Complexity Science Seminar Series - Abstracts
Title: Heterogeneity in Nash Networks
Presenter:
Sudipta Sarangi, Dept. of Economics, Louisiana State University
Date and Time: Tuesday, October 30, 2007, 3:30-4:30PM
Abstract:
Non-cooperative models of network formation involve decisions by agents regarding link formation. Agents in this simultaneous move game are endowed with information that can only be accessed by an act of costly link formation. Stable networks in this context must have the Nash equilibrium property, i.e., be stable against unilateral deviation by any player. In this context efficiency is defined as the network that maximizes the sum of all payoffs. Heterogeneity in Nash networks with two-way flow (undirected networks) can arise due to differences in the following four variables: (i) the value of information held by agents, (ii) the rate at which information decays as it traverses the network, (iii) the probability with which a link transmits information, and (iv) the cost of forming a link. In this talk we show that with heterogeneity all networks can be supported as Nash or efficient. This sharply contrasts with the homogeneous cases. We then identify conditions under which efficient networks are also Nash. We also find that cost heterogeneity is crucial for non-existence of Nash networks in pure strategies. We show that there is in general no relationship between the decay and probabilistic models of network formation. Finally, we show that on reducing heterogeneity the earlier "anything goes" result disappears.Seminar Location: The seminars are held at:
Virginia Tech, Corporate Research Center
1880 Pratt Drive, Building XV
NDSSL Conference Room, first Floor
Directions: Map (PDF)
Back to: NDSSL Seminar Page