Interdependent Choices: Studying Alliances Using Stochastic Actor-Oriented Models of Network Co-Evolution

The formation, maintenance, and dissolution of alliances is the result of strategic action among states. Recent works have highlighted that the formation of an alliance between two states does not occur in isolation from the alliance behavior around these states (Warren, 2010; Maoz, 2011; Cranmer et al., 2012). In this paper, I build on this literature and argue that states’ alliance behavior is not only affected by characteristics of individual states and dyads but also by the behavior of other dyads in the global alliance network. I also maintain that patterns of cooperation in other international networks, such as those constituted by states’ shared memberships in intergovernmental organizations (IGOs) and preferential trade agreements (PTAs), exercise powerful influence on how alliances are formed. In other words, I argue that alliances form as interdependent ties among states and that alliances co-evolve with other forms of international cooperation. In order to estimate these within and cross-network interdependence effects, I use a stochastic actor-oriented model of network co-evolution (Snijders 1996, 2001). This allows me to avoid the assumption of independence of observations which underlies the vast majority of estimation techniques employed by international relations scholars and to directly model interdependence effects of triadic and higher order within the alliance network as well as the co-evolution of the alliance network and other networks of international cooperation. I test my argument using a dataset that combines information about alliances, IGOs, and PTAs between 1955 and 2005. Results suggest that both within and cross-network extra-dyadic dependencies exercise powerful influence on states’ alliance choices indicating that strategic action does neither stop at the dyad nor at the network level of analysis.