國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
7
behaviours. Moreover, Barber et al. (2008) added that when retail investors buy certain stocks in crowd to raise the stock prices, the price of the stocks would fall later.
Andrade, Chang, and Seasholes (2008) reported that the herding phenomenon exists in the Taiwan stock market.
1.4 Short-term Price Momentum and Long-term Price Reversal
Daniel, Hirshleifer, and Subrahmanyam (1998) proposed a hypothesis that irrational investors could influence short-term stock prices, which can go above or below their fundamental value. They believed that the phenomenon will last for a long time but eventually the stock prices will reverse back to their fundamental value.
1.5 Stock Trading Volume
A number of empirical studies have documented that there is a positive correlation between stock trading volume and the stock’s absolute price changes (see Karpoff,
1987). Lee and Swaminathan (2000) pointed out that the trading volume of a stock could represent the past performance of the stock. They added that if the past performance of a stock is good, the stock would attract more attention and make investors buy the stock more, leading to an increase in trading volume.
2. Related Works
There are other works that also used an empirical approach to study the impact of
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
8
investors’ attention and sentiments on stock markets. First, we focus on works using Google SVI to study different markets, and then compare their results with ours. Next, we discuss about works using trading volume and various market interview sentiment indexes to predict stock performance.
2.1 Google Search Volume Index
Da, Engelberg, and Gao (2009) is the first to conduct research using Google SVI data on the U.S. stock market. They found there is a positive relationship between the SVI and the returns of Russell 3000. They pointed out that the increase of the SVI would increase the turnover of Russell 3000 for two weeks. For the Taiwan stock market, our investigation indicates that the change of SVI can predict the average returns of TAIEX from the first to the fifth week, except the second week. Another study using Google SVI data is on the largest 30 stocks traded in NYSE by Vlastakis and Markellos (2012). They showed that SVI is positively related to the stock trading volume and the stock return volatility of the U.S. stock market. Moreover, they reported that there is a positive link between the information investors received and the risk aversion, based on the expected variance risk premium.
Focused on the German stock market, Bank, Larch, and Peter (2011) also paid attention to risk liquidity and reported that the higher the Google search volume is, the
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
9
higher the trading volume and the more improved stock liquidity are, leading to higher returns of the German stocks in the short run. They believed that attention-grabbing stocks are the subjects of temporary buying interests, causing price pressure and then leading to higher stock prices.
For the Japan stock market, Takeda and Wakao (2014) focused on the relationship between the intensity of SVI, the returns of Nikkei 225, and their trading volume. They reported that the relationship between the SVI and the trading volume is strong but the relation between the SVI and the Nikkei 225 return is not significant, probably because the sampling period included major negative economic shocks, such as the 2008 world financial crisis and the 2011 Great East Japan Earthquake.
For the France stock market, Aouadi, Arouri, and Teulon (2013) conducted a research on the relationship between investors’ attention, based on Google search volume, stock liquidity and volatility, using the CAC 40 index data. The research pointed out that more attention is given to larger sized firms, leading to more liquidity.
Although our research focus on the average returns of TAIEX, not on the stock liquidity or volatility, we decide to apply their research results and used the company size as the weight to compute the weighted sum of SVI to represent investors’ attention.
For the Taiwan stock market, Fan, Liao, and Chen (2014) showed that using the
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
10
top-50 company names as keywords for Google Search, the search volume could predict the average returns of TAIEX. In our research, we found that search volume based on company names of all ranks, not just the top-50, have the power to predict the average returns of TAIEX.
There are many studies that used company names or ticker symbols as keywords in Google search to represent investors’ attention to study different stock markets. Among them, Fan, Liao, and Chen (2014), Vlastakis and Markellos (2012), Bank, Larch, and Peter (2011), and Takeda and Wakao (2014) used company names as the search keywords. They reported that SVI is significantly correlated with the stock returns in their studied markets. By contrast, Da, Engelberg, and Gao (2011) used ticker symbols as the search keywords. They reported their SVI is also significantly correlated to the Russell 3000 index. For the Taiwan stock market, we find that SVIs based on company names of all ranks have the power to predict the average returns of TAIEX.
Additionally, SVIs based on the ticker symbols of companies that are ranked higher than 100 also have a similar predictive power.
2.2 Trading Volume and Market Sentimental Indexes
There are also works focusing on the impact of investors’ sentiments on the financial stock markets. For example, Chuang, Ouyang, and Lo (2010) used much
‧
theory to support the use of trading volume as an investors’ sentiment index and found that trading volume is a suitable proxy as sentiments. So, trading volume reflects investors’ expectation of the stock prices; hence influences the average returns of the stocks in the Taiwan stock market. Another work is by Chung and Yeh (2009), who did research on the U.S. stock market using many sentiment indexes, such as consumer confidence level1, the VXO2 (the old VIX (Volatility Index)), Baker and Wurgler's orthogonal sentiment index3 to capture consumers’ and investors’ sentiments. They reported that sentiments could be used to predict the stock returns in the U.S. stock market. Our research uses J.P. Morgan confidence indexes as investor’s interview sentiments and also found that two particular investors’ interview sentiments can help predict the average returns of TAIEX.