Printer-friendly version

The research activity of the Coalition Theory Network is centred on the theoretical and empirical analysis of socio-economic networks and groups. This pertains to different aspects of social and economic systems, among which the management of international public goods, the governance of economic unions, the formation of industrial cartels and collaborations, the patterns of racial integration in social networks, and the endogenous evolution and structure of institutions, etc.

There is now a broad consensus within the international scientific community that such crucial phenomena call for a unified theoretical approach, based on the development of behavioural models and equilibrium concepts, and for an extensive investment in empirical research. The last two decades have witnessed a huge academic commitment in this direction, especially within the disciplines of economics, game theory and behavioural sciences in general. 

Moreover, a growing interest in social and economic networks has been fostered by recent developments in the analysis of strategic network formation, and by an increasing intense integration between this new body of research and earlier work on the formation of large networks, developed in the mechanical physics literature.


A review of the state of the art

The theory of coalition formation focuses on the analysis of the incentives of economic agents to communicate and on how these incentives act on the determination of equilibrium configurations of social and economic groups. It has developed as the natural extension of traditional cooperative game theory, which focused mainly on the problem of how the surplus of cooperation "should" or "will" be shared among cooperating agents, while the theory of coalition formation has extended the set of questions in two principal ways. First, it investigates how groups form, and what configuration of groups will emerge in equilibrium. The formation of multiple groups is consistent with the recent experience of international negotiations on environmental standards and on trade issue, where "small" coalitions are the rule rather than the exception, even when full cooperation would be desirable. A second important extension of the set of questions addressed by coalition theory is the study of problems in which the welfare of a group can be affected by the actions taken by agents who are members of other groups, that is, when there are externalities among groups. Such externalities are typical, for example, of cartel formation in oligopolies, cooperation on environmental policies, and economic unions.

In order to account for these empirical regularities, theorists have taken new approaches to the problem, which can be classified as follows.

Formulation of new game-theory models of coalition formation and networks. In these models, the choice of forming a coalition is modelled as a strategic variable of each player emerges as the outcome of purely non-cooperative behaviour. We can further distinguish between static models (e.g., Hart and Kurz (1983)) and dynamic models of coalition formation (e.g., Chatterjee, Dutta, Ray and Sengupta (1993), Bloch (1996)).

Formulation of new solution concepts. Here, the effort is aimed at identifying equilibrium coalition structures that satisfy consistency requirements both within and outside forming coalitions (e.g., Ray and Vohra (1997)). Other new solution concepts address situations where agents are far-sighted (eg. Dutta, Ghosal and Ray (2005), Page, Wooders and Kamat (2004)). There is also a need for dynamic models and solution concepts that adequately capture the most salient features of real-world situations.

The interest of economics and other behavioural sciences in network theory is more recent, and has mainly developed since the seminal works of Myerson (1991) and of Jackson and Wolinsky (1996). This interest is motivated by the number of economic phenomena that are affected, in one way or another, by the communication network in which agents are embedded. Job contact networks and trade networks are only two of the many examples that may be cited. What is important here is that both the aggregate welfare and its distribution seem to depend on how information flows on the network, thus providing a clearer understanding of how networks form a key issue in economics.

Models of network formation have been previously developed within mechanical physics literature, in addressing the issue of how large networks form as the outcome of purely mechanical random processes. More recent economics literature focuses instead on the incentives of rational agents to establish links and form networks as the outcome of optimal behaviour, and on the welfare implication of such behaviour. There is now a consensus within the scientific community that a bridge between these two literatures needs to be built, in order to study incentives and welfare within complex architectures of the type encountered in social networks, where randomness clearly plays a role. This tendency is confirmed by the large number of new contributions on strategic models of network formation that borrow tools and concepts from mechanical physics literature.

Recent research on networks has had a strong empirical component, particularly in topics related to labour economics and consumer behaviour. Peer and neighbourhood effects on youth have been studied by Sacerdote (2001) and Katz et al. (2001), and social interactions and crime are the focus of Glaeser (1996) and Kling et al. (2005). The possibility of social interactions in labour supply is discussed by Grodner and Kniesner (2006), Weinberg et al. (2004), and Woittiez and Kapteyn (1998), and the impact of custom on contract design is analyzed by Young and Burke (2001). The econometric problems in identifying network effects are complex, and discussed in Brock and Durlauf (2001), Manski (1993), and Moffitt (2001). Glaeser and Scheinkman (2002) provide an overview of some of this literature, which is clearly in its infancy but ripe for future research. An area of particular promise is the study of the evolution and consequences of new institutional arrangements in the transition economies.


Advancing the state of the art

Theoretical and empirical research on networks and groups is moving in several directions. Some of these are natural developments of the existing state of the art, such as the formulation of more general models and more satisfactory solution concepts that would obtain cooperative outcomes as the result of individual equilibrium actions and decisions. Other lines of research respond to the emergence of new and unprecedented social, economic and political frameworks at both national and international levels, which raise new theoretical and policy issues. From the theoretical standpoint, networks and network formation pose the most new and challenging issues for researchers. Social agents are in many ways organised in a network of relationships, taking the forms of friendship ties, professional relations, strategic interaction, physical connections, etc… Network theory needs to be able to provide a unified framework for analysing the relation between agents’ position in the network and their actions and welfare. Even more generally, a model is needed to explain how the whole structure of the network (or the beliefs that agents hold in this structure) affect agents’ behaviour and welfare. The study of network formation and of games played on networks under local and limited information is indeed one of the most challenging and frequently studied issues at present. 

Another important theoretical problem is how to model the formation of large networks and, in particular, how to integrate random network formation models with strategic models in which agents form links rationally. A bridge between these two strands of modelling is needed,in order to make predictions on actual social networks, and to say something about incentives within such structures. 

The lack of data on social networks also suggests that experimental work in this area will play an important role in empirical research. Experimental design in this specific topic is therefore another main direction in which research will make a great effort. This is true also for the problem of coalition formation, for which a larger and more established body of theoretical work exists.

Many interesting questions arise from more applied problems, in which the general models of coalition and network formation are able to provide more specific predictions. For instance, a recent successful approach to group formation is the design of procedures to match agents together in real-life problems such as academic job markets, admission of students in schools or universities (e.g., Roth (1984, 2004)). Recent applications of matching theory raise new issues such as the problem of strategic behaviour and asymmetric information in large markets. Also, the design of efficient and stable voting rules can be fruitfully applied to EU decision-making problems. More in general, the design of stable and efficient self-enforcing institutions seems to be among the main objectives of coalition theory, especially in an evolving world context in which political equilibria are still to be found. In this respect, a fruitful area of research is the investigation of how domestic political incentives and institutions affect the behaviour of countries at the international level, and thereby the outcome of treaties and cooperation. Here the theory of coalition formation can be fruitfully integrated with the theory of networks, the latter studying the formation of bilateral international relations, while the former studying the stability of economic and political unions.

Developed by Paolo Gittoi