Examining market efficiency for large- and small-capitalization of TOPIX and FTSE stock indices

Abstract

This article uses parametric and nonparametric Variance Ratio (VR) tests of Lo and Mackinlay (1988) and Wright (2000) to re-examine the weak-form Efficient Market Hypothesis (EMH) for the large- and small-capitalization stock indices of TOPIX (Tokyo Stock Price Index) and FTSE (Financial Times Stock Exchange). Unlike the previous studies, the multiple VR test of Chow and Denning (1993) is the first extended to the nonparametric VR test of Wright (2000) as suggested by Luger (2003). The empirical results show that the weak-form EMH is supported for large-cap stock indices, but rejected for small-cap ones. This conclusion is further confirmed by using a rolling multiple VR tests.

Acknowledgements

The author would like to thank Jorge Belaire-Franch for kindly sharing their computer codes and also two anonymous referees for their helpful comments

Notes

¹ Contrary to Liu and He (1991), Fong et al. (1997) found much weaker evidence against the null hypothesis once the multiple variance ratio test is applied.

² As suggested by Luger (2003), the multiple variance ratio tests of CHODE (1993) and Cecchetti and Lam (1994) could be extended to Wright's nonparametric variance ratios to control the joint test size.

³ We follow the works of Belaire-Franch and Opong (2005a, b).

⁴ CHODE (1993) modified and extended the variance ratio of LOMAC into multiple setting. As indicated by CHODE, a nominal 100 percent critical value is not appropriate for each q selected due to failing to control overall test size for multiple comparisons, hence leading to an inappropriately large probability of Type I error. That is, the application of VR tests for multiple q values induces over-rejection of the null hypothesis, above the nominal size.

⁵ The detailed introduction TOPIX can refer the ‘New Index Series Guidebook’ of TSE.