WTO membership, ownership deregulation, and market efficiency: evidence from China

Abstract

We assess the impact of accession to the World Trade Organization (WTO) and stock ownership deregulation on the presence of anomalies in stock returns in China. We partition our data into sub-periods that mirror WTO accession and Investor-Base Expansion (IBE) in all share markets. Unlike prior research, we consider all mainland and Hong Kong exchanges from market inception to 2008, and find that Chinese stock markets may have become increasingly more efficient. However, we also document a positive Friday and turn-of-the-year effects that we attribute to greater integration with world stock markets. The daily and monthly anomalies documented here may well guide global investors and portfolio managers to strategically time the market, take suitable short and long positions, and shuffle assets to possibly outperform the market.

Keywords:

• WTO
• calendar anomalies
• market efficiency
• Chinese stock markets

JEL Classification:

• G14
• G15

Acknowledgements

We would like to thank Glenn Pettengill for his valuable comments. The usual caveat applies.

Notes

¹ The literature on the day-of-the-week anomaly provides several explanations for the Monday effect, which include stock settlement procedures that increase the number of sell transactions as compared to buy transactions on Monday, the release of unfavourable financial information by corporations during weekends, or excess demand on Friday and excess supply on Monday that result from speculative short positions as sellers close their positions on Friday and re-open them on Monday (e.g. Chen and Singal, 2003; Blau et al., 2009; Christophe et al., 2009).

² Two main hypotheses generally explain the higher return of stocks in January: tax-loss selling and window-dressing. According to the tax-loss selling hypothesis, investors sell individual stocks that are performing poorly in December to take in the capital loss for tax purposes and reduce their annual tax liability. The window-dressing hypothesis argues that institutional investors prefer to sell losing stocks in December before the holiday and to buy them back in January.

³ See for example, Aggarwal and Rivoli (1989) for Hong Kong, Singapore, Malaysia, and the Philippines; Davison and Faff (1999) for Australia; Wong and Yuanto (1999) for Indonesia; Brooks and Persand (2001) for South Korea, Malaysia, The Philippines, Taiwan and Thailand; and Hui (2005) for the USA, Japan, Hong Kong, Korea, Singapore and Taiwan.

⁴ The tax-loss selling hypothesis used to explain the turn-of-the-year effect in Western markets is irrelevant in China as there are no taxes on capital gains.

⁵ In their first study, Mookerjee and Yu (1999a) use 759 and 727 daily closing values for Shanghai and Shenzhen indices, respectively, and in the second study, they employ 833 and 904 daily closing values, respectively, for the same indices.

⁶ Similar to Chen et al. (2005), we exclude observations around 21 May 1992, when the government removed price limit restrictions on shares and market indices exhibited large jumps in returns. Although the government reintroduced price limits on 15 December 1996, we do not find similar gaps in returns around this date probably because the markets had already gained more depth by then.

⁷ The Shanghai ‘A’ shares index started on 19 December 1990; the Shanghai ‘B’ shares was developed on 21 February 1992; the Shenzhen ‘A’ shares opened on 3 April 1991; and the Shenzhen ‘B’ shares began on 28 February 1992; the Hang Seng Composite Index was first published on 24 November 1969, and the Hang Seng China Affiliated Corporations opened on 3 January 2000.

⁸ For robustness, we estimate an alternative equation to (1) that includes a constant in which the number of dummy variables is four instead of five for Tuesday through Friday. In these regressions, the constant denotes the omitted dummy variable for Monday, and we notice that its parameter estimate is identical to β₁. The other β coefficients differ in size but not in sign or significance, so that our main results hold.

⁹ For robustness, we also estimate an alternative equation to (3) that includes a constant and 11 dummy variables instead of 12 for all months except July. Our main results hold.

¹⁰ It is instead replaced by a marginally positive Monday effect.

¹¹ This view is consistent with the informational advantages of domestic investors (Hau, 2001; Chan et al., 2008).