Multifractal Analysis of the Korean Stock Market

Abstract

We analyze the multifractal properties of the daily prices of individual stocks in the Korean Composite Stock Price Index (KOSPI) from 1980 to 2003 by using raw price data and two artificial data sets: one from which a temporal autocorrelation in the original data was eliminated by randomly shuffling the time, and one from which the nonlinearity in the original data was eliminated by randomly reshuffling the phase. Individual stock return time series were multifractal in all data sets, but the artificial data sets had a, narrower spectrum than the original data. To observe the possible factors that cause multifractality, we took the KOSPI market index as a common factor. We generated an artificial time series reflecting the market factors from the stock market by using the one-factor model and calculated its multifractal spectra. We also found that although the market factor could generate the complexity, not all of the complexity of the KOSPI market could be explained.