국제주식시장지수의 장기기억성에 관한 실증연구

The literature on market efficiency is vast as the theme is of interest for both practitioner and academics. Since it is a very intriguing issue, a big part of this literature focuses on seeking long memory dependency in asset return. Actually, if the returns present long range dependence, the random walk hypothesis is not valid anymore and neither does the market efficiency hypothesis. Moreover, the presence of long range dependence in asset returns contradicts the weak form of market efficiency which states that, under the information contained on the set formed by past returns, future returns are unpredictable. This paper investigated the long-term memory for international stock market indices, using Hurst exponent estimated by DFA(detrended fluctuation analysis) method proposed by Peng et al.(1994) and tried to observe results in case of total period and moving period for robustness. The observed results is that some market indices(Chile, Indonesia, Thailand) possess strong long-term memory based on estimated Hurst exponent, while almost market indices don't reject random process in established Hurst exponent range(). And in the various results, it is not to be denied that characteristics of long-term memory is related their stage of development.