The Out-of-Sample Failure of Empirical Exchange Rate Models: Sampling Error or Misspecification?

Abstract

A companion study (Meese and Rogoff 1983) compared the out-ofsample fit of various structural and time-series exchange rate models and finds that the random walk model 1 performs as well as any estimated model at one-to twelve-month horizons for 1970s dollar/mark, dollar/pound, dollar /yen, and trade-weighted dollar exchange rates. 2 The structural models perform poorly even though their forecasts are purged of all uncertainty concerning the future paths of their explanatory variables by using actual realized values. The present study demonstrates that the dismal short-to medium-run forecasting performance of the structural models is not attributable to the sample distribution of the coefficient estimates. We rule out that explana

Perspectives on the recent currency crisis literature

“…35. The classic reference is Messe and Rogoff (1983). Of course, exchange-rate models are not the only disappointing ones.…”

The Role of National Debts in the Determination of the Yen‐dollar Exchange Rate

“…Given our relatively successful empirical estimates of the model coefficients, we also conduct both an in-sample and out-of-sample forecasting analysis to see if the model has any useful forecasting ability in comparison to the Random Walk forecast made famous by Rogoff (1983a, 1983b). To do this, we look at four possible models encompassed by Equation (25): 1st Model: le t = Ω + δ 1 lM t + δ 2 lM * t + δ 3 li H t + δ 4 i * t + δ 5 ly t + δ 6 ly * t .…”

“…The original popular models used to determine and forecast the nominal exchange rates include the flexible price monetary models such as Frenkel and Johnson (1976), Musa (1976), and Bilson (1978a and1978b). The early tests of the monetary models using traditional econometric procedures were not particularly favorable either in terms of significance of the coefficients or in their in-sample or out-of-sample ability to forecast exchange rates, as shown by Rogoff (1983a, 1983b) who showed that exchange rate models fail to outperform a simple random walk. The early tests of the monetary models using traditional econometric procedures were not particularly favorable either in terms of significance of the coefficients or in their in-sample or out-of-sample ability to forecast exchange rates, as shown by Rogoff (1983a, 1983b) who showed that exchange rate models fail to outperform a simple random walk.…”

Exchange Rate Predictability in Finite Samples

“…In the seminal paper by Meese and Rogoff (1983a,b), the authors find that the random walk model can outperform all of the exchange rate models of the seventies in the out‐of‐sample fit. The failure to find evidence connecting economic models to the exchange rate has inspired subsequent researchers to analyse the robustness of these findings based on different samples, model specifications and explanatory variables.…”