Stock returns around nontrading periods: evidence from an emerging market

Abstract

I examine intraday stock returns in the Istanbul Stock Exchange (ISE) around nontrading periods – weekends and holidays – by utilizing the exchange's structure of two trading sessions. I find that returns are generally more positive in the last session on Fridays and more negative in the first session on Mondays. The results also indicate that the weekend effect has disappeared in the ISE in recent years. I further find some evidence that there is a relationship between the length of a holiday nontrading period and returns around it. The longer a nontrading period is, the more positive the returns are in the morning session before the holiday and the less positive the returns are in the morning session after the holiday. My findings indicate the importance of the uncertainty imposed on stock returns by the length of a nontrading period.

Keywords:

• day of the week effect
• holiday effect
• calendar anomalies
• market efficiency
• nontrading period
• session returns

JEL Classification::

• G14
• G15

Notes

¹ Zhang et al. (2008) argue that political window dressing can also create financial anomalies. Additionally, Ortiz et al. (2010) examine the Spanish market for anomalies and find that tax considerations extend the life of the January effect in the Spanish market.

² Unlike some stock markets, the ISE did not impose any restrictions on short selling during the global financial crisis of 2008.

³ I should note that the first session return following a holiday is less positive compared to the second session return.

⁴ The ISE does not have an index for small stocks. It is possible that the results presented here are confined to large stocks only.

⁵ The period from 9:30 am to 9:45 am was the call auction period with no trading. There was no opening procedure for the second session during the sample period.

⁶ It is also possible that a religious holiday and an official holiday may occur close to each other and the exchange may be closed up to 10 days or more.

⁷ Labor and Solidarity Day was declared an official holiday in 2009.

⁸ Recent studies indicate that seasonal anomalies have been disappearing or their effects have been diminishing in recent years. For example, Steeley (2001) shows that the weekend effect has disappeared in the UK market, and Coutts and Sheikh (2002) provide evidence for the disappearing weekend, January and holiday effects in the Johannesburg Stock Exchange. Similarly, Gu (2003) reports that the January effect in US markets is declining and has even disappeared for some indices. Recently, Marquering et al. (2006) note that several anomalies, including the weekend effect, the holiday effect, the time-of-the-month effect and the January effect have disappeared after being documented in academic studies.

⁹ It can be argued that the use of monthly short interest data to examine returns around nontrading periods is problematic. However, as argued in Chen and Singal (2003), the short interest data are only used to identify stocks with high short interest and thus should not bias the results.

¹⁰ In untabulated analyses, I use all the stocks and obtain quantitatively similar results.

¹¹ There are 12 holidays between January 2009 and October 2010 and nine are short and three are long holidays.

¹² The tables are available upon request.