Country differentiation in the global environmental context: Who is ‘developing’ and according to what?

Several multilateral treaties and International Governmental Organizations have introduced different legal obligations for countries based on the developing/developed (or equivalent) dichotomy. Such differentiation can (re)produce a range of material and symbolic consequences for those labelled developing or developed. Much has been researched about this topic in the environmental regime yet an important gap remains: what does this differentiation look like empirically? This article answers this question through a qualitative analysis that compares about two dozen multilateral environmental agreements (MEAs) on (1) how they specify what makes a country be developing (or not) and (2) the result of this choice, that is, exactly which countries are labelled developing under each MEA. The research reveals at least four important points: (1) the absence of any converging approach to classifying countries in the global environmental context; (2) almost 1 out of 4 countries in the world have mixed classification (developing or developed depending on the MEA); (3) ‘switching’ groups is relatively infrequent, but can be both moving to or away from the developing label; and (4) most countries with mixed classifications appear to be comfortable in the situation. This research contributes to a finer-grained understanding of differentiation in global environmental governance.


Introduction
The division of the world between 'developed' and 'developing' countries has been present in the international system since the end of World War II. This dichotomy has been used as the base for differentiating countries in legal documents, i.e. establishing different rights and/or responsibilities for countries depending on which of the two groups they belong to. Such division is easily found in the context of International Governmental Organizations 1 3 (IGOs) and multilateral treaties in a range of different issue areas, such as trade, development assistance, and environmental matters.
This article seeks to answer two key questions: (1) in the context of multilateral environmental agreements (MEAs), which criteria are used to classify countries as 'developing' (or 'developed') and, as consequence, (2) which countries are labelled 'developing' (or 'developed')? 1 To find these answers, this article qualitatively compares and analyses data from selected MEAs, engaging with cases of explicit and formal (legal) country differentiation, as opposed to implicit and/or colloquial uses. In doing so, this article speaks to (1) country differentiations in environmental governance and also to broader questions over hierarchy--especially those interested in how and why hierarchies are deliberately erected by specific actors (see Lake, 2011Lake, , 2017aLake, , 2017bMattern & Zarakol, 2016;Zarakol, 2017;Fehl & Freistein, 2020)--and (2) to the challenge of classification and (self)identification of/in the social world (e.g. Bowker & Star, 1999;Hopf, 2002;Broome & Quirk, 2015;Michaelowa & Michaelowa, 2015). The novel empirical findings set out below open new paths for research into country differentiation/classification in the international system in general and MEAs in particular.
This analysis reveals four main findings. First, there is no standard approach to classifying countries in the global environmental context. In fact, there is no convergence over which criteria or thresholds are used in this space to label countries as developing. Borrowing from Wendt (1992), developing countries are basically what individual MEAs make of it. The second finding relates to the reflection of this dissonance in the criteria and thresholds used for country classification. A country list comparison reveals a variation in the classification of over 40 countries: almost one quarter of all countries in the world have a mixed classification. They are developing or developed depending on the list. Third, changing classification and 'switching' groups within an existing list is relatively infrequent, especially when the criteria are based on self-identification. Yet, there were a few noteworthy instances where this has taken place, with countries asking to be moved to or from the developing country classification. Finally, some countries (mostly non-G77 members) have prompted their own mixed self-identification, coming forward as developed in some documents and developing in others. This signals the existence of distinct calculations vis-à-vis self-identification and recognition of the potential benefits of being developing (or developed), such as skirting compulsory financial burdens, seeking recognition of a 'higher' status as developed, and political complexities particular to each document, etc.
To develop this analysis, this article is divided into five parts. Following this Introduction, section two provides a background to the case selection. The next two sections are the heart of the paper, exploring the empirical data gained by analyzing the selected MEAs. Section three explores the criteria used by the selected MEAs to classify countries. Section four looks at the resulting classification from such criteria. It compares which individual countries are considered developing or not pertaining to global environmental governance. In these latter sections, special attention is given to classification in three cases: (1) the Montreal Protocol on Substances that Deplete the Ozone Layer; (2) the United Nations Framework Convention on Climate Change (UNFCCC); and (3) the Convention of Biological Diversity (CBD). The paper concludes with thoughts on the importance of country classification in the analysis of differentiation in the environmental context.

Case selection: method
The selection of documents for this analysis derives from existing research (Farias & Roger, 2023) on the use of developing/developed classification language found in documents in the International Environmental Agreement Database (IEADB). This research revealed that of the 1305 MEAs signed between 1945 and 2020, 2 only 81 (6%) MEAs mentioned this dichotomy. 3 Drawing from this latter group, the first step in this research was narrowing down the number of documents for analysis. Of the 81 documents, 52 were original treaties, 12 protocols, and 17 amendments. The next step focused on identifying MEAs whose text indicated differential treatment that could be considered compulsory action. In this pool, the focus was on documents that indicated a strong impact for differentiation, i.e. it set distinct levels of financial or legal commitments. This sat in opposition to language where differentiation was simply recommended, for example, where differentiated treatment was undertaken only if/when a developing country requested it, or when developed countries were directed that they should consider differentiated treatment or take into account, whenever possible the situation of developing countries when making national decisions. If any passage in a document suggested compulsory action, the document was included in this sample. This process narrowed the list of 81 documents down to 23 (Box 1). It should be noted that protocols and amendments that simply repeated what was already said in the original treaty (vis-à-vis developing/developed countries) were not included. The United Nations Environment Program (UNEP) was not added to the analysis because (1) it is not in the IEADB and (2) this research was not able to identify which criteria it used (if any) nor a list with individual countries' classification. 4 The United Nations Development Program (UNDP) bears some connection to this discussion, as it the organization 'responsible' for the Sustainment Development Goals (SDGs) and has its own list of developing countries. It is referenced sometimes in the text as an additional layer of information but not explored in depth since it is not a MEA (and not in IEADB).
The next sections compare and analyse these cases in relation to the criteria design (how countries are differentiated) and outcome (which countries are considered developing or developed).

Criteria: designs
This section focuses on the classification of countries under the developing/developed dichotomy or equivalent terms in MEAs. Its goal is to describe, compare, and analyse the criteria used by over a dozen MEAs. 5 Analysis of the selected cases revealed three main approaches to determining which countries were entitled to differentiated treatment. These are internal, external, and hybrid. For this analysis, 'internal' criteria means that the MEA has developed its own unique way of classifying countries under the developing/developed dichotomy. 'External' indicates the choice to copy an existing classification, created by another IGO or treaty. 'Hybrid', as the name indicates, combines both internal and external criteria.
One of the first assessments from the comparative analysis is that there is no standard approach to classifying countries in the global environmental context. Before detailing the variation, it is important to highlight three points. First, not all MEAs use the developing/ developed dichotomy (or equivalent terms). In fact, use of this dichotomy is an exception in MEAs. Out of the total 1305 MEAs, 511 were original treaties, 228 protocols, and 566 amendments. Of the first group, only about 10% (52) used this world division. Of these, more than half had over 100 signatures, signalling that country differentiation is disproportionately used in MEAs that involve larger groups of states (which is to be expected, given their more global reach). For these large MEAs (as well as newer and more ambitious MEAs), Farias & Roger (2023) have demonstrated that differential treatment plays an important role for participation: when there are no provisions for differentiation, greater 'depth' reduces participation, whilst when such provisions are present, it becomes possible to sustain high levels of participation and depth. 6 Second, this research identified cases that refer to the developing/developed country distinction in their text but that have no criteria to separate the groups, i.e. the text mentions the distinction but has no formal rule/structure to allocate countries into each group. As per their documents and/or secretariats, this was found to be the case for: the International Convention for The Protection of New Varieties of Plants; the Barcelona Convention; the International Tropical Timber Agreement, UNCLOS, 7 and the Protocol to the Convention on the Prevention of Marine Pollution by Dumping of Wastes and Other Matter. As an example, the BSR Secretariat responsible for the Basel, Stockholm, and Rotterdam Conventions informed that their texts 'do not define the term "developing countries," nor have the respective Conferences of the Parties adopted guidance on this matter'. 8 This invites further empirical research over why and how certain MEAs have managed to not (yet?) formally differentiate countries notwithstanding explicitly different legal implications, which could possibly be due to lack of interest in--or of political capacity to--change the status quo, internal case-by-case resolutions, etc.
Third, the role of the Group of 77 (G77) is important, as empirically detailed in the next section. Since its creation in 1964, the G77 has been the convergence point for countries that politically self-identify as developing, including in universal environmental negotiations (Kasa et al., 2008;Vihma et al., 2011;Blaxekjaer & Nielsen, 2015;Bueno, 2020). Notwithstanding the cacophony of uses and definitions of the developing country label, membership to the G77 has remained remarkably stable, with only a handful of countries leaving it. 9 Yet this research was unable to find any MEA that made G77 membership an explicit criterion for determining developing countries. Therefore, the G77 is an actor in negotiations but G77 membership is not used as a formal classification reference in MEAs. One possible explanation for this nominal absence is that when the G77 is directly involved in negotiations of environmental agreements, the criteria used to differentiate countries are designed in such a way that all G77 members are considered as developing. However, further research is necessary for finding if this explanation holds, or if G77 membership is simply correlated to countries sharing a common identity and/or interests.

Internal criteria
For MEAs, adopting internal criteria for differentiation is more of an exception than a rule. This research identified two cases with unique approaches to classifying countries (as developing/developed) in MEAs: (1) the Montreal Protocol and (2) the CBD. Along with the UNFCCC ( to be further discussed as a Hybrid case), they share a common element: their approaches to classification emerged from negotiations among member countries during the creation of the document itself or in subsequent negotiations as part of regular meetings-justifying the choice to analyse these three MEAs with greater detail.
The Montreal Protocol was established to tackle the impact of harmful substances on the ozone layer. Signed in 1987, it was created under the 1985 Vienna Convention for the Protection of the Ozone Layer, yet only the former introduces a developing/developed country differentiation structure. Article 5 ('Special situation of developing countries') of the Protocol lays out the general different treatment for developing countries. This explains the equivalent use of 'Article 5 countries'/'non-Article 5 countries' to designate developing/developed parties. The Article sets out two conditions for different treatment as follows: 'Art 5.1. Any Party that is a developing country and whose annual calculated level of consumption of the controlled substances in Annex A is less than 0.3 kilograms per capita on the date of the entry into force of the Protocol […]' (emphasis added). As such, it introduced two conditions for different treatment, where each is necessary but neither is sufficient: (1) being 'a developing country' and (2) demonstrate eligibility under an 'objective' criterium based on per capita consumption of certain substances. Yet, the Protocol did not establish the parameters for the first condition.
In the Montreal Protocol's 1st Meeting of the Parties (1989), a decision was made regarding the clarification of terms and definitions of 'developing countries': Decision 1/12E listed 130 countries 'considered developing countries for the purposes of the Protocol', although no details were given on what element(s) was used to inform the list. In the 3rd Meeting of the Parties (1991), this issue was explicitly addressed. Decision III/5 established that (1) from then on, the consideration of requests by States for classification as developing countries would be on an individual basis and only when such requests were made (e.g. Malta, Bahrain); (2) it would accept Turkey's classification as a developing country, and (3) an open-ended Working Group of the Parties would be set up 'to study and fully define the criteria which will be applied in the future in case of applications for classification as a developing country for the purpose of the Montreal Protocol'. This Group's recommendation was presented at the 4th Meeting of the Parties (1992) and recommended 'that no criteria for future classification as a developing country' be adopted and that the Parties should consider individually applications for classification as developing countries 'as and when such applications are made' (emphasis added). As of 2022, the Montreal Protocol has 147 developing and 51 developed parties (UNEP, n.d.). 10 Eleven requests for reclassification have occurred over the years: seven countries asking to be considered an Art. 5 Party (developing) and four applying to be removed from the developing list. 11 The CBD, signed in 1992, found its own route to a developing/developed classification. 12 The Convention's Art. 20 § 2 establishes that 'developed country Parties' should provide new and additional financial resources to help developing country Parties meet the agreed costs of implementing measures to fulfil the obligations of the document. In other words, being designated as developing means the country has no compulsory obligation to make financial contributions. The CBD's approach to classification was to establish 'a list of developed country Parties and other Parties which voluntarily assume the obligations of the developed country' (CBD, Art 20 § 2, emphasis added). Thus, the responsibility of classification is placed on countries, which must come forward individually to self-identify as developed. The list of developed Parties was established in 1994 and updated in 2006. The Convention determined that the developed country list would continue to be open to any country wishing to come forward as such. However, no country has asked to be added to this list. As the next section will detail, a close observation of the list reveals several noteworthy absentees.
The Montreal Protocol and the CBD's own paths to differentiating countries have the advantage of allowing more room for adaptation to new circumstances over time. However, by itself, this approach has not meant an absence of contestation: countries have negotiated to push or halt reclassifications. As the next section will detail, there are material and symbolic power considerations at play when countries try to push or halt change to a 'higher' (or 'lower') status in each MEAs hierarchical setting. Finally, it is worth adding that several UN bodies (indirectly related to environmental matters) rely on an internal ('technical') decision-making process for classifying countries, each one embracing different criterium and producing different country lists. Here, country lists are presented via a 'top-bottom' process, as opposed to being negotiated among countries 'bottom-up' (see Farias, 2022: 2-4). Such is the case for the UNDP/SDGs, UN Department of Statistical Division (UN-DESA) and the UN Standard Country for Statistical Use list (a.k.a. UN M49) --none of which have criterium/a to justify their differentiation nor are intended to have legal purposes (idem). As the next subsection will show, these lists matter for the environmental space, as they have been chosen as external sources of country classification by MEAs.

External criteria
Many MEAs have chosen to not develop their own method for classifying countries as developing/developed. Instead, they have chosen to copy another organization's existing approach. This is not unique to the environmental context (Farias, 2022). In fact, this is a frequent practice among IGOs and treaties of all topics and is also used by many countries when determining eligibility for development assistance and/or Generalized System of [trade] Preferences (GSPs). Analysis revealed there is no single go-to source for copying differentiation lists in the environmental context. In other words, MEAs using existing external approaches to differentiate countries do not draw on the same source. This reflects the bigger picture: a dissonance among universal IGOs and treaties over which approach to the developing/developed classification is chosen. No single IGO claims to have (or is perceived as having) the authority to define which countries are developing or not. Yet many IGOs are considered an authority on the topic, creating the heterogeneity of criteria and lists (Farias, 2022: 7).
This research found that both the Minamata Convention on Mercury and the Convention to Combat Desertification in those Countries Experiencing Serious Drought and/or Desertification Particularly in Africa use the same classification as the UN-DESA, while the IRENA copies UN's M49 differentiation list. Since 2021, the developing/developed dichotomy is no longer present in this list, which now only separates countries by geographic regions (UN, n.d.). Both UN-DESA and M-49 approaches have not relied on any formal criteria for differentiation. Rather, from internal bureaucratic processes that simply allocate countries to each group in a top-bottom process designed for analytical purposes only (i.e. not legal). The World Bank's quantitative GDP per capita country ranking (high-, medium-, and low-income groups) has been embraced by the International Treaty on Plant Genetic Resources for Food and Agriculture, and the WHO Framework Convention on Tobacco Control. However, since 2016, the World Bank also no longer uses these income level groupings as proxies for developed (high-income) and developing (low-and midincome) countries (see Fantom & Serajuddin 2016;Farias, 2019). Nevertheless, the World Bank's old approach continues to be embraced by these said MEAs. Finally, the Hazardous Wastes Protocol (1996) 13 established in its Art.1 that members of the OECD are developed countries 14 and those which are not are developing. Unlike the G77, OECD membership was never based on self-identification as 'developed', although it quickly become colloquially known as 'Club of the Rich'. The expansion of OECD membership to Latin America in the past 15 years, including the recent entry of Colombia and Costa Rica and ongoing negotiations with Brazil and Peru (as well as Bulgaria, Croatia, and Romania), will increasingly put pressure on using this organization as an 'easy' proxy for developed countries. Regardless of the IGO chosen as reference for differentiation, this research has found no explanation for how (or why) each MEA chose to borrow its specific external criteria.
The 'outsourcing' of country differentiation can be explained through IGOs' role in the classification and organization of information and knowledge. As argued by Barnett & Finnemore (1999: 707), IGOs' (perceived) authority can derive from both '(1) the legitimacy of the rational-legal authority they embody, and (2) control over technical expertise and information', wherein 'the ability to classify objects, to shift their very definition and identity' is one of their greatest sources of power (pp. 710-711). While it is 'simpler' for an MEA to use an existing approach to differentiating countries, this choice also entails problems, as changes and reclassifications that make sense for the original IGO might have negative (and unintended) spill-over effects to an MEA's member countries--as exemplified by the case of growing OECD membership and changes made in the World Bank and with the UN's M49 lists.

Hybrid criteria
The UNFCCC's country differentiation is the most well-known case in the environmental context. Established in 1992, it mentions 'developing country Parties' and 'developed country Parties' numerous times. But the most recognized division falls under the Annex and non-Annex countries' lists 15 . Bodansky (1993: 506-8) details the different criteria proposed for differentiating countries at the time of the UFCCC negotiations, with the negotiating committee deciding to 'use lists rather than definitions to fix the scope of application of the Convention's specific commitments' (p.507). There are two Annex lists: Annex II lists developed countries, Annex I is made of all Annex II countries plus countries with economies in transition (EIT Parties). 16 EIT parties sit in limbo: they are neither defined as developing nor developed. 17 UNFCCC's Art 4(2) speaks of developed country Parties and other Parties included in Annex I (emphasis added) as opposed to the simpler wording of: "Parties listed in Annex I. Bodansky (1993: 508) explains: Th alternative namely, to refer to the Parties listed in Annex I without characterizing them as 'developed' was not used, because some developing countries, for political reasons, wished the Convention to state explicitly that the countries bound by the specific commitments are 'developed', thereby maintaining the developingdeveloped dichotomy which was central to the NIEO [New International Economic Order].
In the past 30 years, the Annex I list has grown, with the addition of non-EIT countries and (former) non-Annex countries that petitioned to change lists. A noteworthy case of the latter is Kazakhstan, which in 1997 requested to be moved to Annex I. As Depledge (2009: 279) explains, this 'proposal met with opposition when it was formally considered at COP-5 in 1999, with some influential developing countries fearing the precedent that 15 In practice, 'non-Annex' and 'non-Annex I' encompass the same group of countries: i.e. 'developing' (under the UNFCCC). 16 Only those listed in Annex II are mandated to finance UNFCCC's mechanisms. 17 In the 1990s, a few IGOs and treaties adopted specific classifications for post-Soviet countries (e.g. IMF, between 1993 and2003). The UNFCCC is the only notable case of use in a universal treaty; UN-DESA uses the category only for analytical purposes (i.e. no legal implications) in its annual World Economic Situation and Prospects report. would be set if Kazakhstan were allowed to join Annex I'. Concern over precedent meant that this country's proposal could have prevailed if put to vote, but 'the reluctance of parties to resort to a vote under the climate change regime meant that this was not politically feasible'. As of 2021, the country remains non-Annex in the Convention. 18 Such political considerations have (re)enforced the tendency for a locked-in status of non-Annex I countries, especially G77 members.
While the Annex I list has grown, Annex II has remained (practically) unchanged since 1992. 19 The baseline criteria, of OECD membership in 1992, chosen to designate countries as developed was relatively simple. Thus, differentiation in the UNFCCC was initially structured via a 'reverse' and indirect self-identification approach. Instead of setting out criteria for defining which countries were developing, it started by establishing those that were not. Countries can petition to change their listing but cannot be moved without their consent. 20 As such, the UNFCCC developing/developed country division started with an external criterion (OECD membership in 1992) but has evolved into a unique ad hoc list. Thus, change is legally possible but the reality of political pressure against voluntarily 'graduating' from non-Annex to Annex (I or II) has made it less likely.

Criteria: outcomes
Not surprisingly, variation in the criteria used for classifying countries has resulted in individual countries being labelled differently. This is what happens in the global environmental context. There is visible heterogeneity in the labelling of a considerable number of countries. Drawing on the previous sub-section, the following analysis is based on the comparison of seven lists: two use internal criteria (Montreal Protocol and CBD), four source external criteria (UN-DESA, UN-M49, World Bank income groups, and OECD membership), and one uses a hybrid approach (UNFCCC). G77 membership and the UNDP/SDG developing countries list are mentioned to further illustrate the complexity of country classification.
Focusing initially on the three lists using internal criteria and the UNFCCC reveals two nodes of homogeneity. The first pertains to the developing countries, which overlap strongly with G77 membership. As previously stated, G77 membership has not been found to be explicitly used in any documents analysed as a criterium for developing country classification. Nonetheless, G77 membership is implicitly present and strong: (1) All 134 G77 members are developing in the CBD and UNFCCC lists; and (2) All but three G77 18 Kazakhstan is an Annex I Party for the Kyoto Protocol but is a non-Annex Party under the UNFCCC. 19 Turkey is the only odd case. As one of the OECD's founding members, Turkey did not agree with the idea of membership to this organization serving as the criteria for developed countries. At the time it 'unsuccessfully proposed using economic criteria, rather than membership in the OECD, to differentiate between developed and developing countries' (Bodansky, 1993: 507;see Schneider, 2017). The country only joined the UNFCCC in 2004, following a decision adopted in the Marrakesh COP (2001), when it was deleted from Annex II. However, it is still seeking non-Annex status (see UNFCCC, 2021). 20 As explained by Depledge (2009: 278), 'any party can propose to amend Annex I (or Annex II). That proposal must be adopted by the COP, either by consensus or by a three-quarters majority of parties, which ensures that parties cannot unilaterally delete themselves from the annexes. The amendment then enters into force automatically 6 months after being communicated to the UN Secretary-General. However, parties may notify, in writing, that they do not accept the amendment, which provides a safeguard against the status of parties being changed against their will'. members are developing under the Montreal Protocol: only Azerbaijan, Tajikistan, and the State of Palestine are both G77 members and non-Article 5 Parties (developed).
The second node pertains to developed countries. As previously mentioned, the CBD's approach is based on developed countries' self-identification. This arrangement has created an odd situation if one interprets countries not coming forward as 'developed' as being implicitly labelled developing. So far, only 25 countries have come forward as developed. Among the countries who have not (yet) self-identified as developed in the CBD are numerous OECD and EU members. The most curious case is Liechtenstein: the country with the highest GDP per capita in the world. 21 Drawing from an email communication with a CBD Senior Legal Officer in October 2021, 'in practice, the Global Environmental Fund (GEF) does not extend financial assistance' to Liechtenstein or any one of these countries: Bulgaria, Croatia, Cyprus, Estonia, Hungary, Latvia, Lithuania, Malta, Poland, Romania, and Slovakia. However, as these countries have not asked to be on the developed list they do not have to contribute financially to the CBD. A similar scenario can be teased out of the UNFCCC. The number of Annex I countries has grown since 1992, with the addition of eight new non-EIT members. 22 Yet, the list of Annex II countries those with compulsory financial contributions to the implementing the objectives of the UNFCCC has not expanded. 23 An interesting line for further investigation is understanding why some developed countries have resisted self-identification in such settings.
The CBD developed countries and the UNFCCC Annex II lists overlap very closely. Yet some interesting discrepancies arise when individual cases are compared (Fig. 1). Three countries (Czech Republic, Monaco, Slovenia) have come forward to self-identify as developed in the CDB but have not asked to be placed on UNFCCC's Annex II list. The EU, while a signatory to the CBD, is also not officially present in the list of parties voluntarily assuming obligations to the CBD, and the US is not a party to the CBD. Figure 2 provides another way of observing variation among countries' (self-)classification. It starts with a comparison of two dozen selected countries that have not come forward as developed in the CBD. It then compares their classification under the Montreal Protocol and the UNFCCC (Annex I vs. non-Annex lists)-with the UNDP/SGD classification and G77 membership used as additional information. This exercise teases out interesting discrepancies concerning which countries are not classified as developed in the environmental context. The overlaps (and lack thereof) in how these countries self-identify in the Montreal Protocol, UNFCCC, and CBD show that countries can embrace competing identities in global environmental governance. The puzzle for further inquiry is what is/are the main element(s) that explain a country's choice of one identity over another in different agreements? Possible answers for future investigation include free-riding as well as other dynamics 24 --such as status-seeking, path dependency, and domestic and/or external pressures to maintain a (lack of) legal commitment contained in a particular agreement, etc.-all which feed into broader discussions over differentiation, classification, and hierarchy.
Finally, it is worth comparing all internal and external lists mentioned in the previous section. Some caveats are warranted before engaging in this comparison. As presented in the previous section, the lists draw on diverse sources and understandings over how to classify countries as developing or not. Labelling a country as developing can refer to an 'objective' criterium (e.g. GDP per capita), a subjective criterium (e.g. identity), a combination of criteria (objective and/or subjective), or no clear criteria. While they all seek a similar outcome the classification of countries using developing/developed labels each list is designed for internal use. Akin to meta-analyses, this paper is aware of the heterogeneity regarding specific definitions adopted for developing. Nevertheless, this comparison is justified because all of the lists share the same goal of classifying countries for differentiation purposes, and are used in the context of global environmental governance. For example, being a member of the OECD or ranked high income by the World Bank are not formally the same as being classified as a developed country, but the fact that they have been adopted as proxies for differentiation justifies their presence in this comparative space. Consequently, the exercise below is meant to detail variation in individual country classifications within the environmental context while being aware of each list's peculiarities.
As Table 1 shows, both the number of countries considered developing and developed in these lists vary significantly. For example, a developed countries list using the CBD as reference has only 25 countries, while the one informing the WHO Framework Convention on Tobacco Control (which relies on the World Bank) has over 60. In tandem, these two lists are also the most distinct in relation to the number of countries that are not developed: 170 versus 136. By itself, this should serve as an indication of the different, perhaps even conflicting, conclusions that can be drawn when using the developing/developed dichotomy depending on the database used.Of the 193 UN members in these lists, 25 fall homogeneously under the developed label, 25 and 116 under the developing one. 26 Thus, 52 countries, roughly one-quarter of all UN members, have a "mixed" classification. An analysis of this group shows that none of them have chosen to come forward as developed in the CBD. Six of them Estonia, Hungary, Latvia, Lithuania, Poland, and Slovakia fall under the developed label in all lists aside from the CBD's. Croatia, Cyprus, and Malta are not developed under the CBD but are also not OECD members. The next ten countries are the most mismatched in their classification, as detailed in Table 2.
The last 24 countries in the mixed classification group are labelled developing in all but one of the lists. The data shows that they are all part of the developing (or not developed) group in the CBD, and also the UNFCCC (non-Annex), the UNDP, and the UN-DESA lists. There are some other interesting points of convergence among these two-dozen countries. Four post-Soviet republics are developing (or not developed) in all lists but are non-Article 5 (developed) countries in the Montreal Protocol. These are: Azerbaijan, Kazakhstan, Tajikistan, and Uzbekistan. 27 The UN-M49 list is the only one analysed where all former Yugoslavian republics and former Soviet republics in Europe are labelled developed. This means Albania, Bosnia and Herzegovina, Moldova, Montenegro, North Macedonia, and Serbia are developing in all lists but the UN-M49. If the World Bank's income thresholds are used to classify countries as a proxy for developing/developed, then a dozen countries are included in the mismatched group. In other words, they are high income yet labelled developing in all other analysed  ' (or 'not-developed') in the UNFCCC, Montreal Protocol, CBD, and UNDP. *G77 members; **In the UNFCCC, Turkey is currently awaiting a decision over its request to be removed from Annex I and to acquire non-Annex I status (see UNFCCC, 2021)   three of its members Colombia, Costa Rica, and Mexico are considered developing in all other lists. As said, the developing label is in the eyes of the beholder.

Conclusion
This paper sought to contribute to an empirical gap in the understanding of country differentiation in the international system. Focusing on the context of global environmental governance, it asked: which countries are considered 'developing' (and according to what)?
The response was relied on a qualitative comparative analysis of MEAs that established different compulsory obligations for countries classified as developing or developed.
The results revealed an absence of any converging approach to classifying countries in the global environmental context. This resonated with variation in country differentiation in the international system as a whole, as no IGO is the authority in this space. Unlike WHO for health or WTO for trade, there is no go-to IGO for development. The conflicting understandings over what constitutes 'development' directly impact how development levels are assessed. It can be based on a parsimonious economic understanding like GDP per capita and/or a few additional indicators. But 'developing' can also represent an identity, an indication of belonging to a group of have-nots whose best chance at changing historically unbalanced structures is by leveraging power as a group. Or, on the flip side, be perceived as or used to reproduce the civilized/uncivilized hierarchy. The analysis of MEAs reflects this challenge, also demonstrating that 'objective' criteria for differentiation are not less sheltered from controversy. Criteria can be clearly defined in terms of boundariese.g. 'all below threshold X'-yet there is always subjectivity in choosing which criteria should be used and why. Thus, power dynamics are inextricably attached to any criteria chosen, so the lack of (apparent) conflict in classification does not mean a lack of power considerations by countries. From this angle, it is natural that any proposed definition for who should be benefited from CBDRs is bound to be contested, yet not necessarily based on the same reason.
Once the different lists were compared, a considerable number of countries ended up having mixed classifications. A few were found to have opted/requested to be reclassified, but perhaps most importantly, most appear to be unbothered. This points to the need for further analyses over why countries might choose to keep or modify their classification (in general and in MEAs). Finally, it signals that contention over differentiation in the international system and in environmental governance is far from over.
Funding Open Access funding enabled and organized by CAUL and its Member Institutions.
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/.