| During financial bubbles, many agents reenter the inflated market after having exited on time. Using dynamic game theory I show how this strategy can be rationalized.;The thesis extends Abreu and Brunnermeier (2003) to allow for (1) heterogeneity in agents' prior beliefs about the fundamental value of a stock and (2) uncertainty about other agents' heterogenous priors.;The first part shows that, with heterogenous priors and common knowledge, the higher the number of optimistic agents in the market is, the longer all the agents tend to stay invested.;The second part shows that, with uncertainty about the priors, for some values of the parameters, there exists a unique equilibrium with reentry. Otherwise, the equilibrium does not exhibit reentry.;The key innovative feature of the model is that investors use prices to learn each other's beliefs.;The result is important in three ways. First, heterogeneity in beliefs influences the length of a financial bubble. Second, it sheds light on the optimal strategy of a central bank that wants to limit the size of a bubble using the risk free interest rate. The most surprising finding is that the central bank does not always prefer to "lean against the wind".;Finally, when there is uncertainty about the priors, price manipulation can act through a channel not previously considered. A manipulator, misleading agents' learning about others' beliefs, could make a mispricing last longer. |
