Fuzzy cluster analysis using spectral analysis: With an application to stock price comovement

Kaiji Motegi, Kimiaki Shinkai, Hajime Yamashita, Shuya Kanagawa, Hiroaki Uesu

Research output: Contribution to journalArticlepeer-review


This paper investigates comovement of international stock prices from the viewpoint of both region and industry. Specifically, we apply the cluster analysis to recent daily stock price data of Asian and U.S. firms. The measure of comovement in this paper is the coherence, a well-known similarity measure in the frequency domain literature that can distinguish short-run and long-run comovement. The empirical results indicate that the country effect surpasses the industry effect in both short run and long run, i.e., shares from the same country tend to move together but shares within the same industry do not. This finding provides portfolio managers with a practical implication that choosing a country and then many kinds of industry therein is a riskier investment strategy than choosing an industry and then many countries. The dominant country effect also highlights a slow process of globalization. Another interesting finding is that both short-run and long-run correlations among shares have surged since the subprime mortgage crisis. This implies that the portfolio diversification effect has become smaller in recent stock markets.

Original languageEnglish
Pages (from-to)1417-1423
Number of pages7
JournalICIC Express Letters, Part B: Applications
Issue number5
Publication statusPublished - 2015 Jan 1
Externally publishedYes


  • Connecters
  • Fuzzy cluster analysis
  • Partition tree
  • Spectral analysis
  • Stock price
  • Zadeh’s method

ASJC Scopus subject areas

  • Computer Science(all)


Dive into the research topics of 'Fuzzy cluster analysis using spectral analysis: With an application to stock price comovement'. Together they form a unique fingerprint.

Cite this