Learning and complementarities in speculative attacks

We study a model where the aggregate trading of currency speculators reveals new information to the central bank and affects its policy decision. We show that the learning process gives rise to coordination motives among speculators leading to large currency attacks and introducing non-fundamental volatility into exchange rates and policy decisions. We show that the central bank can improve the ex ante effectiveness of its policy by committing to put a lower weight ex post on the information from the market, and that transparency may either increase or decrease the effectiveness of learning from the market, depending on how it is implemented.