ABSTRACT
Most of the empirical studies that have examined the weak-form EMH have investigated whether stock price changes follow a Random Walk 3. Given the assumption underlying this version, they have employed statistical techniques aimed at detecting linear structure in time series, such as the autocorrelation test of Box and Pierce (1970). However, as emphasized
by Granger and Anderson (1978), absence of linear dependence does not necessarily imply independency, but merely a lack of linear autocorrelation. Furthermore, several empirical studies have applied tests, which were originally designed for fi elds such as physics, capable of detecting linear as well as nonlinear patterns in data, and report evidence of nonlinear structure in economics and fi nancial data (see, e.g., Hsieh, 1989; Scheinkman and LeBaron, 1989).
