全球变化知识资源中心

The Intergovernmental Panel on Climate Change (IPCC) has completed its Fifth Assessment Report (AR5). Here, we explore the social scientific networks informing Working Group III (WGIII) assessment of mitigation for the AR5. Identifying authors’ institutional pathways, we highlight the persistence and extent of North–South inequalities in the authorship of the report, revealing the dominance of US and UK institutions as training sites for WGIII authors. Examining patterns of co-authorship between WGIII authors, we identify the unevenness in co-authoring relations, with a small number of authors co-writing regularly and indicative of an epistemic community’s influence over the IPCC’s definition of mitigation. These co-authoring networks follow regional patterns, with significant EU–BRICS collaboration and authors from the US relatively insular. From a disciplinary perspective, economists, engineers, physicists and natural scientists remain central to the process, with insignificant participation of scholars from the humanities. The shared training and career paths made apparent through our analysis suggest that the idea that broader geographic participation may lead to a wider range of viewpoints and cultural understandings of climate change mitigation may not be as sound as previously thought.

The IPCC is widely accepted as the authoritative voice of scientific knowledge on climate change. However, what is often missing in the popular depiction of the IPCC as a scientific collaborative endeavour is a reflection on the array of social processes involved in its assessments. Thus, alongside the institutionalized scientific practices for writing, reviewing and revising the reports, there are social processes and scientific conventions for selecting authors and recognizing authoritative knowledge. These social conventions privilege certain institutional affiliations and pre-existing patterns of scientific collaboration and are tightly coupled with economic resources and political power.

Twenty years of social science research into the IPCC has identified two ongoing biases within the organization and its assessment reports: the disparity in participation between Northern countries and the global South, and a hierarchy in disciplinary knowledge¹. As the organization recognized during its First Assessment Report (FAR), unequal participation impacts the legitimacy of the IPCC process and the authority of its assessment products, with countries reluctant to accept assessments of climate change when their expert involvement is limited^{2, 3, 4, 5, 6}. Disciplinary biases, on the other hand, constrain how climate change is known and acted on, with only certain forms of knowledge and expertise authorized to construct a problem with global implications^{7, 8}. Given the hegemonic status of the IPCC within the production of climate change knowledge, exploring questions of participation and scientific legitimacy remains an important task.

Earlier insights on these issues have been largely qualitative in nature, providing detailed descriptions of the establishment of the organization and identifying the North–South divide structuring the Panel and authorship of IPCC assessments^{2, 3, 9, 10, 11}. Further research identified the scientific interests shaping the construction of climate change as a global political issue, highlighting the dominance of global climate models in Working Group I’s (WGI) conceptualization of climate change, and the natural sciences more broadly across the three WGs (refs 7, 8, 12, 13, 14, 15). More recently, a variety of statistical techniques have been employed to quantify the disciplinary and geographic bias within the IPCC and environmental science research more broadly^{16, 17, 18, 19}. The paucity of the social sciences has been of particular concern to those critical of the IPCC’s technical and managerial framings of the issue and seeking to broaden social understandings of the climate change problem^{7, 8, 20}.

Here we contribute to these studies by applying for the first time social network analysis to the writing team of WGIII’s contribution (that is, Mitigation of Climate Change) to the AR5. Although previous studies have illuminated the disciplinary biases within WGIII, there has been no detailed analysis of the actors authoring the report—the patterns of training, disciplinary backgrounds and institutions connecting this group of actors. Exploring these patterns provides evidence on how disparities between Northern countries and the global South structure the authorship of WGIII’s assessment, and may facilitate a deepened understanding of the controversy surrounding governments’ acceptance and approval of the final reports²¹.

Figure 1 represents the network of countries involved in the WGIII AR5 writing process, showing the patterns of connection between countries through the key moments in an author’s career trajectory, including the country they undertook their PhD and current country of work. These patterns reveal a more complex picture than a simple North–South divide. The figure indicates the existence of a US- and UK-based dominance in authors’ careers, identifying them as principal training sites for scientists involved in the production of the WGIII AR5. Previous studies have highlighted the US and UK dominance in the authorship of the report and identified statistically significant drivers of this, including per capita GDP (gross domestic product), English-speaking status, and levels of tertiary education¹⁷. The historic role these countries had in the development of early scientific interest in climate change and the establishment of the IPCC is also likely to be a factor^{12, 13, 22, 23}.

Lines indicate at least 2 authors connecting each country. Node size indicates the betweenness score and thus centrality in the network (see Methods). The number on each line, as well as line thickness, indicates the number of authors connecting each pair of countries. Data used to construct this figure can be found in columns D, E, G, and I of the ‘CV Data’ table in the Supplementary Database.

Another way of exploring the network of AR5 WGIII authors is through the institutions in which authors have worked and studied. Table 1 shows the top 30 institutions ranked by their betweenness scores. The analysis again reveals the dominance of US- and UK-based institutions but also identifies, perhaps surprisingly, the disproportionate role that international organizations play in connecting up the network, despite the fact that only 7% of authors work in international organizations (18/273). The World Bank is the most connected institution in the network, followed by the University of California at Berkeley, and with the UNFCCC (United Nations Framework Convention on Climate Change), UNEP (United Nations Environment Programme), FAO (Food and Agriculture Organization of the United Nations) and the International Institute for Applied Systems Analysis (IIASA), an international research institute, all in the top ten. This suggests that at some point in their career, substantial numbers of IPCC authors pass through these institutions in some capacity, a dynamic that is not captured effectively in the country network (although the relative prominence of Austria is an effect of IIASA being located there) and indicates that these institutions have an influential coordinating role in climate mitigation and policy research.

Full table

Exploring further the co-authoring network within WGIII enables us to think about pre-existing and ongoing patterns of collaboration amongst authors. As noted above, there are a small number of authors co-authoring regularly with one another, who are central to the network. Whereas the mean betweenness score for authors is 312, the median is 11, indicating a highly skewed distribution with a relatively small number of high betweenness individuals and most authors on the margins of the network. Taking a closer look at these ‘top 20’ indicates that although overall 73% of AR5 WGIII authors were new to the process²⁴, 85% of these 20 individuals have been in a previous assessment, and half of them have contributed to three of the five assessment cycles (see ‘Top 20 IPCC Part Analysis’ tab of the Supplementary Database). This might suggest that there are a small number of WGIII authors whose careers have become structured around the IPCC’s assessments process and in producing papers oriented towards the reports.

At the same time, the co-authoring network of the ‘top 20’ authors (Fig. 3) indicates that the writing collaborations of these individuals extend across the WGIII team, and are indicative of regional patterns of climate change research collaborations. Figure 3 is a two-degree network, with each line indicating that two authors have co-authored at least two papers together. The network shows that European researchers (largely excluding UK researchers) dominate the co-authoring network, and that their collaborations are as much with BRICS authors as US ones. This concentration of co-authoring among a small number of authors seems to reflect research funding criteria and institutionalized incentives to collaborate across EU boundaries and with developing and emerging economies. The network’s existence may be indicative of a core group within the AR5 WGIII writing team, composed of a small group of authors regularly co-authoring with each other and regularly contributing to the IPCC’s constructions of mitigation. However, further research is required to illuminate the dynamics of this group and to determine its fit with existent concepts that aim to capture the authority and influence of scientific networks, such as ‘epistemic community’, ‘core set’ or ‘invisible college’^{25, 26, 27}. This inference is reinforced by the understanding that these authors also come from a narrow range of disciplines—all either economists or engineers.

Lines between nodes indicate that they have co-authored at least two papers—single collaborations are excluded. Node size indicates the betweenness score. Shape indicates gender (triangle, men, circle, women). Colour indicates the world’s region, attributed according to their IPCC representing country (green, Europe; orange, North America; blue, BRICS countries; other regions in white). Some nodes have been moved slightly from their mathematically determined location to enable visibility of the label. Data used to construct this figure can be found in the ‘CV Data’ and the ‘Co-Authoring Patterns Analysis’ tabs of the Supplementary Database.

Turning to gender and disciplinary make-up, women represent 18% (50/273) of AR5 WGIII authors, and only two women appear in the ‘top 20’ co-authoring network. In line with an analysis of the Third Assessment Report¹⁶ (TAR), trained economists are dominant in the report, followed by engineers, and then by scholars trained in energy studies, mathematics and physics. Trained economists and engineers represent 49% of the authors for whom we were able to get data for their highest academic training (125/253), whereas 15% (39/253) were trained as social scientists. This pattern is more marked among the 35 Coordinating Lead Authors (CLAs), among which 14% are female and 58% are trained as economists or engineers. Only 3 CLAs have been trained in social sciences other than economics (that is, political science, geography and law), and none in the humanities. This is significant because CLAs, along with WG bureau and TSU members, make up the drafting team for the Summary for Policymakers (SPM), which constitutes the document through which the main findings of the assessment are distilled for wider social and political attention.

We also analysed authors’ academic ‘disciplines’, taking into account their highest educational qualification and contribution to one or various research fields (see Methods). This enabled greater sensitivity to deviations in authors’ career pathways and avoided equating the highest academic qualification to the author’s current research orientation. The results amplify the dominance of economics and engineering-based disciplinary backgrounds, representing 56% of the authors’ sample. Sixteen per cent of all contributing authors have well-defined disciplinary backgrounds on physical, natural and applied sciences, and 4% on environmental sciences or related degrees with multiple climate mitigation research foci. Social science scholars represent 22% of this sample, and only 2% are trained in the humanities (see Supplementary Fig. 1).

Figure 4 shows the two-degree co-authoring network with our academic disciplines codes overlaid to see if and how authors cluster along disciplinary lines. It also shows the number of co-authorships between nodes and thus the strength of the collaborative relationship. It can be observed that the core of the co-authoring network is composed of multidisciplinary teams dominated by engineers (blue) and economists (orange), with one physical scientist (Smith, green) well connected to this network. This scientist, in a one-degree network (see Supplementary Figs 2 and 3), seems strongly linked to a larger cluster of other physical scientists, most of whom worked on one specific chapter—chapter 11 on Agriculture, Forestry and Land Use or AFOLU—suggesting strong coordination between a group of closely connected researchers who have been able to join the IPCC to work on these issues. In disciplinary terms, Fig. 4 also suggests that engineers are slightly dominant over economists and, when explored in relation to WGIII AR5 chapters, chapters 5, 6, 7 predominate in the co-authoring network. These are the large modelling chapters (drivers of emissions, transformation pathways, energy systems) that are at the ‘heart’ of the mitigation report. To the extent there is an epistemic community as suggested above, Fig. 4 depicts the episteme informing the knowledge production of this community.

Lines indicate that the two authors co-authored at least two articles together. Node size indicates the betweenness score. Shape indicates gender (triangle, men, circle, women). Colour indicates the author’s academic discipline (blue, engineering; green, physical, natural and applied sciences; orange, economics; yellow, social sciences; other disciplines in white). Data used to construct this figure can be found in the ‘DegreeDisciplines Analysis’ and the ‘Co-Authoring Patterns Analysis’ tabs of the Supplementary Database.

There are three key messages that can be distilled from our analysis. First, despite historical improvements in the relative participation of scientists from the global South, those who have participated in WGIII are mostly trained in northern institutions, overwhelmingly in the USA and UK. In addition, some countries in the global South, notably India and Brazil, are more connected to the core of the IPCC research network than others. Second, there are clear patterns in the authors’ network showing the importance of specific international organizations in shaping the field of climate mitigation policy research as represented in the IPCC, and suggestive of the existence of a core network of researchers whose careers are centred on the IPCC and the research represented within it. Third, scholars from the humanities remain marginalized from the IPCC’s assessment of climate mitigation in comparison with economists, engineers and applied scientists. These issues can be further investigated and tested exploring the IPCC as a social network across all three working groups and across the five assessment reports since 1990, as well as through interviews and ethnographic research²⁸.

The shared institutional pathway identified may indicate why WGIII’s contribution to the AR5 met with reluctant acceptance in the governments’ approval session²⁹. Although geographic representation has increased, our analysis makes apparent that actors and institutions in the North continue to play a dominant role in constructing the IPCC’s assessment of mitigation and thus their influence on the UNFCCC process. These findings combined might also suggest why the WGIII AR5 presents a fairly strong harmonization of views, compared with the diversity one finds across the social sciences of climate change more broadly, and explain why WGIII finds it difficult to effectively incorporate many important questions into its discourse, notably questions of justice or governance, because the disciplines dominating the WGIII author team do not have these questions at their core and frame important questions narrowly. For example, referring to ‘behavioural change’ instead of ‘consumption practices’ ignores important insights from disciplines (in this case sociology) that remain largely excluded from the process³⁰.

The conclusion of our analysis, however, is not necessarily that incorporating these broader perspectives would produce a ‘better harmonization’ of views within the IPCC mitigation reports because the Panel’s capacity to produce a ‘consensus’ might be dependent precisely on these exclusions. Rather, broadening the geographical and disciplinary basis of participation in the IPCC may help unearth the key conflicts and choices to be made in climate change mitigation policy, between different values and interests. This might in turn enable the IPCC to increase its policy usefulness by emphasizing the important political choices societies confront as they respond to climate change, particularly mitigation. Changes in the author team composition and the effects on the report’s contents of a broader disciplinary engagement would thus probably not make the approval process of the SPM any easier in the future. However, the mitigation report and its ensuing SPM might well become more legitimate and encompassing to more countries and scientific audiences²¹.

We focused specifically on the Working Group (WG) III writing team for the IPCC Fifth Assessment Report (AR5), dedicated to overview the most recent research related to climate change mitigation. Two of us were Lead Authors in the full report, which was officially published in September 2014 (the Summary for Policy Makers and the full report chapters are available at http://www.mitigation2014.org). WGIII vice-chairs and co-chairs and a scientific supporting team unit, who are part of what is known as the Technical Support Unit (TSU), edit the final report. Each of its chapters is produced by two Coordinating Lead Authors (CLAs) who steer and contribute to the writing-up process, a