Learning and Complementarities: Implications for Speculative Attacks∗

We study a model where the aggregate trading of currency speculators reveals new information to the central bank and affects its decision on whether to abandon a fixed exchange rate regime. We show that the learning process gives rise to informational complementarities among traders, which generates coordination motives in their trading and reduces the informational content of the market signal and the efficiency of the central bank’s decision. We analyze the tradeoff between information and coordination and derive comparative statics regarding the efficiency of the central bank’s decision. We show that by committing to overweight its own beliefs, the central bank can improve the ex ante efficiency of its decisions. We also find that transparency may reduce the ability of the central bank to learn from the aggregate action of the speculators. Overall, our results apply to a wide range of problems, where decision makers learn from the actions of heterogeneously informed agents. Journal of Economic Literature Classification Codes: G12, G15, F39, C7.