For decades, the success and popularity of the efficient market hypothesis (EMH) lies in its ability to explain the lack of predictability in liquid asset returns, meanwhile traditional
“search for value” is clashed by many finance practitioners. More recent analysis has discussed how such traders acting sentiment might induce systematic risk and affect asset prices in equilibrium. For example, as the noise trader models of De Long, Shleifer, Summers, and Waldmann (1990a; 1990b) suggest that if informed arbitrageurs know that prices may diverge further away from fundamentals before they converge closer, they may take smaller positions when betting against mispricing. If these uninformed noise traders base their trading decisions on sentiment, sentiment may have predictive power for asset price behavior. The noise trader model of De Long, Shleifer, Summers, and Waldmann (1990a) has motivated empirical attempts to substantiate the proposition that noise trader risks influence price formation.
Most papers test whether sentiment can predict returns or volatility. They attempt to explain this correlative relationship through the role of noise traders whose changes in sentiment can influence subsequent returns and volatility. If it is true, we can use sentiment as an indicator to forecast not only the returns but volatility as well. Many papers in the past have used sentiment to forecast return or volatility, but rarely both at the same time.
The model of De Long, Shleifer, Summers, and Waldmann (1990a) predicts that the direction and magnitude if changes in noise trader sentiment are relevant in asset pricing, the subsequent empirical testing focused on the impact of sentiment either on the mean or variance in asset returns alone, such testing are mispecified and at best can only be considered as incomplete. The “price-pressure” and “hold-more” effects capture the impact of noise trading on excess returns resulting from lagged changes in investor sentiment. The
“Friedman” and “create-space” effects reflect the impact of noise trading on excess returns associated with the influence of the magnitude of sentiment changes on the future volatility of returns.
The “hold-more” effect implies that noise traders’ increased holdings of risky assets when their sentiment becomes more bullish raises market risk and thereby increases expected returns; and vice versa, when they are bearish. However, noise traders overreact to good and bad news. Asset prices are either too high or too low depending on where noise traders are on average optimistic or pessimistic. Such overreaction lowers expected returns. This
“price-pressure” effects and market returns will correlate with changes in investor sentiment and the direction of the correlation depends on which effect dominates.
In addition, the magnitude of the changes in perceptions about the asset’s risk by noise traders associated with their shifts in sentiment also impact expected returns. Noise traders usually have poor market timing because of their tendency to trade together with other noise traders. Their capital losses are larger due to poor market timing and the magnitude of losses regards the magnitude of the change in their misperceptions. The Friedman effect implies that this changes result in higher market risk and lower expected returns. There is an adverse impact that the Friedman effect has on expected returns depending on the “space” the noise trading creates. A rise in noise traders’ misperceptions increases price uncertainty and crowds out risk-averse informed investors. Therefore, the greater is the proportion of noise trading, the higher will be expected returns.
Lee, Jiang, and Indro (2002) employ a generalized autoregressive conditional heteroscedasticity (GARCH) in-mean model (Engle, 1982; Bollerslev, 1986; Engle, Lilien, and Robins, 1987) to show that both the conditional volatility and excess returns are affected by investor sentiment. In this paper, we also use a GJR-GARCH in-mean (Glosten, Jagannathan, and Thaler, 1993) to show such a relationship. It is different with Lee et al.’s
model which includes contemporaneous shifts in investor sentiment within the mean equation, while our model includes lagged shifts in investor sentiment in the mean equation.
We examine the relationship between volatility of market excess returns, excess returns, and investor sentiment for three different market indices, the DOW Jones Industrial Average (DJIA), the Standard and Poor’s 500 (S&P500) and the NASDAQ. In this paper, we use a lot of sentiment indices to proxy the noise traders’ sentiment. For the daily data, we use the OEX put-call trading volume ratio (PCV), the OEX put-call open interest ratio (PCO), and the ARMS index for NYSE as the sentiment indices. For weekly data, we use the bullish percentage of sentiment indices of Investors’ Intelligence (II) and the American Association of Individual Investors (AAII). And for monthly data, we use the initial public offering first day returns (IPORET) and the number of offerings (IPON).
Our main findings suggest that sentiment is a significant factor in explaining equity excess returns and conditional volatility of the excess return. In addition, we find that the magnitude of shifts in sentiment has a significant impact on the formation of conditional volatility of excess returns and excess returns. Bullish (bearish) shifts in sentiment lead to downward (upward) revisions in the volatility of returns.
Furthermore, we find that PCO, AAII, II, and IPON can be used to forecast the future returns and ARMS, PCO, PCV, AAII, and IPON are good proxy to forecast the volatility of returns. Some of the indices are useful in forecasting the one of the large and small capitalization stocks.
Since we find the sentiment of noise trader has an impact on the conditional volatility of the excess return. And few researches discuss that noise trader risk whether is only a transitory phenomenon. We try to use the component GARCH (Engle and Lee, 1999) to divide the noise trader risk into two components which are the transitory component and the
permanent component. We find that effect of sentiment in the transitory component is larger and more significant than in the permanent component. That is to say noise trader risk should be a transitory phenomenon in the conditional volatility, and the stock market will recover in the future (long-term).
The remainder of the paper proceeds as follows. Section 2 of the paper discusses the noise trader risk and the relationship among the sentiment, return, and volatility and introduces the literatures of GARCH models. Section 3 of the paper presents the data source. Section 4 presents the empirical model. Section 5 discusses the empirical results. The last section provides some conclusions.