Finite State Dynamic Games with Asymmetric Information: A Computational Framework

We present a simple algorithm for computing an intuitive notion of MPE for ?nite state dynamic games with asymmetric information. The algorithm does not require; storage and updating of posterior distributions, explicit integration over possible future states to deter- mine continuation values, or storage and updating of information at all possible points in the state space. It is also easy to program. To il- lustrate we compute the MPE of a collusive industry in which ?rms do not know each other?s cost positions. Costs evolve with the (privately observed) outcomes of their investment decisions. Costly meetings are called when a ?rm perceives that its relative cost position has improved. The meetings reveal information and realign pro?ts accod- ingly. We show that parameters determining information ?ows can e¤ect market structure and through market structure, producer and consumer surplus.