Correlation, Network and Multifractal Analysis of Global Financial Indices

TL;DR

This study analyzes global financial indices using correlation, network, and multifractal methods, revealing how crisis periods affect clustering and multifractality, with implications for understanding market interconnectedness.

Contribution

It introduces a comprehensive analysis combining RMT, network, and multifractal methods to study global financial indices before and during the 2008 crisis, highlighting sector formation and multifractality variations.

Findings

Network clustering reflects geographical and crisis-related changes.

European markets form tight clusters at high thresholds.

Indices show varying degrees of multifractality linked to market stability.

Abstract

We apply RMT, Network and MF-DFA methods to investigate correlation, network and multifractal properties of 20 global financial indices. We compare results before and during the financial crisis of 2008 respectively. We find that the network method gives more useful information about the formation of clusters as compared to results obtained from eigenvectors corresponding to second largest eigenvalue and these sectors are formed on the basis of geographical location of indices. At threshold 0.6, indices corresponding to Americas, Europe and Asia/Pacific disconnect and form different clusters before the crisis but during the crisis, indices corresponding to Americas and Europe are combined together to form a cluster while the Asia/Pacific indices forms another cluster. By further increasing the value of threshold to 0.9, European countries France, Germany and UK constitute the most tightly linked markets. We study multifractal properties of global financial indices and find that financial indices corresponding to Americas and Europe almost lie in the same range of degree of multifractality as compared to other indices. India, South Korea, Hong Kong are found to be near the degree of multifractality of indices corresponding to Americas and Europe. A large variation in the degree of multifractality in Egypt, Indonesia, Malaysia, Taiwan and Singapore may be a reason that when we increase the threshold in financial network these countries first start getting disconnected at low threshold from the correlation network of financial indices. We fit Binomial Multifractal Model (BMFM) to these financial markets.