This paper is concerned with issues of model specification, identification, and estimation in exchange rate models with unobservable fundamentals.We show that the model estimated by Gardeazabal, Reg´ulez and V´azquez (International Economic Review, 1997) is not identified and demonstrate how to specify an identified model in-keeping with their intended approach.Estimates of the identified model are reported for five currencies over two time spans, and a restriction suggested by the asset market view of exchange rate determination is not rejected for any currency or time span.The forecasting performance of the model is also examined and is found to compare favourably with forecasts generated by a random walk with drift.
|Place of Publication||Tilburg|
|Number of pages||25|
|Publication status||Published - 2004|
|Name||CentER Discussion Paper|
- exchange rate