Signaling game:
Here we analyze the most widely applied class of games of incomplete information: signaling games. Stated abstractly, a signaling game involves two players (one with private information, the other without) and two moves (first a signal sent by the informed player, then a response taken by the uninformed player). The key idea is that communication can ocqur if one type of the informed player is willing to send a signal that would be'too expensive for another type to send. We first define perfect Bayesian Nash equilibrium of signaling game and then give an example for it.
Definition of Signaling Game
A signaling game is a dynamic game of incomplete information involving two players: a sender (S) and a receiver (R). The timing of the game is as follows:
1) Nature draws a type ti for the sender from a set of feasible types T = {t1, ..., ti) according to a probability distribution p(t1), where p(ti) > 0 for every i and p(t1) + ... + p(tr) = 1 .
2) The sender observes ti and then chooses a message mj from a set of feasible messages M = {m1, ..., m2) .
3) The receiver observes mj (but not t,) and then chooses an action ak from a set of feasible actions A = {a1, .... ak) .
4) Payoffs are given by Us(ti, mj, ak) and UR(t1, mj, ak).
The figure below gives an extensive form representation (without payoffs) of a simple case: T = (t1, t2), M = (m1, m2), A = (a1, a2), and Prob(t1) = p. Note that the play of the game does not flow from an initial node at the top of the tree to terminal nodes at the bottom, but rather from an initial move by nature in the middle of the tree to terminal nodes at the left and right edges.
Before we go to give the formal definition of perfect Bayesian Nash equilibrium of signaling game, we must give four requirements needed to define perfect Bayesian Nash equilibrium of signaling game. Because the sender knows the full history of the game when choosing a message, this choice occurs at a singleton information set. Thus, requirement 1 is trivial when applied to the sender. The receiver, in contrast, chooses an action after observing the sender's message but without knowing the sender's type, so the receiver's choice occurs at a nonsingleton information set.