Reviewed by
Camille Roth
CAMS Centre d'Analyse et de Mathématique Sociales, CNRS/EHESS
With Connections, Sanjeev Goyal successfully offers a pedagogical introduction to key concepts of network-based approaches in economics. The book constitutes a detailed and step-by-step overview of the possibilities and advantages of introducing networks in a large range of economic research topics - spanning from bilateral trade agreements to job markets, through research collaboration between firms. More to the point, Connections provides a thorough demonstration of how concrete economic situations and corresponding agent behaviours and market mechanisms can be appraised under the lens of networks - or, rather, how useful such a formalism can be in contexts where horizontal (agent-to-agent) cooperation, interaction and exchange crucially impact economic phenomena. Network topology and topology-dependent actions are generally connected to various outcomes in terms of payoff maximization, social welfare, reliability, system-level equilibria and local vs. global efficiency. Readers interested in policymaking issues would also probably appreciate frequent discussions regarding the consequences of modifications on network shape and/or interaction behaviours.While the earlier chapters focus on how macro and micro behaviours are influenced by topology, the book progressively intertwines topology and processes, so as to focus eventually on how topology itself is influenced by or even emerges from locally- or globally determined actions. For instance, specific games played on various simple network configurations are first shown to lead to significantly diverse outcomes in terms of both individual payoffs and global welfare (Chap. 3). This argument is further developed in an evolutionary setting when topology is shown to influence cooperation and coordination dynamics (Chap. 4) or information diffusion phenomena (Chap. 5). The immutability of network configuration is then questioned when Chapter 6 begins to explore correlations between agent features, positions and actions in the context of job seek and recommendation-based employer/employee matching. Later, topology becomes properly a dependent variable when networks stem from strategic decisions, notably in contexts where agents create or delete links to maximize utility (Chap. 7), or choose partners in a unilateral (Chap. 8) or bilateral (Chap. 9) fashion.
On the whole, the book appears to be organized in such a way that both network topology and processes are explored by increasingly emphasizing their interrelated and co-evolutionary nature.
Connections is therefore certainly likely to become a reference book for economists curious of network-based approaches - especially as Goyal generally respects a convenient balance between formal models & equations, and clear, qualitative and often intuitive explanations of the economic mechanisms they represent.
The book admittedly targets economists, focusing on their issues, using their vocabulary and borrowing their notations; it also undoubtedly provides a practical translation into economic terms of several concepts and recent advances in the broader interdisciplinary area of network research.
Could Connections work the other way around, by pleasing other audiences as well and bringing them economic intuitions? There are clearly a few obstacles for that matter. First, the reader used to the traditional (social) network analysis literature would certainly need some time to adapt to economic notations and payoff-based thinking. Similarly, a sizeable portion of the book earnestly deals with economic technicalities which could appear as an in-depth inspection of issues of minor importance to most social modellers - strategic network formation, for instance, occupies a small half of the book.
Most important, the reader interested in empirical applications or results might be deceived by the fact that analytical solutions and therefore analytically solvable models are typically preferred over results based on simulation and/or empirical datasets. From this perspective the book nonetheless provides a robust and consistent repository of formal stylized facts from which to get inspiration to conceive simulation-based models when coping with more realistic situations. Additionally and more broadly, by apprehending network dynamics as the optimal and stable result of some quantity maximization and by encompassing nodes as strategic actors where link formation is induced by trade-offs and clear-cut rational preferences, Connections could plausibly provide a fresh outlook on networks to non-economists as well.
Return to Contents
of this issue
© Copyright Journal of Artificial Societies and Social Simulation, 2010