Abstract
A standard rational expectations model would give strong predictions about the behavior of the nominal exchange rate at the beginning of a disinflation (a rise in interest rates): a substantial initial appreciation, followed by a steady depreciation. It largely conflicts actual observations, like the recent experience of Poland, Hungary, and Chile, where an initial appreciation was not followed by any systematic depreciation. The paper tries to explore whether rational expectations can be rescued by introducing noise and parameter learning. An optimistic learning case (worse than expected inflation data every period), or the combination of a pessimistic learning case (better than expected data every period) and a declining proportional risk content of the interest rate offers a potential explanation.
WP2003_1