Title: Dynamic correlation network analysis of Japanese stock returns
Authors: Takashi Isogai - Tokyo Metropolitan University (Japan) [presenting]
Abstract: The dynamic correlation network of Japanese stock returns is analyzed to study the correlation structure as well as correlation dynamics of the stock market empirically. Stock groups are generated by correlation network clustering in order to work around the high dimensionality problem due to the large number of stocks. Such data-oriented group definition is more reliable than the existing sector classification. Homogeneous groups of stocks in a balanced size are created by segregating the whole stock returns by recursive modularity optimization; a single portfolio that comprises group portfolio returns is also created. Thus, within- and between-group dynamic correlation networks are built, respectively. We, first, confirmed that a higher level of correlation is observed during the crisis periods, namely after the Lehman shock and the Great East Japan Earthquake. We also identify significant differences in the pattern of correlation dynamics between groups. Then, dynamic changes in the network topology measures including density, centrality, and heterogeneity are examined in each correlation network; significant changes in network topologies are detected.