Stock Return Predictability and the Drift Between the Outcomes of Portfolio Investment Strategies

Anomalies found in tests of market efficiency do not necessarily imply that security prices do not reflect all available information, as the asset-pricing model used to describe the return generating process might also be false. In the present study, this joint hypothesis problem does not arise, because no use is made of an asset-pricing model. Instead, stock return predictability is tested by verifying whether the underlying variables of the drift between different types of indexes are correlated. This unambiguously tests for the sources of return predictability, which can be related to empirical anomalies, such as the "firm-size effect" and the "winner-loser effect". The drift between indexes is large if the (cross-sectional) variation of the underlying variables is large relative to their mean values, and vice versa. The size-related drift, for instance, is shown to be particularly large, but it also appears to be easily rendered statistically insignificant