Empirical Models of the Exchange Rate: Separating the Wheat from the Chaff

摘要: Various popular exchange rate models (a standard monetary model, a portfolio balance and sticky-price models) are estimated evaluated using U.S.-Canadian data for the 1970s. Nonnested hypothesis tests demonstrate that none correctly specified. The suggest: 1) persistence observed is not fully explained by any of models; 2) small Durbin-Watson statistics indicate longer lags required; 3) current account useful explanatory variable; 4) model fits well, but has potentially serious problems measuring stock foreign assets.(This abstract was borrowed from another version this item.)