On a Dynamic Public Insurance Game with Heterogeneous Agents
Dr Timothy Kam (ANU)
MSI Computational Mathematics Seminar SeriesDATE: 2009-10-12
TIME: 11:00:00 - 12:00:00
LOCATION: JD G35
CONTACT: JavaScript must be enabled to display this email address.
ABSTRACT:
We develop a class of dynamic taxation and public insurance games between a government, whose aim is to maximize a measure of discounted expected total social welfare, and a continuum of private agents. Private agents are endogenously distributed on a finite set of individual states. We show that the sequential equilibrium payoff correspondence exists and can be found recursively. If the government can commit, there exists an optimal fixed tax-and-insure policy for any initial distribution of private agent states. We also prove that a socially optimal steady state exists in such a case. In the latter problem with commitment, the solution can be found using standard numerical dynamic programming techniques. In the former problem, sequential equilibria have to be found by successive approximations of convex correspondences (with compact graph) in a high-dimensional Euclidean space. The sequential- equilibrium operator defining these successive approximations involve solving constrained op- timization problems at all possible game states. We are currently modifying approximation techniques that may allow for this problem to be computable.
BIO:
http://cbe.anu.edu.au/staff/info.asp?Surname=Kam&Firstname=Timothy
