Defending Against Speculative Attacks: Reputation, Learning, and Coordination

How does the central bank's incentive to build a reputation affect speculators' ability to coordinate and the likelihood of the devaluation outcome during speculative currency crises? What role does market information play in speculators' coordination and the central bank's reputation building? I address these questions in a dynamic regime change game that highlights the interaction between the central bank's reputation building and speculators' individual learning. On the one hand, the central bank has private information about its value from the currency peg and decides whether to maintain it. By defending against speculative attacks, it can build a reputation of defending, which may deter future attacks. On the other hand, speculators individually learn the central bank's value, and such learning may encourage speculators to coordinate an attack. I show that though learning makes the central bank's value approximate common knowledge over time, there is a unique equilibrium when learning is slow. In this equilibrium, no speculator attacks and the central bank sustains the currency peg forever, because the central bank obtains commitment power through the incentive to build a reputation. When learning is fast, there may be equilibria with attacks. In any equilibrium with attacks, the onset of the attack depends on the entire learning process. Once speculators attack, they attack frequently and infinitely often. Consequently, the central bank has no incentive to build a reputation and abandons the currency peg almost surely.