Long-memory of Shanghai stock market: a wavelet-based approach

The detection of long memory processes has crucial implications for the measurement of the efficiency of financial markets. But parametric methods are computationally expensive and subject to mispecification. So in the paper, a semi-parametric estimation with maximal overlap discrete wavelet transform (MODWT) is used to handle 1-minute index time series of Shanghai stock market of China. Multiresolution analysis of index data reveal this stock markets could consist of multiple layers of investment horizons. With the help of wavelet variance, empirical result shows the presence of long-memory.