会議名: 第23回バイオメディカル・ファジィ・システム学会
回次: 23
開催地: 北九州
開催日: 2010/10 -
This paper applies the fuzzy cluster analysis to six international stock markets (i.e., France, Germany, the U.S., Canada, Japan, and China). We analyze the similarity values calculated from the monthly stock returns, using both Zadeh's and Ward's clustering methods. These two methods induce different partition trees. Both the Minkowski distance and cophenetic correlation coefficient choose Zadeh's method over Ward's method. The optimal partition tree and the associated optimal number of clusters suggest (1) the similarity of European countries, (2) the institutional regularity of the Chinese stock market, and (3) the strong influence of the U.S. These implications are consistent with standard economic theory, and are preserved under the addition of three countries (i.e., the U.K., Mexico, and South Korea).