continuously in several specialized organizational schemes, especially on climate funding, cooperation mechanisms, and compliance management. Organizational decisions are heavily influenced by autonomy-creating organizational factors, such as path dependence, fundamental organizational norms and dense sets of decision criteria. We conclude that this organization, and the organizational components of other multilateral environmental agreements, point at important organizational effects, which merit further attention.


Introduction
With the adoption of the UN Framework Convention on Climate Change (UNFCCC) of 1992, the Kyoto Protocol of 1998, and the Paris Agreement (PA) of 2015, the member states have created a peculiar organization that does not fit established categories of political science and international law (Churchill & Ulfstein, 2000). Its supreme decision-making body is the Conference of the Parties (COP), but it comprises also several subsidiary bodies and manages international funds and specialized schemes like the Clean Development Mechanism (CDM). As a "complex organization" (Abbott et al., 2016: 251), it has "its own separate legal personality" (United Nations, 1993: 428-429). Like full-fledged international organizations (IO), it allows adopting collective decisions with binding effects on member states and converts inputs into outputs (Rittberger et al., 2019: 57). However, in contrast to traditional IOs like the World Bank or the World Health Organization, it lacks the formal status of an IO, comprises only a limited secretariat with predominantly (although not entirely) servicing functions, and is not even denominated as an organization. This arrangement reflects the intention of the member states to establish a permanent organizational apparatus for the management and development of cooperation under the constitutive treaties, while avoiding the rigidities and inefficiencies of bureaucratic IOs. It resembles the organizational components of other multilateral environmental agreements (MEAs), whether following the convention-protocol model like the Vienna Convention and Montreal Protocol protecting the ozone layer or based on a single international treaty like the Convention on International Trade in Endangered Species of Flora and Fauna (CITES). Legal scholars conceive of these institutions as 'virtually identical to international organizations' (Brunnée, 2002: 16) and conceptualize them as 'new international law-makers' (Wiersema, 2009). While political science scholars point at possible risks for member states (Morin et al, 2022), International Relations theory has not yet developed concepts to grasp the organizational implications of this peculiar type of organization (Green & Hale, 2017: 474).
In this article, we develop an analytical framework for the analysis of the organizational components of MEAs and argue that the treaty management organization established under the UNFCCC, the abandoned Kyoto Protocol, and the Paris Agreement has become an international actor that can influence international politics in its own right. To examine the sources and effects of its agency, we conceive of it as an organization, although less formal than traditional IOs, and refer to it as the Climate-IO. IOs are increasingly recognized as autonomous international actors, rather than mere arenas for the struggle among member states. As 'governors' of areas of international relations (Lake, 2010: 590-600), they constitute separate sources of influence on world politics. However, current IO theory assigns IO agency primarily to IO secretariats and other non-state institutional agents (Barnett & Finnemore, 2004;Hawkins et al., 2006). Therefore, scholarship interested in organizational agency tends to focus on activities of IO secretariats (Biermann & Sieberhüner, 2013;Mai & Elsässer, 2022). Yet, major organizational implications of the Climate-IO do not arise from secretariat activities, but from decision-making of member-dominated organizational bodies, i.e. the COP and its subsidiary committees and boards. Accordingly, we conceive of the Climate-IO as a predominantly member-dominated organization and examine sources and effects of its agency beyond secretariat activities. Hence, our paper complements secretariat-focused scholarship and helps advance the general understanding of a specific type of organization that is particularly widespread in global environmental governance.
To theorize the agency of international environmental institutions of the Climate-IO type, we draw on novel concepts of corporate agency (Gehring & Urbanski, 2023;Hofferberth, 2019). Like other actors, IOs generally need action capability and autonomy to become actors in their own right. The ongoing discussion on international political authority (Hooghe & Marks, 2015;Zürn et al., 2012) suggests that IO agency relies on the relation of IOs with their member states, not on the existence of institutional agents (Lake, 2010). IOs gain action capability whenever their members authorize them to make decisions with external effects, e.g. regulate a given area of international politics or allocate commonly managed funds, and submit to their rulings. IOs may also gain some autonomy from organizational constraints (as opposed to member state preferences) that influence organizational decisions.
Empirically, we examine the sources and effects of Climate-IO agency along the two dimensions of authority and autonomy, based on official documents. The Climate-IO has gained action capability to a remarkable extent from its authority to adopt decisions with external effects in two core areas of activity, namely COP decision-making and the operation of specialized organizational schemes. The member states have invested the COP with incredibly wide powers to flesh out the sparse provisions of the constitutive treaties (Brunnée, 2002: 15-33;Rajamani & Bodansky, 2019). The COP has used its powers to establish under the UNFCCC and the PA, inter alia, a multifaceted financial mechanism, a cooperation mechanism that follows the Clean Development Mechanism (CDM) of the Kyoto Protocol, and a transparency and compliance mechanism. Several of these schemes empower specialized organizational bodies to continuously adopt case-specific decisions, say on project funding and compliance matters. The Climate-IO has also gained considerable autonomy. Although adopted by consensus of member states, COP decisions are tightly embedded in an increasingly dense institutional structure that creates a specific organizational rationale and causes the typical path dependence of organizations (David, 1994). Decisions in the realm of specialized organizational schemes are made by specialized bodies according to procedures designed to drive actors toward criteria-based (technical) decision-making, instead of preference-based bargaining.
Conceptualizing the Climate-IO as an IO sheds light on important organizational consequences of MEAs and emphasizes the relevance of post-agreement politics (Kalfagianni & Young, 2022: 251-252). When authorizing the organization to adopt decisions with external effects, the member states establish an additional international actor and empower it to exercise international authority. When adopting organizational decisions, procedures, or criteria, they establish increasingly dense sets of institutional constraints, within which subsequent organizational decisions must be accommodated. These institutional constraints become sources of organizational autonomy and create an organizational rationale without which organizational decisions cannot be understood.
We proceed as follows. In the next section, we develop a conception of member-dominated IOs as actors and trace their sources of action capability and autonomy. Subsequently, we examine important domains in which the Climate-IO has exercised international authority through organizational decisions under the UNFCCC and the Paris Agreement, especially by establishing the financial mechanism, the cooperation mechanism, and the transparency and compliance mechanism. Finally, we examine the sources of organizational autonomy both in the realm of COP decision-making and regarding the operation of specialized organizational schemes. We conclude that the Climate-IO is not merely a secretariat plus an arena for interaction among member states and non-state actors. Like the organizational components of other multilateral environmental agreements, it has become an important international actor with significant autonomy and considerable power to influence international politics by its own organizational action. To grasp its organizational effects, it should be analysed as a dynamic international organization.

Member-dominated international organizations as actors
As opposed to simple treaties, IOs are particularly useful where cooperation projects require regular adjustments, continuous collective decision-making, or permanent operational activities (Abbott & Snidal, 1998). Ideally, we can distinguish between two stages of governance through IOs. At stage 1, a group of actors creates an IO by concluding a constitutive treaty or agreement that defines, at the minimum, purpose, membership, and some basic decision rules, including organizational bodies. At stage 2, IO bodies adopt decisions and operate according to organizational procedures. The Climate-IO follows this basic model. The member states have founded and empowered it by concluding the UNF-CCC, the Kyoto Protocol, and the PA (stage 1). Based on these constitutive treaties, organizational bodies, especially the COP and its subsidiary bodies, operate and adopt secondary decisions (stage 2, Depledge, 2017). The member states are involved at both stages, but in different roles. At stage 1, they negotiate the constitutive treaties and decide unilaterally whether to ratify or not. At stage 2, they act (alongside other actors) as members of IO bodies, which adopt collective decisions according to institutionalized procedures (Schermers & Blokker, 2011: 61-62). To study the organizational implications of the Climate-IO, we focus on stage 2.
IOs become international actors in their own right, if they gain action capability and a minimum of autonomy. These are the most general prerequisites for any type of actor (Hofferberth, 2019, 132-134;Gehring & Urbanski, 2023). No entity qualifies as an actor if it is incapable of performing activities that make a difference beyond its confines. What matters is not success of its action, because action may always fail, but the ability or potential to influence world politics. Likewise, IOs cannot become actors in their own right without a minimum of autonomy to determine their action. Without any autonomy, they would be entirely instruments of other actors, especially their member states. Accordingly, we distinguish between two dimensions of organizational agency: Action capability addresses the question of what an organization can do to influence its environment, while autonomy relates to the issue of how it decides on its action. We discuss the two core dimensions of organizational agency separately.

Authority to adopt collectively binding decisions empowers IOs to act
How can member-dominated organizations like the Climate-IO gain action capability despite the central role of their member states in organizational operations?
Generally, IOs act through decisions and operational activities of their authorized bodies. Whenever their members authorize them to adopt decisions intended to create effects beyond their confines, IOs gain action capability. By authorizing an IO to decide, the member states pool part of their sovereignty and transfer it to the IO; and they submit to IO decisions made according to valid procedures (Lake, 2007: 231-232). This transfer of authority inevitably involves a loss of sovereignty by member states, because it changes the default condition, i.e. the outcome that occurs if organizational decision-making fails. If a group of states cannot agree on coordinating their international climate funding policies, states may enact their own policies unilaterally. In contrast, funding is blocked, if members have pooled money in an international climate fund and funding decisions cannot be adopted.
If invested with decision-making power, IOs can exercise political authority and become governors of areas of international politics. Political authority is rightful, legitimate rule (Bogdandy et al., 2010: 11). It arises from a (possibly implicit) contract between a governor and the governed. The governor provides political order in exchange for the acceptance of its decisions by the governed (Lake, 2010: 591). Hence, establishing an IO as a governor of an area of international politics creates a form of accepted hierarchy that replaces international anarchy (Lake, 2010: 608). This distinguishes political authority from other forms of authority, in particular epistemic authority, i.e. the acceptance of an individual or organization as an authority due to her expertise (Katsikas, 2010: 116-119).
The exercise of political authority by an IO does not depend on particular organizational decision-making arrangements (see also Zürn et al., 2012: 88), although this is occasionally assumed (Hooghe & Marks, 2015). Member states can choose among different institutional forms of IO decision-making to establish international political authority (Cooper et al., 2008: 506), in particular between delegation and pooling (Hooghe & Marks, 2015;Lake, 2007: 232). Delegation denotes the grant of authority by principals to an institutional agent, e.g. a secretariat, whereas pooling occurs when actors transfer authority to a collective decision-making body, of which they are members. Even state-dominated IOs like the Climate-IO, which rely predominantly or entirely on pooling, may exercise political authority and become governors of areas of international relations. Therefore, it is premature to attribute IO agency primarily or entirely to their secretariats, as is frequently done (see Barnett & Finnemore, 2004;Biermann & Sieberhüner, 2013;Mai & Elsässer, 2022).
Decisive is the effective authorization of an IO to adopt decisions intended to create external effects (Cooper et al., 2008: 505). This is not limited to decisions with a formal legal status, because obligation is gradual (Abbott et al., 2000, 408-412). Even formally non-binding IO decisions may be authoritative (expected to be followed) or self-enforcing, so that their effect cannot easily be avoided (Bogdandy et al., 2010: 11-15). For example, IOs may be authorized to adopt authoritative interpretations of treaty provisions that do not change formal legal obligations, but modify their implications (Churchill & Ulfstein, 2000: 641-642). IOs may also incorporate funding schemes that do not need enforcement because they cannot be used without complying with relevant rules and conditions (Schermers & Blokker, 2011: 760-761). In contrast, decisions of purely internal relevance, e.g. election of committee chairpersons, and symbolic decisions reflecting disagreement among the constituent actors, do not create IO action capability.
The nature and extent of political authority varies because the members empower IOs selectively (Schermers & Blokker, 2011: 157-158). IOs may be authorized to make rules applying to a wider range of addressees or cases, e.g. fleshing out basic provisions of the underlying treaty, or modifying certain obligations (Green & Colgan, 2013: 477). For example, the Meeting of the Parties to the Montreal Protocol is authorized to 'adjust' (speak: tighten) certain emission reduction obligations. IOs may be empowered to adopt case-specific implementation decisions with immediate effects only for those involved in a particular case, for example on project funding under the UNFCCC or on the listing of protected species of flora and fauna under CITES. They may be authorized to manage compliance (Goeteyn & Maes, 2011), e.g. by appraising relevant information, facilitating compliance or imposing sanctions on non-compliant member states. Finally, IOs may be empowered to conclude agreements with other public and/or private domestic or international institutions.

The autonomy of IOs
How can member-dominated organizations like the Climate-IO gain autonomy, although their member states play an overwhelming role in the organizational decision process? By autonomy, we mean the influence, which an IO exerts on organizational decisions (Reinalda & Verbeek, 1998: 3). Like authority, organizational autonomy is gradual. The more autonomy an IO has, the more important are organizational factors in explaining an organizational decision, and the less influential are member state preferences in the particular decision situation.
Generally, organizational autonomy originates from institutional rules and procedures, which shape and constrain action within organizational decision processes and influence IO decisions (Rittberger et al., 2019: 57). These opportunity structures originate from the constitutive treaties, on which IO activities are based, and from secondary IO decisions. Even the most powerful member states operate under institutional constraints that may privilege weaker members, provide selected participation rights for private actors and institutional agents like secretariats, or envisage multi-stage decision processes involving several organizational bodies. Because of this heavily institutionalized setting, even treaty bodies composed of all IO member states differ from stand-alone diplomatic conferences (Bradley & Kelley, 2008: 4-9;Churchill & Ulfstein, 2000: 623).
Delegation and voting procedures are two widely discussed sources of organizational autonomy (Hooghe & Marks, 2015). Delegation of decision-making power to institutional non-state agents, like secretariats, creates institutional agents that may drive progressive cooperation (Biermann & Sieberhüner, 2013;Mai & Elsässer, 2022). For member states, it increases the risk of agency slack (Hawkins et al., 2006), because it introduces a source of organizational activity not directly arising from their action. Voting procedures allow member state bodies to adopt decisions against the preferences of some members (Cooper et al., 2008: 506). However, even unanimity or consensus decision rules affect the prevailing constellation of power and interests, because they convey power to weaker actors. Hence, the choice between majority voting and unanimity or consensus reflects the balance between the risk of undesired IO decisions and the risk of blockade, not between autonomy and non-autonomy.
There are several other sources of organizational autonomy. First, path dependence may generate IO autonomy even in the absence of delegation or voting. Due to their lasting nature, organizations are generally known as 'carriers of history' (David, 1994). Organizational path dependence occurs if preceding organizational decisions shape subsequent ones. Past decisions may stabilize collective choice in indeterminate decision situations with multiple equilibria (Battle-of-the-Sexes situations, Snidal, 1985: 931-936). They may also affect organizational development and shape opportunities for organizational adaptation to new circumstances. Frequently, it is easier to complement existing arrangements with new ones than to replace them entirely. This strategy of 'layering' is well known from historical institutionalist analyses (Mahoney & Thelen, 2010: 16).
Second, deeply engrained fundamental organizational norms introduce a systematic selection bias into the organizational decision process. Defining organizational purpose, for which the member states invest an IO with action capability (Hooghe & Marks, 2015: 311-312), emphasizes specific problems, like climate change. Therefore, other issues do not gain intra-organizational relevance in this particular IO, except as constraints of IO action. Fundamental organizational norms influence systematically how organizational purpose is pursued, as Graham and Serdaru (2020: 680-682) demonstrate for different IOs involved in climate finance.
Third, collectively agreed decision criteria help accommodate divergent preferences in repeated decision situations because they provide common points of reference for the appraisal of decision proposals (Gehring & Ruffing, 2008). Pursuing incompatible preferences is rendered difficult, because disadvantaged or indifferent members will resist. In climate funding, for example, industrialized countries as contributors tend to advocate restrictive funding decisions, whereas developing countries as recipients favor more flexible ones. However, neither side is likely to agree on outcomes that are less beneficial than rule-based decisions. This effect is reinforced in decision processes involving specialized committees or other IO bodies with a focus on particular subsets of issues, e.g. on scientific, technical, or legal matters (Bradley & Kelly, 2008: 15-25, Green & Colgan, 2013. Hence, IOs may gain considerable autonomy from member-dominated decision processes. Even fully member-dominated IOs without delegation to secretariats may develop distinct organizational rationales (cultures) according to which IO decisions are made.

Authority in practice: what the climate-IO can do to influence its environment
The Climate-IO gains action capability from its authority to adopt decisions with external effects mainly in two ways. First, it acts through COP decisions with external effects for member states (and possibly other addressees). By concluding the UNFCCC (Brunnée, 2002), the Kyoto Protocol (Oberthür & Ott, 1999), and the PA (Rajamani & Bodansky, 2019), the member states have authorized the COP to flesh out the skeletal framework of these treaties through secondary rules in numerous areas (Depledge, 2017). They have even authorized the COP to "exercise such other functions as may be required for the implementation of [the PA]" (PA art 16.4.b, see also UNFCCC, art 7). As the Kyoto Protocol has been abandoned since, current organizational activities are conducted under the UNFCCC and the PA. The Climate-IO has made use of its authority primarily by elaborating specific rules for several specialised organisational schemes, of which three areas stand out, namely the financial mechanism, market mechanisms, and compliance management. Other areas, like loss and damage (PA, art 8) or the technology mechanism (UNFCCC, art 4) are less well developed (Obergassel et al., 2021;Oh, 2020). Second, the Climate-IO acts through operational decisions of designated bodies under these schemes. Whereas COP decisions often emerge from highly contentious and time-consuming political negotiations, the specialized organizational schemes are designed to implement previously achieved political compromise.

Creating and operating the financial mechanism
The COP has made extensive use of its authorization under the UNFCCC to adopt decisions on the financial mechanism (UNFCCC, art 11.1). Based on initial member state agreement to assign the operation of the financial mechanism to the Global Environment Facility (GEF) (Bodansky, 1993: 536-537), it concluded an extensive memorandum of understanding with the GEF, which defines funding policies, eligibility criteria, and even a redress mechanism for aggrieved developing countries (Decision 12/CP.2). It also adopted extensive policy guidance for GEF funding (Decision 11/CP.1), which is regularly updated.
To complement the traditional GEF focus on mitigation activities, it subsequently established two other funds with a focus on adaptation of developing countries to climate change and assigned their operation to the GEF, namely the Special Climate Fund and the Least Developed Countries Fund (Decision 7/CP. 7; Ciplet et al., 2013: 60). In these areas, the Climate-IO acts as an 'orchestrator' (Abbott et al., 2015) of another IO, namely the GEF. Two additional funds established by COP decisions operate within the UNFCCC institutional framework (Ciplet et al., 2013, 61-63). The Adaptation Fund supports adaptation projects in developing countries (Decision 10/CP.7) and is designed to be largely independ- The operation of this broad and complex financial mechanism requires extensive continuous implementation decisions with significant effects on member states-donors and recipients alike. Whenever the Adaptation Fund Board or the Green Climate Fund Board as subsidiary bodies of the COP decide on the application to finance a climate-related project in a developing country, the Climate-IO exercises political authority. States and private actors contributing to these funds sacrifice unilateral control over these assets and authorize the respective bodies to allocate this money according to organizational procedures-even if they earmark their contributions for certain tasks and thus limit the organizational autonomy of funding decisions (Graham & Serdaru, 2020). As of 2021, the Adaptation Fund has financed more than 100 adaptation projects with a volume of about US$ 830 million (Decision 13/CMA.3). The GCF has approved 190 adaptation and mitigation projects and programmes in 127 countries with a volume of about US$ 10 billion (Decision 6/CP.26), which will avoid over 1.8 billion tonnes of CO 2 emissions and increase resilience for over 498 million beneficiaries (GCF Handbook 2021, iii). The two fund boards also exercise political authority when deciding about the accreditation of multinational, regional or national implementing agencies, which are eligible for conducting funded projects. In contrast, project decisions of the GEF-operated funds do not reflect Climate-IO action because the GEF is an independent organisation with its own membership.

Creating and operating article 6 market mechanisms
The COP has also exercised authority through its decisions elaborating the details of several market mechanisms under the PA, which allow member states and private entities to cooperate on emission reduction projects (Depledge et al., 2022;Obergassel et al., 2021). It has adopted extensive rules for the cooperation mechanism of article 6.4 PA (Decision 3/CMA.3, Annex), which closely resembles the Clean Development Mechanism (CDM) of the Kyoto Protocol. A share of benefits will continue to be levied for the Adaptation Fund. The COP has invested a Supervisory Body with far-reaching competencies to make implementation rules and procedures and decide on the approval of projects. It has also established an emission transfer system that allows transferring emission reduction certificates (so-called 'internationally transferred mitigation outcomes', ITMOs) among member states (PA, art 6.2). Furthermore, it has extensively regulated the nature of ITMOs ('real, verified, and additional'), conditions to be fulfilled by member states for participation, and administrative, tracking and reporting obligations of participating states (Decision 2/ CMA.3, Annex).
The operation of the article 6.4 mechanism will require continuous implementation decisions with effects for member states and other actors desiring to use the mechanism. The newly established Supervisory Body is empowered to decide on the registration of projects, to certify realized emission reductions, to approve methodologies for the calculation of realized emission reductions, and to accredit technical agencies as evaluators of cooperation projects (Decision 3/CMA.3, Annex). With more than 8.000 projects approved between 2008 and 2021 with an emission reduction potential of about 9.5 billion tons of CO 2 emissions (CDM, 2021), the previous CDM had been very attractive (Yamin & Depledge, 2004: 159-187

Creating and operating the transparency and compliance mechanisms
Authorized under the PA (art 13.13), the COP has also acted on compliance management (Rajamani & Bodansky, 2019: 1033-1040. It has adopted an extensive set of rules governing the 'modalities, procedures and guidelines for the transparency framework' (Decision 18/CMA.1). These rules include extremely detailed provisions on the standards for assessing and reporting of information by the member states in their reports and define a multistage procedure of a technical expert review. They are highly relevant for member states because they specify mandatory reporting and transparency obligations, while a failure to adhere to these rules will trigger the compliance procedure (Decision 20/CMA.1, Annex, para.22 (a) (ii)). The COP has also adopted the modalities and procedures for the effective operation of the compliance committee as the main body of the compliance procedure (Decision 20/CMA 1).
The transparency and compliance mechanisms empower the Climate-IO to continuously adopt decisions on the performance of member states. Under these mechanisms, the Climate-IO will appraise the reporting of member states on their activities regarding mitigation, adaptation, and climate finance, but not the ambition or achievement of their Nationally Determined Contributions (NDCs, Decision 20/CMA.1, Annex). The newly established compliance committee will deal, inter alia, with cases, in which member states do not fulfil their reporting obligations or do not participate in the transparency mechanism. It is authorized to adopt 'decisions' (para.13), 'take appropriate actions' (para.30), recommend plans of action for the respective parties, and issue findings of fact. Although it shall not function as an enforcement mechanism or impose sanctions, findings of noncompliance may provide the foundation for ensuing political pressure or deprive noncompliant members of IO-generated benefits (Schermers & Blokker, 2011: 924-926). For instance, participation in the cooperation mechanism requires that member states have prepared, communicated and maintained their NDCs (Decision 3/CMA.3, Annex, para. 26). Emission trading presupposes that parties have provided the most recent national inventory report according to established procedures and guidelines (Decision 2/CMA.3, Annex). Projects financed by the Green Climate Fund must be in line with the applicant country's NDC. Unlike negative sanctions, such as trade restrictions, these effects do not constitute strong and punitive enforcement, because they are based on the eligibility and desire of member states to receive organizational benefits. However, they may provide sufficient incentives to motivate non-compliant parties to get back on track.
Considering the afore-mentioned COP decisions as Climate-IO actions requires some explanation. In contrast to the treaty management organizations of some other multilateral environmental agreements like the Montreal Protocol or CITES, the Climate-IO is not authorized to impose new obligations on member states. However, COP decisions have a 'fundamental legal character' (Yamin & Depledge, 2004: 546-547) and are adopted with the collective intention to shape international climate policy. Some of them, like compliance management, introduce specifications to existing obligations whose implications member states cannot easily avoid (Bogdandy et al., 2010: 11-15). Other decisions, such as those on the Green Climate Fund or the market mechanisms, establish specialized organizational schemes, which affect the costs and benefits arising from the sectoral international order (Schermers & Blokker, 2011: 760-761). All these decisions arise from a duly authorized organizational body and enter into force for all member states without ratification or opt-out opportunities. Identical decisions agreed upon by the member states outside the Climate-IO would have different effects.

Autonomy in practice: how the climate-IO exerts influence on organizational decisions
Organizational autonomy reflects organizational influence on IO decisions and arises from institutional constraints on intra-organizational action of member states and other actors. Decisions arising from the COP and from specialized organizational schemes are affected by different sources of organizational autonomy.

Autonomy in the realm of COP decision-making
In the absence of significant delegation and voting procedures, organizational influence on COP decisions arises mainly from path dependence and deeply engrained fundamental organizational norms. Voting does not play a major role because the COP adopts its decisions by consensus. Due to prevailing disagreement among member states, its rules of procedure are merely 'applied' except for rule 42, which would envisage majority voting in some cases (Yamin & Depledge, 2004: 431-432). Moreover, the climate secretariat does not seem to influence COP decisions systematically. Due to the high salience and tremendous costs of climate policy, it has been established as a small servicing unit with strictly administrative functions (Bodansky, 1993: 534). In this realm, it still appears to be largely a 'technocratic bureaucracy' (Busch, 2009, 251) that is "truly … first and foremost the servant of governments" (Biermann & Siebenhüner, 2013: 154) -in contrast to other areas, such as orchestrating non-state actors (Mai & Elsässer, 2022). COP decision-making is heavily path-dependent and reflects a typical characteristic of organizations (David, 1994). Member states do not operate on a clean slate, but in a highly institutionalized environment, which shapes the decision process and constrains their opportunities for action (Depledge, 2017). Actors are aware of these constraints. Typically, COP decisions cite relevant treaty provisions and previous decisions, even if adopted decades ago. For example, COP Decision 7/CP.26 of 2021 on guidance to the GEF refers, among others, to Decision 11/CP.1 of 1995 defining an initial set of guidelines.
The effects of organizational path dependence are illustrated by the development of the funding mechanism. Originally, contributing countries had insisted on assigning the financial mechanism to the mitigation-focused GEF (Decision 12/CP.2). When this arrangement became untenable, the member states did not revise their original decision, but created three additional funds with a particular focus on adaptation projects in developing countries, namely the Special Climate Fund, the Least Developed Countries Fund, and the Adaptation Fund. Moreover, operation of the Adaptation Fund was organized as an activity of the Climate-IO, not the GEF, because it seemed to be largely independent from contributions by donor countries (Ciplet et al., 2013: 62-63). Somewhat later, the COP established the Green Climate Fund, which expands Climate-IO funding activities to mitigation projects. Accordingly, collectively perceived needs for institutional adjustment are not primarily realized by redesigning earlier institutions, but by complementing them with additional ones. Due to this institutional 'layering' (Mahoney & Thelen, 2010: 16), the financial mechanism comprises currently five separate funds, which compete for voluntary contributions (Graham & Serdaru, 2020) -an arrangement that hardly reflects a grand design.
COP decisions are also heavily affected by fundamental organizational norms that shape opportunities for member state interaction. Arguably, the most important fundamental norm of the Climate-IO is the close link between environmental and developmental issues. From the beginning, the Climate-IO agenda does not only include mitigation issues, but also matters of adaptation and historical responsibility as an early result of the struggle between industrialized and developing countries (Bodansky, 1993: 470-474). The UNF-CCC requires developed countries to "assist the developing country Parties that are particularly vulnerable to the adverse effects of climate change in meeting costs of adaptation to those adverse effects" (art 4, para. 4). It affirms that "responses to climate change should be coordinated with social and economic development in an integrated manner" (preamble). It notes "that the largest share of historical and current global emissions of greenhouse gases has originated in developed countries" (preamble).
The close normative link between environmental and developmental issues has a number of long-term implications. First, the broad political agenda attracted many states that emphasized development over mitigation. Consequently, cumbersome decision-making among almost 200 states has always burdened the Climate-IO. This contrasts with the narrow, mitigation-focused approach of the Montreal Protocol, which had started with a limited club of some 24 mostly industrialized member states (Sebenius, 1991: 117). Current initiatives to establish limited climate clubs (Obergassel et al., 2021) constitute selective attempts to overcome this adverse effect. Second, the broadened agenda allowed developing countries to pursue their developmental interests as a separate goal, not merely as a restriction to mitigation action, reflected in the agenda topic of 'loss and damage' and COP decisions to establish adaptation funding (Ciplet et al., 2013: 50-53). Moreover, the financial mechanism operated for more than two decades without significant mitigation obligations for developing countries, whereas the financial mechanism of the highly successful Montreal Protocol has been accessible only for those developing countries that have accepted emission reduction obligations (Hermwille et al., 2017: 160). Third, the UNF-CCC reflects a clear distinction between developed countries and developing countries, of which only the former were committed to establishing national mitigation policies. This distinction contributed to the breakdown of the Kyoto Protocol and was overcome only by replacing the Protocol with the entirely new Paris Agreement.
Hence, even in the area of COP decisions, the Climate-IO has acquired a considerable degree of organizational autonomy. It operates according to a distinct organizational rationale that arises from path dependence and fundamental organizational norms. Even though COP decisions reflect consensus of all member states, they are difficult to explain without considering the organizational constraints, under which the member states operate within the COP.

Autonomy in the realm of specialized organizational schemes
The Climate-IO further gains considerable organizational autonomy through its specialized organizational schemes. These schemes have several features in common. First, decisionmaking authority is assigned to specialized IO bodies composed of a limited number of individual experts from developing and developed countries. Members are nominated by their constituent state groups and have alternates from another country of the same group. Second, the schemes operate according to criteria established by the COP and further developed by its subsidiary bodies. These criteria provide common points of reference for all actors involved in collective decision-making. Third, the schemes involve technical evaluation components that provide a fact-based foundation for organizational decisionmaking. Fourth, delegation of selective tasks to secretariats reinforces largely technical and criteria-based decision-making. These features are designed to promote problem-oriented decisions. They do not remove conflict but change the mode of conflict resolution. Taken together, they limit the room for coalition-building and political bargaining, thus pushing decision-makers toward considering the merits of the case in light of established criteria.
For example, decisions of the Green Climate Fund are made by the GCF Board composed of twelve members from developed countries and twelve members from developing countries nominated by their respective constituencies. Whereas the Board may use a highly restrictive voting procedure (GCF/B.23/23, Annex III; Kalinowski, 2020: 6), it tends to adopt decisions by consensus. Funding proposals undergo a standardized and consistent technical review process comprising four stages (see GCF/B.11/04,[2][3][4]Decision B.7/11,Annex VII). First, the Fund Secretariat conducts a completeness check of documentation. Second, the Secretariat conducts a substantive assessment of how the proposed project performs against relevant criteria using information from the 'accredited entity', which will conduct the project. Third, the independent Technical Advisory Panel (TAP) established by the Board provides an assessment of the expected performance of the project or programme. Fourth, the Board renders the final funding decision.
Numerous guidelines, procedures and funding criteria compiled in the regularly updated 'GCF Handbook' provide common points of reference for all actors involved. For example, funding shall be divided evenly between adaptation and mitigation projects, and at least 50% of adaptation funding shall be allocated to least developed countries, small island developing states and African countries (Decision B.07/11, Annex XIV). The prime indicator for mitigation projects is the projected lifetime emission reduction. For adaptation projects, it is the expected change in loss of lives, value of physical assets, livelihoods, and/ or environmental or social losses. Moreover, projects should trigger 'paradigm shift', catalyse impact beyond a one-off investment and must be in line with nationally determined contributions (NDCs) (Decision B.22/15,Annex VII). Investment criteria indicators shall explicitly guide "(i) the Board when approving projects; (ii) the independent Technical Advisory Panel (TAP) and the Secretariat when reviewing projects; (iii) and the accredited entities (AEs) when developing project/programme proposals" (ibid.).
The new article 6.4 cooperation mechanism is similarly organized (Decision 3/CMA.3, Annex). Its Supervisory Body is composed of twelve experts from regional groupings acting in their personal capacity (i.e. not as state representatives). It shall decide by consensus, but may vote (three-fourths majority), if it cannot achieve consensus. The cooperation mechanism will operate along the lines of the CDM established under the Kyoto Protocol (Gehring & Plocher, 2009). Approved activity-specific methodologies provide criteria for determining the amount of additional emission reductions generated by a mitigation project. Numerous 'methodologies' adopted under the CDM and compiled in a voluminous 'methodologies booklet' may be applied also under the new cooperation mechanism. Project assessment involves technical evaluation by a private technical agency accredited by the Supervisory Body. At the end of a project, the agency evaluates and the Supervisory Body certifies realized emission reductions. The comparatively strong CDM-Secretariat further limits the room for political manoeuvring (Michaelowa & Michaelowa, 2017: 254-256). Hence, the process is highly technical and criteria-based.
The new transparency and compliance mechanism is also designed to operate as criteria-based and technical as possible. Extremely detailed COP provisions on mandatory reporting obligations provide a tight set of criteria. The 199 paragraphs long COP decision on 'modalities, procedures and guidelines for the transparency framework' (Decision 18/ CMA.1) reflects the collective intention to strictly limit the room for creative reporting or tacit escape from NDC commitments. Technical Expert Review (TER) teams composed by the Secretariat from a roster of member state-nominated experts will examine submitted information and identify "areas of improvement, constituting preliminary 'recommendations' (for 'shall' provisions) and/or 'encouragements' (for non-'shall' provisions)" (Decision 18/CMA.1, Annex, para.162d). The reports are then publicly scrutinized in a written question and answer phase and a subsequent working group debate. Cases, in which parties fail to submit their NDCs or mandatory reports or to participate in the transparency mechanism, will be addressed by the compliance committee composed of twelve experts acting in a personal capacity (Decision 20/CMA.1, Annex).
Hence, the Climate-IO has acquired extensive organizational autonomy in the area of decisions arising from these specialized organizational schemes. Decisions follow an organizational rationale that is first and foremost based on organizational criteria, not on the aggregation of member state preferences. While member states play an important role, e.g. as project applicants or when electing board members, organizational decisions are largely determined by the characteristics of cases and organizational rules and procedures, not by case-specific member state preferences. Selective involvement of a secretariat reinforces this effect.

Conclusion
The joint organizational component of the UNFCCC, the Kyoto Protocol, and the Paris Agreement is not merely a secretariat plus an arena for member state interaction. It has become an international actor in its own right with considerable power and significant autonomy, despite its peculiar nature and the central role of its member states. We can grasp its organizational effects, if we analyse it as an IO. Organizations become actors in their own right, if their members authorize them to adopt decisions with external effects, and if they gain a minimum of autonomy. These two dimensions of organizational agency are applicable to member-dominated IOs like the Climate-IO. How an IO can act with external effects depends on its decision competencies. How an IO influences organizational decisions depends on the rules and procedures, which shape and constrain organizational decision processes. Sources of IO autonomy vary across IOs and even across areas of IO activity. Accordingly, secretariat influence on organizational decisions is not a precondition for organizational agency, but one possible source of organizational autonomy among others.
The theoretical concept of IO agency helps identify the sources, effects, and limits of the action capability of the Climate-IO. The member states have empowered the COP as the supreme decision-making body to adopt extremely wide-ranging governance decisions that become effective for member states without ratification. Based on the skeletal rules of the constitutive treaties, the COP has used its powers primarily to establish several organizational schemes that affect the costs and benefits of membership. In other areas, the COP has not acted decisively at the time of writing due to diverging member state preferences; in yet other areas, in particular regarding emission limits, change of treaty provisions, and appraisal of NDC ambition, it is not authorized to act at all. Under some of the organizational schemes established by COP decisions, especially the climate funds operated by the Climate-IO and the transparency and cooperation mechanisms, specialized bodies are authorized to adopt numerous case-specific decisions with immediate effects for their addressees, thus creating additional organizational action capability.
The concept of organizational agency helps to identify the sources and effects of organizational autonomy as well. The Climate-IO has acquired organizational autonomy to a considerable extent. One cannot infer its decisions only from knowledge about constellations of preferences and power among member states. COP decisions are heavily influenced by the highly institutionalized setting from which they emerge, despite the fact that they are consensually adopted by the member states. A specific organizational rationale arises in particular from previous decisions that create path dependence and from deeply engrained fundamental organizational norms that provide the stage for negotiations among member states. Decisions of organizational schemes arise from even more heavily institutionalized settings. Procedures are designed to promote technical and criteria-based decision-making and provide participation rights for non-state actors. They are intended to preclude decisions based on preferences and power, thus deliberately increasing organizational influence.
Prevailing analytical approaches cannot readily grasp these organizational effects. The regime perspective (Levy et al., 1995) conceptualizes MEAs as institutionalized cooperation arrangements concluded by member states, while the regime complexes approach extends this perspective to the interplay of international and transnational regime (Keohane & Victor, 2011). These approaches underestimate the organizational implications of MEAs. We complement them with an organizational perspective that demonstrates how MEAs become semi-autonomous international actors in their own right with a separate capability to act according to a distinct organizational rationale. Likewise, negotiation analysis examines bargaining processes among states (Oberthür & Groen, 2018), identifies possible future topics of COP activity (Kinley et al., 2021), and explores the role of the conference president as agenda manager and mediator (Park, 2016). Generally, it focuses on areas available for collective agreement through negotiated preference aggregation and adopts the perspective of negotiators. Our organizational perspective complements this view in two ways. It elucidates post-agreement IO activities based on previous organizational decisions, such as the operation of an international fund or the Clean Development Mechanism, which diminish the role of bargaining processes and preference accommodation. And it emphasizes the influence of institutional constraints on organizational decision processes arising from procedures, institutionalized decision criteria, or previous organizational decisions, within which intra-organizational activities of member states and other actors are inevitably embedded. Generally, such institutional constraints reflect an organizational rationale that supports actors pursuing compatible preferences, while hampering the advocates of incompatible ones, whether member states, the conference president, an MEA secretariat, or a non-governmental organization.
International treaty management organizations like the Climate-IO are particularly widespread in global environmental governance. Despite their limited bureaucracy, they become important governors of their respective issue-areas. To understand their role and influence in world politics, we need to conceive them as dynamic organizations with action capability and some autonomy. Accordingly, we should examine more closely the power resources of these organizations (and their limits) as well as the sources and effects of their autonomy.
Funding Open Access funding enabled and organized by Projekt DEAL. We gratefully acknowledge funding by the Deutsche Forschungsgemeinschaft (DFG) under Grant number 443226341.
Data availability Data sharing is not applicable to this article as no datasets were generated or analysed during the current study.

Declarations
Conflict of interest All authors declare that they have no affiliations with or involvement in any organization or entity with any financial or non-financial interest in the subject matter discussed in this manuscript.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.