||Thailand’s 2011 Flooding: its Impacts on Japan Companies in Stock Price Data
Thailand’s 2011 Flooding: its Impacts on Japan Companies in Stock Price Data
ムハンマド, フィルダウス ルビス ,
白田, 由香利リリ, フィトゥリ サリー
121 , 2015-10-01
This paper aims to find which Japan companies whose stock prices were damaged by the flood happened in Thailand in 2011. Many Japanese companies were devastated by the floods. We analyze the matrix of return rates of the largest 225 Japanese stock prices. We proposed the following approach: First, the matrices U, W, and VT were obtained by using Singular Value Decomposition (SVD) on the standardized form of the return of stock price matrix. Then, two kinds of eigenvectors are introduced: Brand-eigenvector, obtained by multiplying U with W, and Dailymotion-eigenvector, obtained by multiplying S with VT. Each of them has their own role in the analysis: the Brand-eigenvector decides which company group had been impacted at the time while the Dailymotion-Eigenvector provides solid proof on the decision of the Brand-Eigenvector, showing representative time series fluctuations on the company cluster. The well-known facts that Japan digital camera companies, Nikon, Casio, and Sony, were damaged due to the flood were also used to help the analysis. We use the element of Nikon as the clue to find the damaged Japanese company cluster. By doing these steps, we can identify the correlated clusters of stock return rates. In addition, we utilize Random Matrix Theory (RMT) to identify the random behavior of stock return rates. First, we find the relation between Brand-Eigenvector and crosscorrelation matrix eigenvector and between Brand-Eigenvector singular value and cross-correlation matrix eigenvalue. Then, the distribution of the cross-correlation matrix eigenvalues and the elements of all Brand-Eigenvectors are inspected to see their consistency with RMT. By doing so, we are able to see the flood impact on Japanese company return stock rate as a whole for a certain period. In implementation, several tests were conducted. Finally, the Brand-eigenvector # 9 was found to be the Brand-eigenvector which best expressed the flood effect on the stock price change at the time, compared with other candidates (Brand-eigenvector # 19 and # 27). By inspecting the Brand-eigenvector # 9, we 【102 頁】 found that the cluster included several Japan companies damaged due to the floods. It was also found that there were similarities, in the stock pattern and decline, between the Dailymotion-eigenvector # 9 and the Nikon daily stock price during the last week of September and the first week of October. The dates were the period when the flood damage had intermittently happened owing several typhoons. In addition, through the inspection of the cross-correlation eigenvectors and eigenvalues, we found that the flood damage had a great impact on several Japanese companies during certain period.