Day-of-the-week effect in anomaly returns: International evidence