Advances in Complex Systems

We report a study of the correlations among topological, weighted and spatial properties of large infrastructure networks. We review the empirical results obtained for the air-transportation infrastructure that motivates a network modeling approach which integrates the various attributes of this network. In particular, we describe a class of models which include a weight-topology coupling and the introduction of geographical attributes during the network evolution. The inclusion of spatial features is able to capture the appearance of non-trivial correlations between the traffic flows, the connectivity pattern and the actual distances of vertices. The anomalous fluctuations in the betweenness-degree correlation function observed in empirical studies are also recovered in the model. The presented results suggest that the interplay between topology, weights and geographical constraints is a key ingredient in order to understand the structure and evolution of many real-world networks.

We present an analysis of inter-regional investment stocks within Europe from a complex networks perspective. We consider two different levels: first, we compute the inward–outward investment stocks at the level of firms, based on ownership shares and number of employees; then we estimate the inward–outward investment stock at the level of regions in Europe, by aggregating the ownership network of firms, based on their headquarter location. To our knowledge, there is no similar approach in the literature so far, and we believe that it may lead to important applications for policy making. In the present paper, we focus on the statistical distributions and the scaling laws, while in further studies we will analyze the structure of the network and its relation to geographical space. We find that while outward investment and activity of firms are power law distributed with a similar exponent, for regions these quantities are better described by a log-normal distribution. At both levels we also find scaling laws relating investment to activity and connectivity. In particular, we find that investment stock scales as a power law of the connectivity, as previously found for stock market data.

While networks are widely thought to enhance regional innovative capability, there exist few longitudinal studies of their formation and evolution over time. Based on an analysis of all patenting inventors in the U.S. from 1975 to 2002, we observe dramatic aggregation of the regional inventor network in Silicon Valley around 1989. Based on network statistics, we argue that the sudden rise of giant networks in Silicon Valley can be understood as a phase transition during which small isolated networks form one giant component. By contrast, such a transition in Boston occurred much later and much less dramatically. We do not find convincing evidence that this marked difference between the two regions is due to regional differences in the propensity to collaborate or the involvement of universities in patenting. Interviews with key network players suggest that contingent labor mobility between established firms in Silicon Valley, in particular resulting from IBM's policy as a central player in patenting activity, promoted inter-organizational networking, leading to larger inventor networks.

This paper develops a simulation model of the behavior of industrial districts in the face of changes in their environment. By applying Kauffman's N/K-C model to the question of optimum governance in industrial districts facing an external shock, the paper sheds light on the optimum co-ordination mechanism among agents in interdependent industrial networks. Simulation results indicate that collective governance structures with agents adjusting for the sake of the district as a whole perform best in adaptation, whereas individualistic governance modes deliver the worst results. Alliance and leader firm governance forms position themselves in-between these two extremes. However, both modes of governance can be preferable to the collective one if the observation and punishment mechanisms in the district's local culture are not strong enough to impose solidarity among its agents. In this case, a prisoner's dilemma emerges and the collective governance form is replaced by the inferior individualistic one. Through these results, the model highlights the link between governance and district adaptability. It provides an explanation for the trend in Italian districts towards more hierarchical governance structures. Moreover, the identification of the role played by governance structure for district adaptability in changing environments could serve as guidance for future research.

Technological change is argued to be taking place along ordered and selective patterns, shaped jointly by technological and scientific principles, and economic and other societal factors. Historical, descriptive analysis is often used to analyze these "trajectories". Recently, quantitative methods have been proposed to map these trajectories. It is argued that such methods have, so far, not been able to illuminate the engineering side of technological trajectories. In order to fill this gap, a methodology proposed by Hummon and Doreian (1989) is used and extended to undertake a citation analysis of patents in the field of fuel cells.