Network Analysis of Nikkei 225 Constituents Based on Granger Causality and a Proposal for a New Economic Indicator

Abstract

This study constructs a corporate network among the 225 constituents of the Nikkei Stock Average by conducting pairwise Granger causality tests on closing prices, identifying statistically significant temporal dependencies. Centrality metrics such as PageRank are applied to quantify each firm's influence within the network, serving as a basis for selecting constituents for a novel economic indicator. Comparative analysis using alternative selection criteria, including Sharpe ratio and return, demonstrates that the proposed indicator exhibits higher correlation and explanatory power than conventional average-based indices.