Interconnection and overlap of networks

Researchers

Helen Simpson

Research Grant

Outputs

Project outline

Existing network literature focuses on the effects of membership of ‘single’ networks. By this we do not mean that there is only one network but that membership (e.g., joining a telephone network, neighbourhood, etc.) is treated as an activity in its own right with a social or economic end project derived from membership. However, individuals belong to several networks at one time and there are interesting interrelationships between these networks. Indeed this is explicit in the way that networks are used as signals. This is a common characteristic of networks that interconnect and overlap with neighbourhoods. For example, it is generally believed that neighbourhood (through postcode) conveys a great deal of information about the person.  This may be because neighbourhood operates as an all encompassing network but this is more likely to be because the equilibrium outcomes of social choices over interconnecting networks are far from independent, i.e. knowledge of membership of one network (neighbourhood for example) is informative about membership of other networks that are the real focus of attention. At school, for example, teachers and parents may use neighbourhood as an indication of whether new parents will be likely to join the parent teachers association and hence neighbourhood may affect how the headteacher and current members allocate effort to sustain the membership of the association. If such beliefs are common then they may become self-reinforcing through self-selection and small complementarities may then have large effects. If several networks are used as signals about other networks then these effects may become very strong and self-sustaining. 

The aim of this section of the project is to model and understand interconnection and competition between networks when the networks are far from perfect substitutes.  In particular, we seek to understand why in equilibrium we appear to observe far more overlapping of networks than one would expect to arise purely from differences in preferences and income. This will involve both theoretical analysis and case studies.