Why is it so difficult to outperform the random walk in exchange rate forecasting?

A simulation exercise is used to demonstrate the difficulty to outperform the random walk in exchange rate forecasting if forecasting accuracy is judged by the Root Mean Square Error (RMSE) or similar criteria that depend on the magnitude of the forecasting error. It is shown that, as the exchange rate volatility rises, the RMSE of the model rises faster than that of the random walk. While the literature considers this finding to be a puzzle that casts a big shadow of doubt on the soundness of international monetary economics, the results show that failure to outperform the random walk, in both in-sample and out-of-sample forecasting, should be the rule rather than the exception. However, the results do not imply that the random walk is unbeatable, because it can be easily outperformed if forecasting accuracy is judged according to criteria such as direction accuracy and profitability.     

A reappraisal of the Meese–Rogoff puzzle

Several explanations have been put forward for the Meese–Rogoff puzzle that exchange rate models cannot outperform the random walk in out-of-sample forecasting. We suggest that a simple explanation for the puzzle is the use of the root mean square error (RMSE) to measure forecasting accuracy, presenting a rationale as to why it is difficult to beat the random walk in terms of the RMSE. By using exactly the same exchange rates, time periods and estimation methods as those of Meese and Rogoff, we find that their results cannot be overturned even if the models are estimated with time-varying coefficients. However, we also find that the random walk can be outperformed by the same models if forecasting accuracy is measured in terms of the ability to predict direction, in terms of a measure that combines magnitude and direction and in terms of profitability.     

Demystifying the Meese–Rogoff puzzle: structural breaks or measures of forecasting accuracy?

Structural breaks have been suggested by several economists as a possible explanation for the Meese–Rogoff puzzle, in the sense that an exchange rate model can outperform the random walk in terms of the out-of-sample forecasting error if the period under investigation is free of structural breaks. The results indicate that structural breaks cannot explain the inability of the flexible price monetary model to outperform the random walk. The only plausible explanation for the Meese–Rogoff puzzle is that forecasting accuracy is traditionally assessed by magnitude-only measures. When forecasting accuracy is assessed by alternative measures that do not rely exclusively on the magnitude of error, the monetary model can outperform the random walk regardless of the presence or otherwise of structural breaks.     

Cointegration as an explanation for the Meese–Rogoff puzzle

Some economists suggest that the failure of exchange-rate models to outperform the random walk in exchange rate forecasting out of sample can be attributed to failure to take into account cointegration when it is present. We attempt to find out if cointegration matters for forecasting accuracy by examining the relation between the stationarity and size of the forecasting error. Results based on three macroeconomic models of exchange rates do not provide strong support for the proposition that cointegration matters for forecasting accuracy. The simulation results show that while stationary errors tend to be smaller than non-stationary errors, this is not a universal rule. Irrespective of the presence or absence of cointegration, none of the three models can outperform the random walk in out-of-sample forecasting, which means that cointegration cannot solve the Meese–Rogoff puzzle.     

Why is it so difficult to outperform the random walk? An application of the Meese–Rogoff puzzle to stock prices

Some economists suggest that the Meese–Rogoff puzzle is equally applicable to the stock market, in the sense that no model of stock prices can outperform the random walk in out-of-sample forecasting. We argue that this is not a puzzle and that we should expect nothing, but this result if forecasting accuracy is measured by the root mean square error (RMSE) and similar metrics that take into account the magnitude of the forecasting error only. We demonstrate by using two models for dividend-paying and nondividend-paying stocks that as price volatility rises, the RMSE of the random walk rises, but the RMSE of the model rises even more rapidly, making it unlikely for the model to outperform the random walk.     

Out‐of‐Sample Exchange Rate Forecasting and Macroeconomic Fundamentals: The Case of Japan

This study explores the respective out‐of‐sample exchange rate forecasting abilities of five macroeconomic fundamental models in comparison to a naïve random walk model for Japan during the post‐Bretton Woods era. To assess the influence of major economic changes, we estimate both linear and nonlinear models for all the macroeconomic fundamentals. Overall, most structural exchange rate models outperform a naïve random walk model in terms of forecasting accuracy in the short horizon. When the fundamentals are only linearly modelled, the forecasting ability of the Taylor rule is generally superior to other fundamental models. When the fundamentals are nonlinearly specified, the predictability of some other models rises dramatically to match that of the Taylor rule models in short and/or long horizons. Of importance, we determine that the yen/dollar exchange rate forecasting performance effectively improves in several fundamental models when influential economic changes are incorporated.     

A Comparison of Alternative Exchange Rate Forecasting Models*

This paper compares a number of structural and times-series models on the basis of their accuracy in forecasting the A ustralian-US dollar exchange rate out of sample. Purchasing power parity, forward exchange theory, static and dynamic specifications of both the flexible price and sticky price monetary models, and univariate ARIMA models are considered in the paper. Exchange rate forecasts are generated at horizons of one to four quarters. In contrast to overseas results which support the view that the exchange rate follows a random walk, several models in this study are found to generate forecasts superior to the random walk model.     

Dynamic forecasting of monetary exchange rate models: Evidence from cointegration

The Frenkel-Bilson and Dornbusch-Frankel monetary exchange rate models are used to estimate the out-of-sample forecasting performance for the U.S. dollar/Canadian dollar exchange rate. By using Johansen's multivariate cointegration, up to three cointegrating vectors were found between the exchange rate and macroeconomic fundamentals. This means that there is a long-run relationship between the exchange rate and economic fundamentals. Based on error correction models, two monetary models outperform the random walk model at the three-, six-, and 12-month forecasting horizons. Therefore, monetary exchange rate models are still useful in forecasting exchange rates.     

Can Economists Forecast Exchange Rates? The Debate Re‐Visited: The Case of the USD/GBP Market

In the debate on forecasting exchange rates, critics claimed that traditional macroeconomic models could not outperform a random walk in post‐sample forecasts. Perceived deficiencies include inadequate allowance for simultaneity, and expectations hypotheses inconsistent with the structure of models employed. This paper re‐visits the debate, first to address critics' major concerns, and second because in the view of the present authors, the debate closed on an unduly pessimistic note. This paper develops a simultaneous, rational expectations model of the USD/GBP market, with functional relationships for hedgers, speculators and a spot rate equation. The model is estimated with data contemporaneous to the debate, including a period during which the US Commodity Futures Trading Commission did not collect data on traders' open positions. Using the results of post‐debate research on tests for stationarity with missing observations, the model, using only public information, outperforms a random walk in post‐sample forecasts of the spot rate. Recent microstructure models of the exchange rate based on order flow have re‐kindled the forecasting debate. The model developed here, however, is differentiated from these microstructure models, first because order flow utilises both public and private information, and second because the microstructure models do not directly address critics' concerns.     

Forecasting the yield curve and the role of macroeconomic information in Turkey

In this study we investigate the yield curve forecasting performance of Dynamic Nelson–Siegel Model (DNS), affine term structure VAR model (ATSM VAR) and principal component model (PC) in Turkey. We also investigate the role of macroeconomic variables in forecasting the yield curve. We have reached numbers of important results: 1—Macroeconomic variables are very useful in forecasting the yield curve. 2—The forecasting performances of the models depend on the period under review. 3—Considering the structural break which associates with change in monetary policy leads models to produce better forecasts than the random walk. 4—The role of exchange rate should not be ruled out in forecasting the yield curve in an emerging market like Turkey.     

The random walk as a forecasting benchmark: drift or no drift?

We examine the proposition that the random walk without drift is more powerful in predicting exchange rates than the random walk with drift. It is demonstrated that there is no theoretical reason why the random walk without drift always outperforms the random walk with drift and that this is an empirical issue. The results show that while the random walk without drift can outperform the random walk with drift in terms of the RMSE, it fails to do so in terms of the ability to predict the direction of change, measures that take into account magnitude and direction, and in terms of profitability. If the drift factor is allowed to change over time by estimating the model in time-varying parameter terms, the random walk with drift performs even better.     

Forecasting exchange rates: The multi-state Markov-switching model with smoothing

This paper presents an exchange rate forecasting model which combines the multi-state Markov-switching model with smoothing techniques. The model outperforms a random walk at short horizons and its superior forecastability appears to be robust over different sample spans. Our finding hinges on the fact that exchange rates tend to follow highly persistent trends and accordingly, the key to beating the random walk is to identify these trends. An attempt to link the trends in exchange rates to the underlying macroeconomic determinants further reveals that fundamentals-based linear models generally fail to capture the persistence in exchange rates and thus are incapable of outforecasting the random walk.     

Random Walk or Bandwagon: Some Evidence From Foreign Exchanges in the 1980s

The predicitive performance of the bandwagon expectations model foe weekly spot exchange rates over the 1980–6 period is evaluated. Empirical results generally indicate the presence of significant bandwagon effects in the exchange rate dynamics, as found in survey expectations data. Specifically, we find the the bandwagon forecasting scheme can improve the forecasting accuracy in terms of both mean squared errors and market timing upon the random walk and vector autoregressive models. The results illustrate that bandwagon expectations can be rational, and the exchange rate appears to follow a more general integrated process than a random walk.     

The Taylor Rule,Wealth Effects and the Exchange Rate

In this study, we develop the Taylor rule and Taylor rule‐based exchange rate models that consider wealth effects as represented by both asset prices and asset wealth. Using data for Australia, Sweden, the UK and the USA, we find that effects of asset prices and wealth on the Taylor rule vary depending on the country and on the form that wealth takes. Out‐of‐sample forecasting capacities of the wealth‐augmented Taylor rule model and Taylor rule‐based exchange rate model outperform conventional models and random walk theories for these countries.     

FORECASTING THE UK/US EXCHANGE RATE WITH DIVISIA MONETARY MODELS AND NEURAL NETWORKS

This paper compares the UK/US exchange rate forecasting performance of linear and nonlinear models based on monetary fundamentals, to a random walk (RW) model. Structural breaks are identified and taken into account. The exchange rate forecasting framework is also used for assessing the relative merits of the official Simple Sum and the weighted Divisia measures of money. Overall, there are four main findings. First, the majority of the models with fundamentals are able to beat the RW model in forecasting the UK/US exchange rate. Second, the most accurate forecasts of the UK/US exchange rate are obtained with a nonlinear model. Third, taking into account structural breaks reveals that the Divisia aggregate performs better than its Simple Sum counterpart. Finally, Divisia‐based models provide more accurate forecasts than Simple Sum‐based models provided they are constructed within a nonlinear framework.     

Forecasting US recessions with various risk factors and dynamic probit models

This paper extends probit recession forecasting models by incorporating various recession risk factors and using the advanced dynamic probit modeling approaches. The proposed risk factors include financial market expectations of a gloomy economic outlook, credit or liquidity risks in the general economy, the risks of negative wealth effects resulting from the bursting of asset price bubbles, and signs of deteriorating macroeconomic fundamentals. The model specifications include three different dynamic probit models and the standard static model. The out-of-sample analysis suggests that the four probit models with the proposed risk factors can generate more accurate forecasts for the duration of recessions than the conventional static models with only yield spread and equity price index as the predictors. Among the four probit models, the dynamic and dynamic autoregressive probit models outperform the static and autoregressive models in terms of predicting the recession duration. With respect to forecasting the business cycle turning points, the static probit model is as good as the dynamic probit models by being able to flag an early warning signal of a recession.     

Forecast performance of exchange rate models revisited

This paper re-evaluates the performance of reduced form exchange rate models by updating the Messe-Rogoff study (1983). This paper confirms earlier tests showing that simple monetary models do not perform well, but it finds more positive results for other monetary models that incorporate more dynamic econometric specifications. A simple error correction monetary model out-forecasts a random walk almost half of the time.     

Professional forecasts of interest rates and exchange rates: Evidence from the Wall Street Journal’s panel of economists

We analyze economists’ forecasts of interest rates and exchange rates from the Wall Street Journal. We find that a majority of economists produced unbiased forecasts but that none predicted directions of changes more accurately than chance. Most economists’ forecast accuracy is statistically indistinguishable from a random walk model in forecasting the Treasury bill rate, but many are significantly worse in forecasting the Treasury bond rate and the exchange rate. We also find systematic forecast heterogeneity, support for strategic models predicting the industry employing the economist matters, and evidence that economists deviate less from the consensus as they age.     

Forecasting the price of gold

This article seeks to evaluate the appropriateness of a variety of existing forecasting techniques (17 methods) at providing accurate and statistically significant forecasts for gold price. We report the results from the nine most competitive techniques. Special consideration is given to the ability of these techniques to provide forecasts which outperforms the random walk (RW) as we noticed that certain multivariate models (which included prices of silver, platinum, palladium and rhodium, besides gold) were also unable to outperform the RW in this case. Interestingly, the results show that none of the forecasting techniques are able to outperform the RW at horizons of 1 and 9 steps ahead, and on average, the exponential smoothing model is seen providing the best forecasts in terms of the lowest root mean squared error over the 24-month forecasting horizons. Moreover, we find that the univariate models used in this article are able to outperform the Bayesian autoregression and Bayesian vector autoregressive models, with exponential smoothing reporting statistically significant results in comparison with the former models, and classical autoregressive and the vector autoregressive models in most cases.