Random walk in the forward foreign exchange market

The present study investigates the applicability of the random walk hypothesis to the forward exchange market of British pounds in terms of U.S. dollars. The variables examined are first order daily price differences as well as the weekly first order differences of the net speculative forward exchange function, which is measured by the residual of the traditional Interest parity formulation. The sample periods for the price data cover five periods of approximately 90 days each. Three belong to the narrow gold band period and cover pre and post devaluation (1967) days. The other two belong to the freely fluctuating period. The speculative expectation data are from three periods, two under the fixed exchange system and one tinder freely fluctuating system. Serial correlation as well as distributional evidence, two independent aspects of the random walk hypothesis are examined by a number of statistical tools, parametric as well as non parametric. For price changes data, violation of serial independence is indicated. The distribution appears to be closer to stable Paretian with < = 1 than normal. For expectations data, violation of random walk model is less serious and distributional evidence indicates normality. Therefore, a series of efficient behavioral inputs is seen to give rise to series of non efficient economic output series. This seems to indicate the inefficient aspects of government intervent ion in the forward market.