The Impact of Investor Sentiment

3.1.1 Proxy Variables of the Investor Sentiment

Prior studies classify investor sentiment index, which has been used in studies both domestic and abroad into two categories: the first one is direct index, which is acquired directly by surveying investors; the other one is indirect index, which is calculated by relevant data that can reflect investors’ psychology.

Regarding the indirect sentiment index, empiricists proxy for investor sentiment with market-based measures and weather conditions. Market-based measures such as Volatility Index (VIX), turnover rate, trading volume, IPO first-day returns, IPO volume, or closed-end fund discount, among others (Baker & Stein 2004; Baker & Wurgler, 2006; 2007; Chou, Chang, & Lin, 2007; and Da et al., 2015). Weather conditions such as temperatures, air pressure, humidity, cloud cover and sunshine duration (Hirshleifer & Shumway, 2003).

Hirshleifer & Shumway (2003) point out the famous “Sunshine effect”. It is worth noting that the Volatility Index (hereafter VIX) is calculated and disseminated in real-time by the Chicago Board Options Exchange (CBOE). This study uses CBOE VIX instead of Hong Kong VIX because Hong Kong VIX is high related with CBOE VIX and the CBOE VIX is considered by many to be the world's premier barometer of investor sentiment and market volatility and is designed to reflect investors' consensus view of future (30-day) expected stock market volatility. The VIX is often referred to as the market's "fear gauge. Da et al.

(2015) includes the VIX index as a control variable in most specifications. Because market-based measures have the advantage of being readily available at a relatively high frequency (Da et al, 2015), easy to obtain, neutrality and more complete, there are abundant studies use direct sentiment index proxy for investor sentiment.

Regarding the direct sentiment index, empiricists use survey-based indices such as the University of Michigan Consumer Sentiment Index or investment newsletters (Brown & Cliff, 2005; Lemmon & Portniaguina, 2006; and Qiu & Welch, 2006). Compared with market-based measures of investor sentiment, the survey-based sentiment measure has two advantages of greater influence and prediction on the market. For example, Michigan Consumer Sentiment

Index is widely accepted in the world. Baker & Stein (2004), Baker & Wurgler (2006), and Kaustia & Knupfer (2008) showed the significant prediction ability of investors' sentiments for stock returns or investors' trading behavior. Da et al. (2015) points out search-based sentiment measures are available at a high frequency. Survey measures are often available monthly or quarterly, and they find the daily FEARS index can predict monthly survey results of consumer confidence and investor sentiment.

To sum up, market-based measures and weather conditions are the indirect sentiment index proxy for investor sentiment. Indirect sentiment index has the disadvantage of being the equilibrium outcome of many economic forces other than investor sentiment (Qiu & Welch, 2006). And survey-based index is the direct sentiment index proxy for investor sentiment.

Although survey-based indices have two advantages of greater influence and prediction on the market, but it is too decisive and difficult to use specific proxy variable to stand for psychology states of investor. Because indirect sentiment index is readily available at a relatively high frequency, easy to obtain, neutrality and more complete, especially VIX is now in widespread use and turnover rate is generally recognized proxy for investor sentiment, and there is a debate about the relation between idiosyncratic risk and expected return in the literature, this study consider VIX proxy for investor sentiment investigate the relation between idiosyncratic risk and expected returns. Furthermore, this study considers turnover rate proxy for investor sentiment for robustness.

3.1.2 Importance of distinguishing investor sentiment by VIX on idiosyncratic risk

There are three main topics of VIX researches: analyst recommendation revision (Kliger

& Kudryavtsev, 2013), mutual fund flows (Ben-Rephael, Kandel, & Wohl, 2012), and stock return (Brown & Cliff, 2004 and Cotter et al., 2015), but quite few researches to investigate the relation between idiosyncratic risk and expected return under high and low VIX³.

Investors’ behaviors are affected by investors sentiments (e.g. Brown & Cliff, 2004 and Lee, Yen & Chan, 2014), which will make differences of the relation between idiosyncratic risk and expected returns. Expand on the affection of investor sentiments: prior studies point out that investor sentiments will affect investors’ behaviors (Brown et al., 2013; Da et al., 2015; Fisher & Statman, 2000; Chan & Fong, 2004; Baker & Stein, 2004; Baker & Wurgler,

3 Searching keywords of “VIX”, “investor sentiment”, and ”idiosyncratic risk” in SSCI database, respectively.

10

2006; and Kaustia & Knupfer, 2008) and the stock price (Brown & Cliff, 2004 and Baker &

Wurgler, 2006), which will further differentiate the relation between idiosyncratic risk and expected return.

Investor sentiment represents the expectations of market participants relative to a norm:

A bullish (bearish) investor expects returns to be above (below) average, whatever average may be (Brown & Cliff, 2004). For example, stock prices are rose and fell by the effect of investor sentiment (Baker & Wurgler, 2006; Brown & Cliff 2004; Chung et al., 2012; and Lee et al., 2014). VIX is calculated and disseminated in real-time by the CBOE and is considered by many to be the world's premier barometer of investor sentiment and market volatility and is designed to reflect investors' consensus view of future (30-day) expected stock market volatility. VIX is often referred to as the market's "fear gauge. High VIX means investor sentiment is panic and fear, which usually associated with crisis, crash, asset bubble; on the other hand, low VIX means investor sentiment is non-panic, unfear (D’Anne, 2012 and Treadway, 2013). Because there is a debate about the relation between idiosyncratic risk and expected return in the literature, this study considers the investor sentiment to investigate the relation between idiosyncratic risk and expected returns.

There are two threshold values for distinguishing panic and non-panic. The first one is pointed out by D’Anne (2012), who suggests a helpful categorization of market anxiety based on VIX values: It is extremely high anxiety when VIX above 40; and it is low anxiety or moderate complacency when VIX under 15, as below:

 15-20 low anxiety, moderate complacency

 20-25 moderate anxiety, low complacency

 25-30 moderately, high anxiety

 30-35 high anxiety

 35-40 very high anxiety

 40-45 extremely high anxiety

The second one is pointed out by Treadway (2013). In general, VIX equal to 30 appears to be the first ceiling of the panic index in normal condition. VIX above 30 usually linked with crises, market crashes, panics and asset bubble bursts. Therefore, panic means VIX above 30.

To sum up, this study distinguishes investor sentiment states and periods from VIX into four states, include VIX≧30, VIX<30, VIX≧40, and VIX<15 to investigate the relation

between idiosyncratic risk and expected returns. This study distinguishes VIX into VIX≧30 and VIX<30 to calculate the empirical results, and then extremely distinguishes VIX into VIX

≧40, and VIX<15 for robustness.

Investor sentiment is a short-run effect. Investor sentiment only affects temporary trade behaviors (Chan & Fong, 2004), and it is a short-run phenomenon (Tsai, Wang, & Chung, 2009). Investor sentiment only influences trade decision in short-run, but lower the influence in long-run by getting more information by inputting the time and labor (Chen, Chou, &

Wang, 2008).

The investors are more willing to hedge in a slump, and arbitrageurs are risk averse in the short term so the risk-averse arbitrageurs prefer holding the portfolios on low idiosyncratic risk stocks, and it makes those low idiosyncratic risk stocks to have high trading volume and further have high expected return in the short term (Shleifer & Vishny, 1997 and Lee & Wei, 2012).