The purpose of this paper is to suggest a superior method for assessing mean stationarity of asset pricing effects.
The authors suggest the use of an F‐test to examine mean stationarity of asset pricing effects across subperiods. The superiority of this test is demonstrated through examination of the Halloween Effect using simulated data and the Morgan Stanley Capital International (MSCI) data for 18 developed economies.
It is found that the suggested F‐test provides results superior to a simple examination of the magnitude and statistical significance of estimated regression coefficients across subperiods when attempting to determine mean stationarity.
This paper sheds light on an analytical oversight in the asset pricing anomalies literature and suggests an appropriate test to address this oversight.
