Abstract
Using monthly yield data on straight bonds, this article investigates seasonality in the Japanese corporate bond market. A statistical examination of spreads between the yield of each bond and a bond market index reveals that the yield spread consistently decreases from April to August, whereas it increases from September to December. Because accounting year-ends for most investors in Japan are concentrated in either March or December, this seasonality supports the hypotheses of tax-loss selling and window dressing. Moreover, the seasonality becomes more pronounced as the debt rating declines, consistent with the findings in previous studies investigating the US bond market.
Acknowledgements
An earlier version of this article was presented at the Annual Meeting of the American Accounting Association, Anaheim, 2–6 August 2008, and the 31st Annual Congress of the European Accounting Association, Rotterdam, 23–25 April 2008. The author gratefully acknowledges Yasuhiro Arikawa, Kentaro Koga, Junsuke Matsuo, Kenji Miyazaki, Hatsuru Morita, Seiji Nakajima, Yasuhiro Ohta, Takao Saga, Kazutaka Takechi, Keisuke Takeguchi, Wataru Tanaka, Ken-ichi Tatsumi, Katsuyuki Tokushima, Shinki Tomozuka, Daisuke Tsuruta, Konari Uchida, and Peng Xu for their insightful comments. The research for this article is supported by a Grant-in Aid for Scientific Research (B) (19330066) from the Japan Society for the Promotion of Science. The author is solely responsible for any remaining errors.
Notes
1 Matsui (Citation2006) investigated the issue conditions of corporate bonds determined in the primary market in order to explore the pricing mechanisms of debt Initial Public Offering (IPO) and Secondary Public Offering (SPO) in Japan. He found that the issue price of a bond tends to be significantly overpriced on average compared with the bond's price in the secondary market.
2 On the other hand, substantial empirical studies have investigated seasonality of stock prices not only in advanced countries but also in emerging economies (e.g., Cheung and Coutts, Citation1999; Coutts and Sheikh, Citation2000; Jarrett and Kyper, Citation2005; Asteriou and Kavetsos, Citation2006; Ushad, Citation2009; Chang et al., Citation2010).
3 The null hypothesis remains rejected even if the time trend term, TIMEt , is excluded from Equation Equation2.
4 Moreover, both the LLC and the IPS tests rejected the null hypothesis even if the time trend variable is excluded.
5 JCR and NIS merged in April 1998 to become R&I.