Abstract
In this paper, the analysis of the random character in stock market prices was performed. Firstly, the mechanism of variations in stock prices was considered as a feedback system which consisted of stock prices as andogenous variables and peculiar noises as exogenous variables. Then, impulse response functions and relative power contributions were estimated by using the method of system identification through the multiple autoregressive model and interactions among stock prices were investigated. The results of the analysis for stock prices of three beer companies in Japan showed good agreement with the actual situation. Consequently, it was found that modeling the mechanism of variations in stock prices as a feedback system was effective.
