The Extrapolative Component in Exchange Rate Expectations and the Not-So-Puzzling Interest Parity: The Case of Uruguay

This paper analyses the importance attached to the past behaviour of the exchange rate when forming expectations and tests for the uncovered interest parity hypothesis. Using interest rate dierentials for Uruguay over 1980-2010, we identify a strong and time-varying extrapolative component in exchange rate expectations. Agents attach more importance to the past behaviour of exchange rates the higher the level of in ation is. Yet agents are able to internalise policy announcements and external events that are likely to aect exchange rate fun- damentals. Further, we nd deviations from the uncovered interest parity hy- pothesis. These are lower than those usually reported for developed economies. Also, they tend to be higher for the period of low in ation and freely oating exchange rates. As long as what it takes to predict well is rather simple | i.e. look backwards, follow policy announcements, the interest rate dierential per- forms well. Once the exchange rate determination model becomes more intricate or less familiar to the agents, they tend to fail at predicting exchange rate depre- ciations. These results point to expectational failures as a likely explanation for the `uncovered interest parity puzzle'.