Do Spanish stock market prices follow a random walk?

Abstract

In this article we test the random walk hypothesis in the Spanish daily stock market prices by means of using fractionally integrated techniques. We use a version of the tests of Robinson (1994) that permit us to test I(d) statistical models. The results show that though fractional degrees of integration are plausible in some cases, the confidence intervals are generally narrow, including the unit root in all cases. Therefore, there is very little evidence of fractional integration, despite the length of the series, implying that the standard practice of taking first differences when modelling stock prices is adequate. In addition, the tests cannot reject that the underlying I(0) disturbances are white noise, supporting thus the (weakly) efficient market hypothesis in the Spanish stock market.