THE FORWARD PREMIUM PUZZLE - THEORETICAL ANALYSIS WITH A USE OF MONTE-CARLO EXPERIMENT (Problem premii forward na rynku walutowym - analiza teoretyczna z zastosowaniem eksperymentu Monte-Carlo)

The article deals with Forward Premium Puzzle. The first part of the paper describes the current state of the art. The main focus of the article is the difference in the estimates of two regressions: the level regression and the forward premium one. The article shows that the opposite estimate of the parameter in two regressions may be caused by unobservable stationary variable representing jointly non-rational expectations forecast errors and risk premia. Series of Monte Carlo experiments present that 'level' regression of spot - forward relationship and forward premium regression under presence of unobservable variable may yield estimates suggesting different inference. The level regression shows that forward is a proper predictor of future spot exchange rate and the forward premium regression shows that future exchange rate change will move opposite to the direction suggested by the forward premium. Moreover Monte Carlo experiments suggest that the unobservable variable must be persistent and should also represent a certain level of variability. The properties of unobservable variable may be used in future to match the theoretical values coming from different economic models.