Is South Korea's stock market efficient? A note

Abstract

In an attempt to examine efficiency of South Korea's stock market (SKM), Narayan and Smyth (2004) used a battery of unit root tests to investigate the random walk hypothesis and on the basis of the reported evidence for unit root, they concluded that the SKM is efficient. The authors have unfortunately confused random walk with unit root hypothesis. The present note stresses the fact that it is insufficient to test for stationarity when examining efficiency, casting serious doubt on Narayan and Smyth's conclusions. Furthermore, we provide comments on the shortfalls of the unit roots tests employed and not accounted for in Narayan and Smyth's study which may lead to spurious results.

Notes

¹ Given how broad the original EMH was, Fama (1970) divided the theory into three sub-hypotheses: The ‘Weak’ form states that all past market prices and information are fully reflected in securities prices. The ‘Semistrong’ form asserts that stock prices fully reflect all historical information and all current publicly available information. Finally, the ‘Strong’ form assumes that prices reflect not just historical and current publicly available information, but insider information. Fama (1991) presents a new, more appropriate classification using different types of tests: (1) Weak-form tests: return predictability with any type of information, (2) semi-strong form tests: event study and (3) strong-form tests: private information.

² As an example of tests capable of detecting linear and nonlinear dependence we refer to the BDS test of Brock et al . (1996), the NGEGM test of Nychkra et al. (1992) and the Hinich bispectrum test of Hinich (1982).

³ Kim et al . (2000) showed that Zivot–Andrews approach may provide fallacious results when a true second structural break is not accounted for. Although, N&Y have also used Lumsdaine–Papell test to allow for a second structural breaks, the above critics holds still.