Spurious Correlation in Exchange Rate Target Zone Modelling: Testing the Drift Adjustment Method on the US Dollar, Random Walk and Chaos

The drift-adjustment method estimates the expected rate of depreciation within an exchange rate band by simple equations. Papers applying this method claim that, while forecasting a freely floating currency is hopeless, predicting an exchange rate within the future band is successful. This paper shows that the results achieved by applications to EMS and Nordic currencies are not specific to data of target zone currencies. For example, application to US dollar leads qualitatively to the same result as application to EMS currencies. Simulation evidence suggests that the closer the dominant inverted autoregressive root to unity the higher the chance of reproducing the empirical target zone results, since the finite sample biases of the parameters of interest in the unit root case are such that the random walk seemingly significantly fits the model in the vast majority of experiments. HAC standard errors do not help much in hypothesis testing either. The paper develops a simple model coinciding with stylized facts of target zones that demonstrates the unpredictability of the expected rate of depreciation within the band. Surprisingly, application of drift-adjustment method to a process switching between stationary periods and chaotic periods, the fit is similar to reported target zone results.