We investigated financial market data to determine which factors affect information flow between stocks. Two factors, the time dependency and the degree of efficiency, were considered in the analysis of Korean, the Japanese, the Taiwanese, the Canadian, and US market data. We found that the frequency of the significant information decreases as the time interval increases. However, no significant information flow was observed in the time series from which the temporal time correlation was removed. These results indicated that the information flow between stocks evidences time-dependency properties. Furthermore, we discovered that the difference in the degree of efficiency performs a crucial function in determining the direction of the significant information flow.
Recently, researchers have become interested in the information flow occurring between financial assets or markets in an effort to understand the nature of the interaction between assets and the pricing mechanism in markets [1], [2], [3], [4], [5], [6], [7], [8], [9], [10], [11], [12], [13], [14], [15], [16], [17], [18], [19], [20], [21] and [22]. The relationship between spots and derivatives has represented the normal course of study, particularly the manner in which the derivatives that transact with future prices affect the spots [1] and [2]. The information flow from developed markets to emerging markets is also an issue in which active research is being conducted [3], [4], [5], [6] and [7]. In addition, the information flow with regard to synchronization, integration and segmentation between financial markets by internal and external events has been assessed [8] and [9]. Previous studies have attempted to analyze financial data using statistical method including the Granger causality test, the VAR (vector-autoregressive) model, and the GARCH (generalized autoregressive conditional heteroskedasiticity) type [10], [11] and [12]. However, studies regarding the factors that significantly affect information flow have proven insufficient. Therefore, we have attempted to determine empirically which factors are crucial to the information flow, considering particularly the following factors: the time-dependency property, and differences in the degree of efficiency.
According to the results of previous studies, the financial time series is time-dependent, and the time sequence exerts a significant effect on the information flow. That is to say, in financial markets, there exist many internal and external events which, as time passes, induce price changes via the interactions between stocks at the times that these events occur. In other words, the time scale of return performs a crucial function in the information flow. We have noted that the time scale of return corresponds to the time intervals, particularly when the prices are converted into the returns. Also, the efficiency of information is crucial to the pricing mechanism [23]. The price change in a given individual stock differs from that of others, even though the same information is both instantaneously and fully reflected. This suggests that the degree of efficiency differs for each stock. Therefore, the degree of efficiency significantly affects the information flow. The efficiency we assessed in this study is based on the weak-form efficient market hypothesis (EMH), which assumes that the similarity of past price change patterns are useful in predicting future price changes. We have utilized the approximate entropy (ApEn) method in order to observe the randomness in the time series [24]. The ApEn method quantitatively calculates complexity, randomness, and prediction power. As the frequency of similarity patterns in the price changes is high, both the randomness and the ApEn remain low. Previous studies have argued that the ApEn evidences significant information by which the degree of efficiency can be measured [25], [26] and [27].
We have investigated individual stocks traded in the stock markets of Korea, Japan, Taiwan, Canada, and the USA. The entire interactions between stocks are considered, such that the number of interactions is N(N−1)/2N(N−1)/2, where NN is the number of stocks. This technique provides sufficient information flow in the market, allowing us to discover the characteristics of information flow within the context of the whole market. We detected a negative relationship between the time scale of return and the frequency of significant information flow, which supports the notion that the information flow between stocks evidences a time-dependency property. Also, we discovered that the difference in the degree of efficiency between stocks performs a crucial function in determining the direction of the information flow.
In the next section, we describe the data and the methods of the test procedures employed herein. In Section 3, we present the results obtained in accordance with our established research aims. Finally, we have summarized the findings and conclusions of this study.
