LAUSR

This study revisits calendar anomalies in Japanese stock returns to examine whether they can be explained by market conditions. Results of the OLS and GARCH (1,1) regression models show that most of the well-known calendar anomalies no longer exist in Japanese stock returns when conventional methodologies are used. These calendar anomalies became evident during the Japanese bubble period and disappeared subsequently. To provide new evidence on calendar anomalies in Japanese stock returns, we examine calendar anomalies based on market conditions. We show that the day-of-the-week, January, turn-of-the-month, Halloween and Dekansho-bushi effects became evident in UP market conditions only. They were never evident in DOWN market conditions. All these anomalies are still found to be significant in UP market conditions. Our explanation is consistent throughout the whole sample period and is robust against the choice of index used to measure market returns.