The endurance of the G77 in international relations: South–South ideology and voting at the United Nations 1970–2015

Abstract The voting record of states of the global South at the United Nations General Assembly indicates they are dissatisfied with the US-led liberal international order. Against existing interpretations, this article challenges the notion that states belonging to the Group of 77 (G77) express discontent because they are illiberal and undemocratic. Instead, the article argues that the G77 is composed of a diverse group of states influenced by a common South–South ideology. This foreign policy ideology has a distinct intellectual history and conceptual morphology, grounded in common experience of colonial domination and international peripheralisation. These arguments are tested using a series of multiple regression models, controlling for illiberal characteristics of states and examining the reciprocal influence between G77 membership and voting stance at the United Nations. Disaggregation of General Assembly resolutions and analysis of the text of General Debate speeches corroborates the argument that a coherent set of shared ideas shape how global issues are conceptualised and framed by members of the G77. The results are consistent with the argument that states of the G77 have socialised one another into a shared South–South ideology and that domestic illiberalism is insufficient to explain why they express dissatisfaction with the US-led international order. Ideologies of foreign policy originating in the global South, therefore, should not be overlooked as an influence on world politics.

This article seeks to qualify this interpretation, arguing that it understates degree of cohesion among states of the global South, who are the main occupants of the 'non-Western pole' at the UN. As Bailey et al. indicate, these states are diverse in terms of their political institutions and the ideological character of their regimes. Nonetheless, they have retained a high degree of unity within the UN General Assembly because they share elements of a common outlook on international relations. States of the global South are not opposed to the Western liberal order simply because they are all illiberal or undemocratic; the global South includes both authoritarian regimes and democracies. Lack of domestic liberalism is an incomplete explanation for the South's opposition to aspects of the liberal international order. Rather, they are influenced by a shared ideology of foreign policy that was established during the process of decolonisation and the subsequent efforts of post-colonial and economically underdeveloped states to establish and defend their sovereignty. It continues to enable coordination among states of the global South, despite half a century of change in world politics. Examining the case of the Group of 77 (G77) therefore provides a test-case for evaluating the influence of non-Western and non-Northern ideas in contemporary international relations.
The institutional expression of this South-South ideology is the G77, which was established in 1964 at the end of the first session of the United Nations Conference on Trade and Development (UNCTAD), leading to the subsequent establishment of G77 chapters within other specialised agencies of the UN. Alongside the Non-Aligned Movement, with which it is closely connected, the G77 became one of the primary diplomatic coalitions of states of the global South, nearly doubling in size to 134 states. Although New Zealand was one of the states involved in the formation of the group, and the former Yugoslavia and Romania were once members, the G77 is now composed of non-Western post-colonial states in Asia, Africa and the Middle East, as well as less economically developed countries in the Americas. Upwardly mobile states that have 'graduated' to the European Union or Organization for Economic Cooperation and Development (OECD) have usually left the group. Ministers for foreign affairs of the G77 meet before the regular session of the UN General Assembly each year, and the group issues frequent joint statements via particular member states' permanent missions to the UN.
Although focused originally on issues relating to economic development and the work of UN specialist agencies in this area, the G77 at the UN General Assembly also concerns itself with issues relating to peacekeeping and security. In a joint declaration in the year of the 55th anniversary of the G77 in 2019 under the chairmanship of the State of Palestine, constituent states reconfirmed the role of the group 'in providing the Global South with the means to articulate our shared vision, promote our interests and enhance our joint negotiating capacity within the United Nations system'. The group reiterated that its core concerns were: the imbalances in the global economy and the inequitable structures and outcomes in the trading, financial, monetary and technological systems that led to the establishment of our Group persist to this day and, together with the threats to the rules-based multilateral system (G77, 2019) Yet this seems to understate the coherence of the states of the G77, which according to Panke's data voted alongside one another over 75% of the time from 1999 to 2010. Moreover, in some sessions of the General Assembly, the G77 achieved greater coherency in voting than smaller and more homogeneous regional organisations such as the African Union (Panke, 2013: 288). The G77 seems able to achieve a level of coherency comparable to other groupings in the UN, despite its economic and political heterogeneity-which Panke does not control for in her analysis.
Against the claims that the states of the G77 display little cohesion or coherency in their voting behaviour at the UN or in their foreign policy preferences, this paper argues that although states of the G77 are a diverse group, they remain united by aspects of a common ideology. This ideology originated in opposition to Western colonialism and is based on defence of a certain conception of national sovereignty in combined with a critique of certain inequities of the liberal international order. It provides a degree of international unity among the states of the G77, despite their high degree of heterogeneity.
This article first provides a review of research on voting patterns at the UN and what they reveal about the foreign policy alignments of states, before developing an argument that underlines the ideological continuities within the global South within and beyond the United Nations. It then advances the hypothesis that members of the G77 vote differently to other states at the UN, even when confounding factors are accounted for. Next, a set of multiple regressions are presented to test these hypotheses for the period 1970-2015, and empirical results are discussed. These include a simultaneous equations analysis of the reciprocal relationship between G77 membership and UN voting position, evaluating the argument that G77 states have socialised themselves into a common South-South foreign policy ideology. The regression analysis of voting patterns is followed by an examination of the content of indicative UNGA resolutions and a brief quantitative text analysis of speeches at the UN General Debate 1970-2015, which provides corroborating evidence for the link between South-South ideology and foreign policy stances take at the UN. Concluding, the article makes the case that the influence of ideas originating in the global South should not be overlooked and that scholars should be cautious about dismissing scepticism about the US-led international order as stemming from authoritarian illiberalism.

Literature review
Research on voting positions within the UN General Assembly is well-established within international relations. In an early study of voting at the UN, Alker found two sets of divisions among states at the 1961 session: East-West and North-South (Alker, 1964: 647). Using factor analysis, Russett identified three divisive issues at the 18th session of the General Assembly in 1963: the Cold War, intervention in Africa and supranationalism. These issues grouped states into Western, communist, Western-leaning African, Afro-Asian, Arab and Iberian voting blocs with highly correlated voting patterns (Russett, 1966: 327). Russet suggested that voting data could be used as a source of information about the policy positions of states within the UN and the structure of the international system in terms of its polarity and major coalitions (Russett, 1966: 327).
Subsequent research explored the changing relevance of the East-West dimension of disagreement relative to the North-South dimension. Using multidimensional scaling and hierarchical clustering to identify voting blocs, Holloway (1990) found distinct and very stable Western and Warsaw pact groupings, with a large bloc of states of the global South containing both strongly pro-Soviet and moderately pro-Western states. In a subsequent study of the 'watershed' 46th session of 1991, Holloway and Tomlinson demonstrate that many former communist states altered their voting patterns, bringing about an end to the Warsaw Pact voting bloc. However, despite declarations of a 'New World Order' by US President George H W Bush, they found that 'the gap between the north and the south appears to have actually increased' (Holloway and Tomlinson, 1995: 242). Examining voting during the three sessions of the UN from 1991 to 1993 using factor analysis, Russett and Kim identified disagreements on three 'super-issues' of self-determination and disarmament, political rights and the Middle East (Kim and Russett, 1996). Comparing the three sessions to those held from 1983 to 1985, they found that the East-West had become less relevant and that the North-South 'overwhelmingly defines the terms of political debate in the General Assembly' during these sessions (Kim and Russett, 1996: 636).
Evaluating change over a larger series of sessions than previous analyses, Voeten used the W-NOMINATE scaling technique to provide a more accurate analysis of the dimensions of disagreement during sessions of the General Assembly up to 1996. He finds that a North-South dimension of disagreement emerged in the 1970s alongside the Cold War dimension, although communist countries and the South often voted alongside one another (2000: 197, 202). Voeten identifies continuities between the Cold War and post-Cold War periods, as states of the global South that tended to vote alongside the Soviet Union continued to vote on differently to the USA and its Western allies after the collapse of European communism (2000: 204-205).
A major step forward in the analysis of the expression of preferences at the UN was made by Bailey et al., who use an ordinal item response theory model. This model estimates an 'ideal point' for each state, on the assumption that state voting behaviour is influenced by its position on a single dimension of policy disagreement and the characteristics of the particular vote. By identifying identical resolutions across sessions, they are able to measure changes in states' voting behaviour over time. This allows them to generate an ideal point estimate for each state during each session of the UN General Assembly, reflecting a state's foreign policy preferences (Bailey et al., 2017: 435-437). They argue that the main underlying dimension of contestation at the UN concerns the legitimacy of the US-led liberal international order, state ideal points therefore measuring 'affinity with the Western-led world order' (Bailey et al., 2017: 431). Bailey and Voeten extend the model to generate a two-dimensional analysis, although the second dimension does not have a consistent interpretation from year to year. At various points, it relates to divisions 'lurking below the surface' of General Assembly debate, such as nuclear disarmament (Bailey and Voeten, 2018: 46, 54). They therefore conclude that, whilst secondary issues rise to prominence on occasion, there is a single major dimension of disagreement in the UN that reflects support for or opposition to the US-led international order.

Domestic and international dimensions of ideology
What then drives state preferences, as revealed by voting at the UN? Voeten argues that the answer is ideology. In his Ideology and International Institutions, he argues that much of the politics of international institutions can be understood in terms of a single dimension of ideological disagreement. He defines ideology as: [A] widely-understood set of interconnected propositions about how a set of issues should be resolved and who should resolve them. Ideologies have implications for what is (1) ethically good, (2) how resources should be distributed, and (3) where power appropriately resides. (Voeten, 2021: 21). For Voeten, ideologies play a central role in international relations. This approach echoes that of Owen, who argues that world politics has been driven by a series of contests between transnational ideological movements, and of Braumoeller, who uses case studies and a system-of-equations modelling approach to explore the efforts of the great powers to shape the ideological character of the international system (Owen, 2010;Braumoeller, 2012). In terms of the ideological contest that animates contemporary post-Cold War international relations, liberal internationalism is something of a status quo ideology. Liberal ideology has influenced major aspects of the international order and many specific international institutions, supported by the USA as a superpower as well as its liberal democratic allies (cf. Narizny, 2012). But, as a result of the end of the Cold War, contestation at the UN is 'no longer is defined by an ideologically consistent alternative to Western liberalism. Instead, a hodgepodge of countries that have little in common other than their opposition to Western liberalism occupy the non-Western pole' (Bailey and Voeten, 2018: 48).
In the twenty-first century, overlapping ideological tendencies of nationalism, statism and populism oppose liberalism without offering their own fully developed programme of proposals for how global issues should be resolved (Voeten, 2021: 22-23). Voeten follows Mudde in describing populism as a 'thin' ideology (Mudde, 2004;Voeten, 2021: 25). Mudde's analysis, in turn, draws upon Freeden's evaluation of nationalism as a 'thin-centred' ideology that cannot offer a comprehensive set of responses to the problems of politics, unlike fully fledged ideologies such as liberalism, socialism and conservatism (Freeden, 2002: 750-751). For Freeden, thin ideologies do possess certain core concepts, yet rely on concepts drawn from other ideologies and from culturally specific belief systems in order to provide concrete answers to political questions (Freeden, 2002: 750-751). Thin ideologies, on this diagnosis, lack the comprehensiveness, depth and universality of 'true' ideologies. This is consistent with Voeten's diagnosis that states influenced by these ideologies have little in common with one another but their rejection of aspects of the liberal international order.
On the question of the reasons why some states have adopted these oppositional international ideologies, Voeten is cautious. On the one hand, he argues that 'a country's ideological perspective is not independent from factors that other liberal scholars have identified as important for institutional politics, such as democracy, economic openness, and left-right ideology' (Voeten, 2021: 39). This is underlined by the observation that resolutions concerning human rights are an important source of division within the UN General Assembly, 'as they signify conflict between liberal and nonliberal states' (Bailey et al., 2017: 431). This suggests that states that do not conduct free and fair elections, that do not safeguard a liberal conception of human rights and that spurn economic liberalism should be expected to oppose liberal internationalism. On the other hand, Voeten notes that 'Foreign policy ideology does not always follow from domestic ideology' (Voeten, 2021: 146). Nonetheless, liberal attributes of states such as democracy and capital openness are indeed statistically associated with a voting record closer to that of the USA (Bailey et al., 2017: 446;Voeten, 2021: 38).
As well as domestic illiberalism, there are other reasons why states might not be content with the liberal international order. A state's foreign policy might be shaped by its level of satisfaction with the international status quo (Organski and Kugler, 1980;Tammen, 2008). All things being equal, states experiencing prosperity might be more contented (Sample, 2018). In a contribution to the liberal peace debate, Mousseau (2003) argues that prosperity conditions the effect of democracy on a state's preferences. Interacting measures of democracy and of economic development in an analysis of voting agreement, Mousseau finds that higher levels of democracy are associated with increased agreement for high-income states only. Mousseau argues that this pattern arises from the distinct cultural values found in societies where citizens can benefit from contractual exchanges in a regulated market (Mousseau, 2003: 488-490;Mousseau, 2009: 62). Social market democracies find it easier to reach agreement with one another, because their citizens share common values. Where a clientelist economy predominates, individuals within a society tend towards parochial value-systems-described by Mousseau in terms that recall Voeten's thin ideologies of nationalism, statism and populism (Mousseau, 2003: 490).
Existing scholarship, therefore, seems to explain the opposition of some states to the liberal international order in terms of those states' undemocratic political institutions, weak protection of human rights, clientelist political economies and lack of success within the global economy.

Ideology and South-South solidarity
These explanations, however, underplay the ideological commonalities in the foreign policies of states that criticise the liberal international order. Both democratic and non-democratic members of the G77 frequently vote on the opposite site of the USA in General Assembly resolutions. Domestic illiberalism cannot, therefore, provide a complete explanation for patterns of disagreement at the United Nations. Scholars of the liberal international order argue that it consists of institutionalised commitment to democracy, human rights, regulated free markets and 'principled multilateralism' (Lake et al., 2021: 232). We would therefore expect domestically liberal states to vote alongside one another on resolutions pertaining to aspects of this order. But these scholars also acknowledge that a central part of this order has been a specific set of security relationships among the high-income democracies of Europe, North America and East Asia focused on the USA (Lake et al., 2021: 232). Those states outside of this club have found much to criticise in terms of both American global leadership and what they perceive as the unequal, exclusive and intrusive character of the liberal international order.
To take one important example, in the 1990s and 2000s Brazil pursued a foreign policy that was distinct from that of the USA but which nonetheless advanced certain liberal goals and priorities. Under the Cardoso and Lula administrations, Brazil joined the Treaty on the Non-Proliferation of Nuclear Weapons, continued to promote disarmament and denuclearisation, and sought to take the initiative on the global issue of hungera foreign policy that was justified by appeal to the values of 'humanism' and 'solidarity ' (de Faria and Paradis, 2013;de Queiroz Duarte, 2017). These are not the features of an illiberal foreign policy, and Brazil seems to have been fully committed to 'principled multilateralism', yet during this period Brazil voted very differently to the USA within the UN.
These differences reflect a North-South divide, not just a divide between liberal and illiberal states. Even relatively liberal states of the global South have been critical of the actually existing liberal international order. As identified by Kim and Russett (1996) as well as Bailey and Voeten (2018), the primary divide in the UN since the end of the Cold War has been between the North and the global South. A majority of UN member states are members of the G77, which has its roots in mid-C20th attempts by states of the global South to speed up decolonisation, resist Cold War polarisation and renegotiate the terms of the global economy. Bivariate analysis by Lees (2021) shows that the gap in UN voting positions between G77 states and members of the OECD was just as large in 2018 as it was in 1980. Furthermore, Levene's test shows that the G77 has not been significantly less coherent as a voting group in the UN than the OECD states (Lees, 2021: 102). Panke's (2013) claim that the states of G77 struggle to coordinate on a common position therefore underestimates the coherence of the group. Given that the states that compose it are indeed quite diverse and that the G77 is not a deeply institutionalised body capable of strong enforcement, this coherence can be best explained in terms of a shared ideology of foreign policy. This ideology has been variously referred to as 'anticolonialism', 'nonalignment', 'positive neutralism', 'Afro-Asianism', 'Third Worldism', 'South-South solidarity', the 'structuralist perspective' and the 'development-multipolar perspective' (Roberts and Parks, 2007;Solarz, 2012;Mielniczuk, 2013;Lüthi, 2016;Lewis and Stolte, 2019). Alternatively, these aspects could be viewed as distinct but overlapping ideological currents within world politics. What matters is that there is a recognisable 'conceptual morphology', to use Freeden's term (2003: 750), to the political claims and arguments made by states of the G77. These arguments emphasise: the impermissibility of white minority rule; the need for rapid decolonisation and dissolution of European overseas empires; the need to renegotiate the unfair and inequitable organisation of the global capitalist economy; the moral and political obligation for high-income states to transfer of resources to address global poverty, the existence of which constitutes a violation of human rights; the need for developing and post-colonial states to cooperate; the commitment to a notion of sovereignty that proscribes external interference within a state's borders but prescribes external support; and opposition to the dominance of the great powers combined with support for multipolarity and multilateralism.
This constitutes a recognisable conceptual core to South-South ideology. Although articulations of this ideology often draw on other ideologies such as socialism and nationalism, it is not merely a thin ideology defined negatively by its rejection of liberal internationalism. As a foreign policy ideology, the perspective has an extensive intellectual hinterland. Notable thinkers that have influenced the development of a distinctive perspective of the global South include Argentinian economist Raul Prebisch, Indian Bengali poet Rabindranath Tagore, Indonesian novelist Pramoedya Ananta Toer, Pakistani development theorist Mahbub Ul Haq, Palestinian scholar of literature Edward Said and Filipino political scientist Walden Bello, among many others. 1 India's Nehru, Ghana's Nkrumah, Tanzania's Nyerere and Brazil's Cardoso contributed to the perspective both as intellectuals and as national leaders. As with other ideologies such as socialism, proponents advocated disparate strategies, ranging from Gandhian passive resistance to the strategy of violence espoused by Fanon (Steger, 2009: 139). Within the discipline of International Relations, dependency theory, structuralism, worldsystems and post/decolonial approaches represent attempts to develop aspects of the ideology as social scientific theories. As a foreign policy ideology, the South-South perspective is therefore considerably 'thicker' than has been recognised.
As well as drawing on a substantive body of thought, South-South foreign policy ideology rests on an established tradition of diplomatic practice. The Southern coalition had a precursor in the 1927 Brussels conference of the League Against Imperialism and Colonial Oppression. Almost immediately after decolonisation, Nehru's India used the platform of the General Assembly to criticise colonialism and racial inequality in South Africa (Mazower, 2009). Representatives from 34 contingents attended the 1947's Asian Relations Conference in Delhi, at which delegates of national independence movements debated the future of Asia and alternative strategies for opposing colonialism (Thakur, 2019). The subsequent Afro-Asian Bandung Conference of 1955 provided an opportunity for a generation of post-colonial leaders to assert their new-found political independence and collectively condemn colonialism and racism. 2 The Cairo Afro-Asian Peoples Solidarity Organisation conference of 1957-1958 featured tensions between communist and non-communist states, but led to the subsequent Tricontinental Conference of 1966 in Havana, which gathered left-wing states and anti-colonialist movements. But it was the leadership of an independent European communist state, Yugoslavia, alongside India and Egypt that led to the crystallisation of a movement based on the principles agreed in Bandung. The resulting Belgrade Conference of 1961 linked the two issues of nonalignment and anti-colonialism, asserting a conception of sovereignty as autonomy from the great powers (Mortimer, 1984: 11-14;Braillard and Djalili, 1986: 91).
The economic dimension of the South-South programme can be traced to arguments put forward by Latin American states in the post-war period. Within international institutions, they argued that the process of post-war economic reconstruction in Europe should lead to a commitment to address problems of economic reconstruction and development worldwide (Murphy, 1984: 25, 30-31). New postcolonial states were receptive to these claims and issued their own demands for compensation for the damages caused by colonialism. Latin American involvement in the Cairo Economic Conference of 1962 brought individuals closely involved with the Non-Aligned Movement together with structuralist economists such as Prebisch, who criticised the Bretton Woods system during his activist leadership of the Economic Commission for Latin America and the Caribbean from 1950 to 1963 (Mortimer, 1984: 16). The result was the issue of a Joint Declaration which set the agenda for the first United Nations Conference on Trade and Development in 1964, and the consolidation of the global South caucus in the General Assembly through establishment of the Group of 77 (Mortimer, 1984: 17).
The subsequent call at the Sixth Special Session in 1974 for a New International Economic Order by the South was ultimately unsuccessful, despite initial Northern engagement. Prashad (2007) argues that, in any case, the political project had already been defeated from within: anti-socialist leaderships seized power within the global South and crushed social movements that had mobilised around the causes of equality and anti-imperialism. Byrne agrees that, after a wave of coups in the 1960s, Third World internationalism lost its subversive character as a transnational movement of social movements and became a closed club of states legitimating one another's sovereignty (Byrne, 2019: 290-291). Nonetheless, attainting influence within multilateral institutions had always been central to the project and this focus of the early South-South coalition has endured. Even as states of the global South abandoned the more ambitious and utopian aspects of this project, patterns of cooperation persisted within the UN and other venues. South-South solidarity seems to have become engrained as an ideology of foreign policy, even as it declined as a mass political movement and basis for domestic political mobilisation.
Using Mann's terminology (1986), South-South solidarity transformed from a 'transcendent' ideology cutting across colonial social boundaries, to an 'immanent' ideology of political leaderships that found themselves in a marginal position in world affairs. Thus, Iida (1988) found that agreement among states of the global South rose during the 1980s, even after the diplomatic defeat of the South at the Cancun Conference in 1981. The endurance of a shared outlook or perspective explains why states once again invested in South-South diplomatic relations from 2003 onwards, following the rise of the emerging economies amidst continuing dissatisfaction with both the international economic architecture and American military unilateralism (Hurrell and Narlikar, 2006;Gray and Gills, 2016). Using spatial modelling techniques to examine the voting records of Brazil, Russia, India, China and South Africa, Binder and Payton find that the BRICS emerging economies and resurgent powers 'have begun to form a bloc of dissatisfied powers in the UNGA ' (2022: 393). Given that four out of five of these states are members of the G77 and that they have periodically taken leadership roles within the group, their findings are congruent with the argument that a common South-South platform still exists. The expansion of South-South trade and cooperation in recent years does not seem to have been accompanied by the sort of liberal conditionalities included in North-South bilateral and plurilateral agreements. Gamso (2019) interprets the absence of labour rights provisions in South-South trade agreements as in keeping with a legacy of non-interference provisions that stretches back to the formation of the Non-Aligned Movement and the Bandung Conference.
To summarise, the foreign policies of a majority of the states of the global South are influenced by a body of ideas about what is ethically desirable, how resources should be distributed and where power should reside in international relations. These ideas are sufficiently coherent and recognisable to constitute an ideology. States with quite different political institutions and domestic political ideologies, from liberal democracies to communist and ultraconservative states, appeal to these ideas. The tendency of states to vote in opposition to features of the US-led international order is not, therefore, simply the product of illiberalism, populism, nationalism or statism. These ideas might also exert an influence, but institutionalised patterns of diplomatic coordination through bodies such as the G77 have sustained the South-South perspective as a foreign policy ideology.
This argument leads to a falsifiable hypothesis: H1: Members of the G77, controlling for domestic illiberalism, have ideal points that on average indicate greater opposition to the US-led international order than non-members of the G77.
Membership of the G77 might not necessarily cause states to oppose US-led international order. States might self-select into the G77, joining the group because they agree with its shared political platform, and leaving when their foreign policy preferences change. However, as well as demarcating a distinct group of states within the United Nations system, nations making up the G77 might also socialise one another into a shared ideology of foreign policy. This implies a two-way causal influence between G77 membership and a state's position in UNGA debates, leading to a hypothesis that may be investigated through a simultaneous equations approach: H2: Membership of the G77 and UN ideal point influences one another reciprocally. Members of the G77 have ideal points that on average indicate greater opposition to the US-led international order than non-members, even taking account of endogeneity.
Issues of unequal economic development have been central to North-South relations, and dissatisfaction with the global distribution of income may be a major driver of voting behaviour in the UN. Therefore, an attempt will be made to replicate Mousseau's findings: H3: The influence of democracy on support for or opposition to the US-led international order is conditional on level of economic development.
The hypotheses will be tested using multiple regression analysis, using data from 1970 to 2015. The following section discusses the data employed in the analysis. The subsequent section discusses the statistical modelling strategy and the findings from a set of multiple regressions.

The G77 and voting at the UN General Assembly: operationalisation and data
The dependent variable is the single-dimensional ideal point of each state, as calculated by Bailey and Voeten based on a roll call of votes at the United Nations. 3 This variable is intended to measure support or opposition to the US-led liberal international order. Votes at the UNGA are not perfect as a measure of a state's revealed foreign policy preferences, especially as resolutions are non-binding and may not involve significant costs for a state. However, if votes at the UNGA had more significant material consequences, states might vote more strategically, for example engaging in 'horse-trading'. Because votes are primarily symbolic, the positions that states adopt can be interpreted as sincere with greater reliability, although an attempt is made in the analysis to control for the most salient forms of strategic voting.

Independent variables
The focal independent variable is G77 membership. Being part of the coalition is associated with different voting behaviour (P < 0.000, Welch's t). The example of Mexico is illustrative: as Figure 1 shows, its voting stance at the UNGA changing dramatically after it joined the G77 in 1965. Subsequent to leaving the G77 in 1995, Mexico altered its voting behaviour and adopted a stance closer to that of the USA. Figure 2 shows the variation in the ideal points of G77 states at five-year intervals. Although there are several outliers, the majority of G77 members have converged around a similar ideal point over time, and this common position seems to have remained stable. However, the bivariate relationship found in voting patterns may arise because states with certain attributes join the G77, and these attributes also influence a state's voting stance at the UN. If this were the case then the relationship between G77 membership and UN ideal point would be spurious. Table 1 shows the characteristics of G77 members compared to non-members for selected relevant characteristics. Although a diverse group, member-states are on average less economically developed, less democratic and larger in terms of population. Including a set of control variables in our regression analysis allows us to investigate whether, net of the influence of other factors, G77 membership is associated with a critical stance towards the US-led order in the UN. If they do not, then the similar voting behaviour of states composing the G77 might be adequately explained by their shared illiberalism after all, rather than the influence of a shared South-South perspective on foreign policy.
Lack of democracy might be a confounding factor for the relationship between G77 and voting at the UN, as undemocratic states might be both more likely to join the G77 and to oppose the liberal international order. Therefore, we include a measure of free and fair elections, the Varieties of Democracy (V-Dem) polyarchy variable, as a control (Coppedge et al., 2020). But even if elections are free and fair, a polity may be illiberal if the majority tyrannises minorities or if individual rights are not safeguarded. Therefore, we include a measure of political terror in our analysis, based on the Political Terror Scale version 1.2. This project defines 'political terror as violations of basic human rights to the physical integrity of the person by agents of the state' (Wood and Gibney, 2010;Gibney et al., 2019: 1). This is an ordinal scale that ranges from 1, where violations of physical integrity of persons ARE rare, to 5, where they affect the whole population. As the data for this variable are based on reports of Amnesty International, a leading liberal non-governmental organisation, this variable should hopefully track respect for a liberal conception of human rights. 3 The data were generously provided on request by Erik Voeten. As Bailey and Voeten state that 'one-dimensional ideal points are likely to be most useful for analyses in which a measure of a country's place in the international world order is needed' (2018: 54).
It is possible that the G77 states vote differently to other states on average because the group has contained many states subscribing to socialist ideology. Tensions between socialist and non-socialist states could provide an alternative explanation for North-South and East-West disagreement at the UN. A continuous variable measuring the importance of socialist ideology to the legitimation of state's  ruling regime was constructed by combining V-Dem measures of the extent to which a regime depends on ideological legitimation and extent to which its ideology is socialist.
As well as variables reflecting the liberal or illiberal character of a state, a set of political economy variables were included. High-income states might be more contented with an international order, whilst low-income states might be more dissatisfied. Therefore, GDP per capita, based on the World Bank's World Development Indicators, was included in the analyses (World Bank, 2019). To take account of Mousseau's arguments, GDP per capita and polyarchy were interacted to create a measure of whether a state is both high income and democratic. If the G77 is merely a collection of poor and undemocratic dissatisfied states, the relationship between membership and voting behaviour might be revealed as spurious. The population size of a country may also be a relevant factor for both G77 membership and UN voting, as the grouping may be composed of smaller states who are dissatisfied with dominance by the major powers.
Opposition to the US-led order at the UN might result from a particular kind of political economy. Economic rents derived from oil might motivate and sustain a state's antagonism towards the USA. The demand for a New International Economic Order was a consequence of the Oil Crisis and the attempt by resource-rich states to wrest economic power from high-income states and multinational corporations. The antagonism between the USA and oil-producing states such as Iran, Iraq, Libya and Venezuela has been an enduring feature of international politics for decades. We include a measure of oil rents as a percentage of GDP from the World Bank to assess whether tensions between the USA and oil producers drive the apparent effect of G77 membership. Additionally, several scholars of international relations have argued that states that are integrated into the global economy are more supportive of the US-led international order at the UN (Gartzke, 2007;Voeten, 2021: 37). To take this into account, we use the standard measure of capital openness created by Chinn and Ito (2006).
Bilateral relationships with the USA could potentially account for the voting behaviour of countries in the UN. Aid is used to shape coalitions and purchase influence, therefore a variable for US aid was constructed using data from the Foreign Aid Explorer of the US Agency for International Development. Security concerns loom large for many states and might hypothetically override ideology, encouraging client states to vote alongside geopolitical patrons such as the US. Notably, Mexico and South Korea, both states with close geopolitical connections with the USA, both left the G77. Data from the Correlates of War Formal Alliances data 4.1 were used to construct a US ally dummy variable indicating whether a state is a treaty ally of the USA in a given year (Gibler, 2009). Another way of examining security relations with the USA is to assess the number of American troops that state hosts. The variable US troops measures the number of troops hosted by a country, based on data gathered by Tim Kane (Kane, 2016). 4

Empirical findings
To identify problems that could potentially arise, a set of initial regression diagnostics were performed after a pooled regression. Results of a link test indicated problems of model misspecification, so skewed variables were transformed by taking the natural logarithm. A check for multicollinearity did not suggest serious issues, although unsurprisingly the variance inflation factors were high for the polyarchy and GDP per capita measures due to the interaction term. Four different panel data modelling approaches were adopted: regression with year-specific fixed effects; regression with a lagged dependent variable (LDV); regression using panel-corrected standard errors (PCSE); and a two-stage simultaneous equations model. Robust standard errors clustered on the country were used where appropriate and all independent variables were lagged one year. Kropko and Kubinec (2020) argue that fixed-effects OLS models remain one of the most useful tools of quantitative social science. However, they caution that, although two-way fixed-effects models that attempt to account for country-and year-specific effects are commonly employed in political science, inclusion of dummy variables for every time-period and every unit is statistically inappropriate. Kropko and Kubinec suggest that researchers to decide whether they are more interested in making cross-case or within-case comparisons, and adopt unit-specific or time-specific fixed effects as appropriate. Such models are included in the replication files for completeness, but for the main analysis this investigation employs decadal and regional fixed-effects in an attempt to enable comparison across and within countries, addressing the issue of unobserved confounding effects whilst avoiding the problems that Kropko and Kubinec highlight.
The LDV model is an autoregressive model, which means that the independent variables affect the long-term average value that the dependent variable tends towards. Our LDV model of UN ideal points therefore estimates the long-term average ideal point of states conditional on the control variables and membership/non-membership of the G77. This model addresses the issue of serial correlation arising from the tendency of states to maintain a similar foreign policy outlook and pattern of voting from year to year. The PCSE model accounts for both correlations across countries in specific years, and correlations between voting positions of individual countries across years (Beck and Katz, 1995).
It is possible and, indeed, theoretically plausible that G77 membership is endogenous to voting record at the UNGA. States of the global South may have joined the G77 because they were similarly dissatisfied with aspects of the US-led post-war international order. However, as this paper has argued, the formation of the G77 facilitated the development and diffusion of a common South-South perspective that influenced voting stances in the UNGA. To take account of this reciprocal influence, Keshk's implementation of a two-stage simultaneous equations model for data where one variable is continuous and the other is binary was employed (Keshk, 2003). This variety of simultaneous equations model has been used to examine the reciprocal relationships between inter-state trade and peace, and between democracy and civil war (Keshk et al., 2004;Choi, 2016).
For the fixed-effects, LDV and PCSE models, both a 'short' and a 'long' regression were run. Methodologists such as Achen (2005) have advocated for short regressions with fewer variables, arguing that interpretation of a model becomes increasingly difficult with each additional variable. Because of data availability problems, introducing additional variables also reduces the number of observations, which may also lead to more unbalanced panels. However, exclusion of a confounding variable as a control can result in inaccurate estimates of the effects of the independent variables included in the model. The long regressions include additional economic and political variables, as well as variables for alliance with the USA and receipt of aid from the USA. A state's relationship with the USA is likely to be highly endogenous to that state's stance on global issues. Therefore, the long regressions should only be interpreted comparatively, providing details on the correlation between G77 membership and voting at the UNGA adjusted for likely confounders. This helps answer the question of whether countries in the G77 have voted differently from other states on average, even if they have similar security relationships with the USA and receive similar levels of US foreign aid. Table 2 shows the results from our seven regression models. G77 membership has a significant effect on voting patterns in all models, net of other effects. The relationship between G77 membership Cluster-robust, panel-corrected or Maddala standard errors in parentheses. ***P < 0.01, **P < 0.05, *P < 0.1. and voting does not appear to be a spurious correlation that can be explained by the illiberal characteristics of states of the Global South, as these are controlled for in the models. The effect of membership is substantively significant: by construction the standard deviation of the UN ideal point measure is approximately one. In model 6, membership of the G77 is associated with a difference in voting stance approximately equivalent to that between Australia and Turkey or Colombia and Cambodia in the 70th session of the General Assembly. Mousseau's argument about the interaction between democracy and economic development is corroborated in most of the model specifications: democracy, measured by free and fair elections, is associated with opposition to the liberal international order for states with low levels of economic development. Yet the tendency for autocracies and lowincome democracies to vote against the USA and its allies does not 'explain away' the effect of G77 membership on voting behaviour. Additionally, states designated as having poor human rights records by Amnesty International seem no more or less likely to vote against the liberal international order, all things being equal. Figure 3 displays the effects of a one-unit increase in each of the independent variables on UN ideal point, based on model 6. For model 7, the simultaneous equations analysis, UNGA ideal point was used as the dependent variable in a first-stage ordinary least squares model and G77 membership was used as the dependent variable in the first-stage probit model. In the second stage, estimates of ideal point and G77 membership were used as instruments substituted for the endogenous variables in the models for one another, before Maddala standard errors are calculated (Keshk, 2003). The models for both variables include controls for time, economic development, polyarchy and their interaction. To meet the order condition, the model for G77 membership does not include capital openness, oil production or socialist ideology, whilst the UN ideal point equation does not include political terror or population size. 5 Figure 3. Coefficient plot of model 6 showing the estimate effect of a one-unit increase in each independent variable on UN ideal point.

5
Alliance and aid variables were not included in the simultaneous equations model because Keshk's implementation only permits two potentially endogenous variables, and other aspects of a state's foreign policy are likely to be endogenous to voting at the UNGA. Domestic-level factors are more plausibly exogenous. Results should be interpreted cautiously, as the firststage probit model seems to be very sensitive to multicollinearity, which is difficult to avoid due to the importance of the Results are consistent with the presence of a reciprocal relationship, but indicate that G77 membership is associated with a UN ideal point indicating greater dissatisfaction with the US-led international order even accounting for endogeneity.
The findings differ little in alternative specifications. The association between G77 membership and UN ideal point remains significant in models where country-specific effects are accounted for and in models where year-specific effects are accounted for. Post-regression diagnostics indicate that there are some outliers in our dataset. To guard against our results being distorted by influential outliers, jackknife resampling was conducted for models 1-6, with each country sequentially dropped from the regression model and the estimated coefficients averaged across repeats. The results were very similar to the original findings and G77 membership remained significant at the P < 0.01 threshold in all models.
To ensure robustness, models 1-6 were re-run sequentially with alternative operationalisations of several of the control variables. A variable measuring political liberalism from the Varieties of Democracy Project, operationalised primarily in terms of constraints on the executive, was used in place of the polyarchy variable. A measure of civil liberties, also from V-Dem, was used in place of both the polyarchy and political terror variables as an alternative indicator of liberalism. As many past studies use the Polity combined autocracy-democracy scale, this measure was used in place of the polyarchy variable derived from V-Dem (Marshall et al., 2019). The log of the number of US troops was substituted for the dummy variable measuring alliance with the USA. As an alternative to the continuous measure of socialist ideology, a dummy variable for communism was substituted. 6 Finally, the models 1-6 were re-run using a new dependent variable, percentage agreement between a state and the USA in roll call votes in the UN General Assembly. In all of these alternative model specifications, G77 membership was robustly associated with voting in opposition to the US-led international order.
It might be objected that the three modelling strategies employed do not adequately address the issue of serial correlation. Foreign policies of states are fairly stable from year to year, and Bailey et al.'s measure of UN ideal points is similar from year to year by construction (2017: 436). Therefore, the best way to evaluate whether G77 states really do vote differently from other states, all things being equal, might be to run separate cross-sectional regressions for each year that we have data for. Figure 4 shows the estimated effect of G77 membership on voting between 1970 and 2012, based on 43 cross-sectional regressions including a full set of controls. 7 The dotted lines in the figure show the 95% confidence interval. As can be seen, apart from in a few years where the effect is not quite distinguishable from zero using the standard threshold for significance, G77 membership is strongly and significantly related to voting behaviour in the UN. Indeed, the effect of G77 membership appears to grow stronger and clearer over time. Over the course of the period that we have data for, a state's attributes and its bilateral relations with the USA do not 'explain away' the effect of G77 membership.

The G77 and the substance of disagreement at the United Nations
This article has argued that there is a distinctive South-South ideology of foreign policy, with its own conceptual morphology distinguishing it from other ideologies such as liberalism and socialism. interaction of polyarchy and income. The model represents a tentative attempt to tease apart the relationship between UN ideal point and G77 membership, appealing to the theoretical argument that oil production, capital openness and government ideology all change more rapidly than G77 membership, which is fairly stable. As indicated by models 1-6, political terror and population size do not appear to directly affect UN ideal point, any effect is indirect through G77 membership. 6 Afghanistan 1978-1992, Albania 1956-1991, Belarus 1946-1991, Benin 1975-1990, Bulgaria 1956-1991, Burkina Faso 198-1987, Cambodia 1975-1989, China 1971-2015, Cuba 1965-2015, Czechoslovakia 1948-1990, Ethiopia 1974-1991, German Democratic Republic 1973-1989, Grenada 1979-1983, Hungary 1955-1990, Laos 1975-2015, Macedonia 1945-1990, Madagascar 1975-1992, Mongolia 1961-1992, Mozambique 1975-1990, North Korea 1991-2015, Poland 1947-1989, Romania 1955-1989, Russia 1946-1991, Somalia 1969-1991, Ukraine 1946-1991, Vietnam 1977-2015, Yemen People's Republic 1967-1989 It meets Voeten's definition of an ideology in that it offers a 'set of interconnected propositions about how a set of issues should be resolved and who should resolve them' (Voeten, 2021: 21). If this explanation is valid, then the content of the resolutions G77 members support or oppose in the General Assembly and the language that states to frame issues should provide corroboratory evidence.
Although they argue that a single dimension of disagreement over the merits of the US-led liberal order captures much of what is important in world politics, Bailey et al. (2017) also provide the means to generate separate measures for the position of states on multiple dimensions of disagreement. These separate dimensions are based on a classification of resolutions into six non-exclusive issue areas: colonial, disarmament, economic, human rights, Middle Eastern and nuclear issues. Most of these dimensions are quite highly correlated, but nuclear issues sometimes divide the General Assembly between larger and smaller powers rather than between North and South. Models 5 and 6 were re-run using the ideal points for colonial, economic and human rights resolutions, and G77 membership was again associated with greater disagreement with the US. G77 states have repeatedly supported, and the USA and its allies have repeatedly opposed, resolutions such as 77/174 'Towards a New International Economic Order' in 2022 calling for a redistribution of resources and a greater focus on the needs of developing countries. In another series of resolutions that G77 states have supported but the USA has opposed, the General Assembly as identified development as a human right on an equal basis with other human rights, for example in 77/212 'The Right to Development' and many previous resolutions. Simultaneously, members of the G77 have supported resolutions such as 77/ 214 'Human Rights and Unilateral Coercive Measures' which reassert state sovereignty and condemn the use of external pressures such as sanctions to enforce human rights.
Beyond voting on resolutions at the General Assembly, states signal their stances on global issues in the General Debate at the beginning of each session. Baturo et al. (2017) provide the UN General Debate corpus, comprising speeches from the start of every session of the General Assembly since 1970. They argue that, in these speeches, diplomats 'can more freely express their government's perspectives on issues deemed important' than in votes on resolutions (2017: 2). Moreover, the speeches contain information about how states conceptualise and frame issues that is pertinent for investigations into the role of ideas in shaping foreign policy stances. Following the analysis in Baturo et al. (2017), correspondence analysis was used to identify the relative emphasis of countries on different issues in their speeches at the General Debate. Replicating their methodology, a set of orthogonal dimensions were fitted to the data using the quanteda package in R (Benoit et al., 2018). Figure 5 is a wordcloud of the stemmed terms that load onto the first dimension, which is the dimension that explains the most variance in the textual data. Speeches that score highly on this dimension are likely to refer to poverty, imperialism, colonialism, racism, interference and development. This dimension plausibly relates to South-South ideology, as it is within this specific ideology of foreign policy that these concepts are linked. The prominence of this dimension within the General Debate corpus is consistent with the argument that South-South ideology is important in shaping debates at the UN. Although space precludes a full investigation of the correlates of the content of speeches, G77 membership was significant at P < 0.05 and P < 0.01 when models 5 and 6 were re-estimated with the primary dimension in the corpus of speeches as the dependent variable. This indicates that G77 states have tended to highlight a similar set of issues in their speeches at the UN, controlling for other factors.
The UN General Debate corpus can also be drawn upon to examine how states of the South frame issues of international concern in comparison to states of the North. Figure 6 is a keyness plot, displaying the trigrams that have the highest distinctiveness for G77 and OECD countries when speeches of members of each group are compared with speeches of members of the other from 2005 to 2015. 8 Trigrams are used to capture distinctive phrases not detected by the correspondence analysis technique used above; the 10-year span is chosen to examine whether the influence of South-South ideology persisted into the final decade of the period examined in the main analysis. As can be seen, aside from discussion of particular countries, members of the G77 have been more likely to discuss nonalignment, the right to self-determination and small island developing states. An extended list of 75 terms with the highest keyness for G77 members (provided as part of the Supplementary materials) includes 'interference internal affairs', 'least developed countries', 'South South cooperation', 'socio economic development' and 'landlocked developing countries'. By contrast, OECD states are much Chile, the only country that has joined the OECD but remained in the G77, has been categorized as an OECD state. more likely to mention terms relating to nuclear proliferation, European security, women and peace, the International Criminal Court and the Human Rights Council. Many of these textual differences seem to be linked to the influence of South-South ideology and liberal internationalist ideology on the two groups of states.

Conclusion
As Kim and Russett argue, the UN is a venue in which 'Different groups of states are able not only to be heard but also to assemble majorities on different kinds of issues that represent critical fault lines of global politics ' (1996: 641). The North-South divide remains one of these critical fault lines. Opposition to the US-led international order does not only come from illiberal and undemocratic states, but also from democratic states within the global South. Despite their differences, the states comprising the G77 share aspects of a common ideology of foreign policy with its own intellectual hinterland and conceptual morphology. This ideology is related to, but distinct from, other ideologies such as socialism and nationalism. It is a distinct phenomenon from populism, not least because the South-South ideology advocates rather than rejects multilateralism. As a result, membership of the G77 remains robustly associated with voting against the USA and its allies even when an extensive set of controls measuring domestic illiberalism and even bilateral ties with the US are included in a multiple regression model. The argument that domestic liberalism explains voting in the General Assembly is not falsified, but rather shown to be incomplete. Mousseau's argument that low-and high-income democracies have opposing preferences is corroborated, but the tendency for low-income democracies to be dissatisfied with the international order does not fully account for the North-South divide. It is also necessary to take into account the shared South-South, pro-development, anti-colonial ideology embodied in the G77.
As the yearly cross-sectional regressions demonstrate, the North-South divide has endured despite repeated declarations about the death or irrelevance of the global South. The evidence provides a further corroboration of Iida's argument that G77 consolidated itself as a coordinated voting bloc in the 1980s, well after the heyday of the demand for a New International Economic Order (Iida, 1988). Supplementary analysis of different dimensions of disagreement in General Assembly voting and of the content of speeches in the General Debate has corroborated the existence of a connection between a shared South-South ideology and the foreign policy positions taken by G77 states at the UN.
The analysis presented in this article suggests that states do not oppose the liberal international order simply because they are illiberal. The findings presented here suggest that domestic illiberalism is only a partial explanation for voting patterns at the UNGA. Greater attention should be paid to the ideas that shape the foreign policies of both democratic and undemocratic states of the global South in order to advance understanding of the global divisions that animate international politics.