Signaling versus Commitment Strengthening: Exchange Rate Insurance against Currency Attacks

Both empirical and theoretical studies suggest that currency attacks can occur even in a fixed exchange rate regime with sound fundamentals. Can mechanisms be designed to prevent such currency attacks? To address this question, we first need a theory of currency crisis. I argue that such a theory must contain two ingredients: the government's lack of commitment and its preferences being private information; and that a successful mechanism must handle both of the problems. With such a theory in mind, I evaluate the proposal that Chan and Chen (1999) and Merton Miller (1998) made during the Asian economic crisis to defend the Hong Kong dollar and Chinese RMB via the government sale of insurance against devaluation. I argue that the proposal will not perform as well as claimed. As the issuance of insurance makes devaluation more costly, the commitment to peg is strengthened. That the insurance scheme serves as a commitment device renders it an ineffective signaling device: in the game where the government's type is private information, a separating equilibrium does not, in general, exist where only the strong type adopts the insurance scheme. Despite this, I also find that it is never a negative signal: that is, it will never be the case that the weak type adopts the proposal while the strong type does not. Therefore, the potential problem of the Miller proposal can be fixed by giving the government one more dimension of choice. One remedy, for example, is to allow the government to also choose whether to sell the insurance or to give it out for free. The dimensionality critique that is identified above is valid for other unidimensional proposals. The recent path breaking work of Morris and Shin (1998) is employed to tackle the coordination problem among speculators.