Network Analysis and Institutional Change in Fragmented Global Governance Architectures

Network analysis holds a significant potential for the study of complex, multi-institutional, global governance architectures. However, it has been underutilized so far. While the identification of institutions of interest is straightforward, edge detection is a major challenge: as network size increases, the viability of manually coding institutional interactions, linkages or overlaps decreases. Making network analysis into a feasible methodology requires automated procedures for determining whether or not edges exist between pairs of institutions. How can we systematically and efficiently identify various types of inter-institutional relationships for larger sets of institutions? We propose and contrast two approaches: inter-treaty citations (e.g., Treaty A cites Treaty B) and institutional overlaps (Treaty A and Treaty B regulate the same subject matter and share common states parties). Both allow for the efficient generation of network data while capturing different aspects of institutional complexity: inter-institutional interplay on one hand, and on the other legal rights and obligations from parallel sources pertaining to the same actors for the same subject matter. We apply the two edge-detection approaches to map, analyze, and compare two evolving networks of 180 international pollution agreements adopted since 1945. In this broader issue area, so far largely neglected in the literature on institutional fragmentation, we find a diverse set of institutions of global and regional scope that deal with subjects ranging from hazardous waste over marine pollution from land-based sources and oil spills up to transboundary air pollution. Focusing on changes in community structures and graph-level properties, we attempt to identify similarities and differences in the patterns of institutional change between the two resulting networks. We discuss what citations and institutional overlap can tell us in a social network analytical setting. We also attempt to link network metrics (such as density, centralization and communities) to theoretical concepts in the field of institutional fragmentation and change. In both the citation network and the overlap network, we see broader patterns of change through punctuated equilibria; the emergence of cores of institutions characterized by relatively high centrality; as well as the formation of sub-clusters of institutions with regional and functional specialization. We conclude that, in addition to addressing the methodological challenge of edge detection, network analysis of fragmented global governance architectures requires the mapping of network-theoretic concepts to key institutionalist concepts.