The random walk versus unbiased efficiency: can we separate the wheat from the chaff?

In the issue under investigation the wheat cannot be separated from the chaff because all we have is chaff. The random walk and unbiased efficiency are equally bad forecasters that should not be used as benchmarks for measuring forecasting accuracy. The simulation and econometric results show that when two forecasters are not independent they produce forecasts of similar quality, in which case it is erroneous to use the forward rate as a forecaster and the spot rate as a benchmark. Theoretical and intuitive explanations are presented for why the random walk and unbiased efficiency produce poor-quality forecasts that are almost identical. The failure of unbiased efficiency is explained primarily in terms of the post Keynesian view of the forward exchange rate.