Πρόβλεψη συναλλαγματικών ισοτιμιών: μπορούμε να νικήσουμε το μοντέλο τυχαίου περιπάτου
foreign exchange forecasting : can we beat the random walk model
Exchange rates are the quintessential international variable, a factor in virtually every international financial market decision. Alas, virtually all structural models of exchange rate behavior are empirically outperformed by a simple random-walk model. This conclusion might suggest that nothing can be confidently said about the economic determinants of exchange rate behavior. Still, the presence of nonrandom-walk behavior offers market economists an opportunity to add value. The analysis shows that news items that appear to be the same may differ in some critical respect, and that this causes the exchange rate to respond differently following what seem to be similar economic events. The bad news of course is that exchange rate modeling-not to mention exchange rate prediction-can be extremely complex. Still one can derive a set of basic principles insofar as exchange rate movement is concerned: 1.Only unanticipated events cause exchange rates to deviate from their expected path movement. 2.Factors that increase the demand for a currency tend to raise the price of that currency. 3.The “character” and the “context” of the economic news item will greatly influence the “nature” of the exchange rate response that follows. Empirical evidence suggests that during some periods-the 1920 hyperinflation, the early 1970s and over the longer run-exchange rate behavior is significantly related to fundamentals. Also, in the very short run exchange rates often closely follow fundamental events interpreted within the context of an asset model. Still, economic models of exchange rates have often been unreliable and unsuitable for forecasting.