I extend the traditional Diamond Dybvig framework with aggregate liquidity shocks to small open economies. Currency board may imply perfect risk sharing (with perfect credit markets), contrary to Chang and Velasco's findings (2000). With interim-date borrowing constraints and fixed exchange rates, Wallace's (1990) partial suspension of convertibility of deposits is obtained. A banking system with an international lender may implement both allocations without runs. Flexible exchange rates with local-currency denominated deposits improves risk sharing relative to fixed exchange rates when borrowing constraints are present.