The central task in financial economics is to identify the systematic risks that drive asset prices and expected returns (Campbell, 2000; Cochrane, 2000). However, in recent years there has been a growing debate on the possible linkages between the behavioral aspects of investors and stock prices. Financial economics has become more receptive to imperfect rational explanations, and investor psychology has emerged as a major determinant of asset prices. After decades of study, the sources of risk premium in purely rational models are well understood; while, dynamic psychology based asset pricing theories are still in the infancy stage. This debate surrounding asset pricing has identified two prime suspects in setting stock prices: fundamentals and investor sentiments. The theoretical framework describing the role of investor sentiments in determining stock prices is provided by researchers such as Black (1986), Trueman (1988), DeLong, Shleifer, Summers and Waldman [DSSW henceforth] (1991, 1990), Shleifer and Summers (1990), Lakonishok, Shleifer and Vishny (1991), Campbell and Kyle (1993), Shefrin and Statman (1994), Palomino (1996), Barberis, Shleifer and Vishny (1998), Daniel, Hirshleifer and Subramanyam (1998) and Hong and Stein (1999). A direct implication of these studies is certain groups of investors (noise traders), who often do not make investment decisions based on a company’s fundamentals, are capable of affecting stock prices by way of unpredictable changes in their sentiments. Following the ‘noise trader model’ of DSSW (1990), several empirical studies examine the influence of investor sentiments on stock prices (Brown and Cliff, 2004a, 2004b; Lee et al. 2002; Fisher and Statman, 2000; Clarke and Statman, 1998; Solt and Statman 1988; De Bondt, 1993). Overall, these studies provide evidence in favor of strong co-movements between investor sentiment and the stock market returns recognizing the existence of individual investor sentiments, as well as institutional investor sentiments. The previous research mainly focuses on the effect of investor sentiments on the U.S. market while less attention has been given to its relevance in the international context. For example, little has been done to understand the degree of the relationship between the U.S. individual and institutional investor sentiments and Pacific-Basin stock returns. Given strong empirical evidence on the strengthening response of Pacific-Basin stock markets to the U.S. market over time (Soydemir, 2005; Kim, 2003; Ratanapakorn and Sharma, 2002; Janakiramanan and Lamba, 1998; Park and Fatemi, 1993) it is important to analyze whether the expectations of the U.S. investors is an important player in propagating U.S. stock market movements abroad. We shed new light on the relevance of the rational expectations and irrational exuberance of the U.S. investors in determining stock returns of Hong Kong, Malaysia, Philippines, Singapore, Taiwan, Thailand, and Korea. Using the investor sentiments data at the individual and institutional level, provided by American Association of Individual Investors and Investors Intelligence and the vector auto regression (VAR) models we find the following results: first, we do not find any significant effect of irrational exuberance of the U.S. investors on Pacific-Basin stock market returns. Second, we find significant relationship of varying degrees of strength between the rational expectations of the U.S. individual and institutional investors and Asian stock returns except in the cases of Taiwan and Thailand. Third, there are greater effects of rational expectations of the institutional investors than individual investors on theses stock returns. These findings suggest that the international effects of the U.S. stock market can be attributed to fundamental trading and not to noise trading in the U.S. market. These results are consistent with the view that investor sentiment in the U.S. is an important player in propagating U.S. stock market movements abroad. Therefore it is important for policymakers to consider such spillover effects in their international policy making decisions and for investors in their portfolio allocation decisions involving stock markets movements. This remainder of this paper is organized as follows: section two reviews the existing literature on investor sentiments and stock prices while section three presents the model. Section four presents the data. Section five reviews the econometric methodology. Section six presents empirical findings. Section seven concludes. 2. Previous work on investor sentiments and stock prices The concept of investor sentiments, noise trading and its role in the financial markets is first given by Black (1986). Black (1986) labels non rational investors as “noise traders”, who have no access to inside information, and act irrationally in response to news that conveys little information about fundamentals. However, there are two opposing views in the literature on the relevance of noise traders in determining stock prices. Based on Friedman (1953) and Fama (1965), it is argued that noise traders are irrelevant and cannot survive since they are driven out of the market by rational arbitrageurs. For example, West (1988) states “there is little direct evidence that trading by naïve investors plays a substantial role in stock price determination”. On the contrary, Black (1986) and Trueman (1988) argue that noise traders induce necessary liquidity in the market, and therefore provide incentives for informed investors to trade. The notable work of DSSW (1990, and 1991) models the influence of noise trading on equilibrium prices. They argue that noise traders falsely believe that they have special information about the future prices and the unpredictability of their sentiments brings an additional risk in the market. They may get pseudo signals from analysts, brokers, consultants and irrationally believe that these signals carry information. Their ‘noise trader’ model shows that a nonfundamental factor exists in the form of investor sentiments that is priced in equilibrium. Furthermore, noise traders as a group can earn expected returns higher than rational investors and can also survive in terms of wealth gain in the long run, due to unpredictability in their sentiments. Campbell and Kyle (1993) model the competitive interaction between noise and informed traders and its consequent effect on stock prices. Shleifer and Summers (1990) present an alternative to the efficient markets paradigm that stresses the role of investor sentiments and limited arbitrage in determining stock prices. They show that the assumption of limited arbitrage is more plausible as a description of risky asset markets than the assumption of complete arbitrage on which market efficiency hypothesis is based. This implies that changes in investor sentiments are not fully countered by arbitrageurs therefore may affect stock returns. Similarly, Shefrin and Statman (1994) show the interaction between noise and informed traders and present the behavioral capital asset pricing theory. They argue that in contrast to information, sentiments of noise traders’ act as a second driver which takes the market away from efficiency.

Lakonishok, Shleifer and Vishny (1991) find that in small markets institutional investors influence prices. Along the same lines, Palomino (1996) extends the DSSW (1990) model for an imperfectly competitive market and shows that noise traders may earn higher return and obtain higher expected utility than rational investors. Wang (2001) examines the dynamics of non-rational investors find that bullish sentiments can survive while bearish sentiment cannot survive in the long run. Overall, these models suggest that the unpredictability in investor sentiments of noise traders acting as a group can introduce a systematic risk that is priced in markets. Following these predictions, several empirical studies have examined the role of investor sentiments on stock pricing. These studies have either used indirect measures or direct measures of investor sentiments. Studies using indirect measures include the following proxies: close-ended fund’s discount (Gemmill and Thomas, 2002; Baker & Wurgler, 2005; Sias, Starks and Tinic, 2001; Neal and Whitney, 1998; Swaminathan, 1996; Elton, Gruber and Busse, 1998; Chan, Kan and Miller, 1993; Lee, Shleifer and Thaler, 1991); market performance based measures (Brown and Cliff, 2004a); trading activity based measures (Brown and Cliff, 2004a; Neal and Whitney, 1998); derivative variables (Brown and Cliff, 2004a); dividend premium (Baker and Wurgler, 2005); and IPOs related measures (Baker & Wurgler, 2005; Brown and Cliff, 2004a). Overall these studies do not provide a consensus on whether the proxies chosen are appropriate measures of investor sentiment and also show mixed results in their debate on the linkages between sentiments and stock returns. Studies using direct measures employ sentiment surveys data that indicate the expectations of market participants. Research related to individual investors sentiments find strong co-movements with stock market returns (Brown and Cliff, 2004a; De Bondt, 1993) and mixed results regarding its role in short term predictability of stock prices (Brown and Cliff, 2004a; Fisher and Statman, 2000). Similarly, studies examining institutional sentiments find strong co-movements with stock market returns (Brown and Cliff, 2004a) and mixed results regarding its short run implications on stock prices (Brown and Cliff, 2004a; Lee, Jiang and Indro, 2002; Clarke and Statman, 1998; Solt and Statman, 1988). Also, Brown and Cliff (2004b) examine the long run implications of institutional investor sentiments and find strong relationships with long horizon stock returns. Overall, these studies provide powerful and consistent empirical support for the hypothesis that stock prices are affected by individual and institutional investor sentiments. 3. Model Previous studies suggest that some shifts in investor sentiments are fully rational i.e., expectations based on the risk factors, while some are irrational exuberance, induced by the noise (Baker and Wurgler, 2005; Brown and Cliff,

2004b, Shleifer and Summers, 1990). Hirshleifer (2001) also relates expected returns to both risks and investor mis-valuation. When an investor is bullish or bearish, then this could be a rational reflection of future period’s expectation or irrational enthusiasm or a combination of both. Therefore it is quite possible that international stock returns are affected by both rational (risk based) and irrational (noise) components of the U.S. investor sentiments. We follow the approach of Baker and Wurgler (2005) to capture the irrational component of investor sentiments by regressing sentiment indicators to a set of risk factors and computing the residuals. Accordingly, we formulate equations (1) and (2): J

Sentt1t = γ 0 + γ j ∑ Fund jt + ξt

Sentt2t = θ 0 + θ j ∑ Fund jt + ϑt where γ0 and θ0 are constants, γj and θj are the parameters to be estimated; ∂ ζt and t are the random error terms. Sentt1t and Sentt2t represent the shifts in sentiments of individual and institutional investors respectively at time t. Fundjt is the set of fundamentals representing rational expectations based on risk factors that have been shown to carry non-redundant information in conditional asset pricing literature. The fitted values of equations (1) and (2) capture the rational component of sentiments (i.e. Senttˆ1t and Senttˆ2t). On the other hand the residual of equations (1) and (2) capture the irrational component of sentiments (i.e. ζt ∂ and t ). Next, we analyze the extent to which Pacific-Basin stock returns are affected by rational expectations and irrational exuberance of the U.S. individual and institutional investors. Accordingly, the sentiment variables are decomposed into the rational and irrational components based on equations (1) and (2) and included in the return generating process as:

Rit = α 0 + α1Senttˆ1t + α 2 Senttˆ2t + α 3ξ t + α 4 where α0 is a constant while α1, α2, α3 and α4 are the parameters to be estimated; ρt is the random error term. Rit is the returns for the ith PacificBasin stock market at time t. Specifically, the parameters α1 and α2 capture the effects of rational expectations on part of individual and institutional investors respectively; while α3 and α4 capture the effects of irrational exuberance of individual and institutional investors respectively. We also place importance on jointly modeling the sentiments of individual and institutional investors to avoid misspecification. Specifically, shocks originating from sentiments of one class of investors not considered might mistakenly be seen as a disturbance originating from sentiments of another class of investors in the analysis.

We obtain all data in monthly intervals from October 1988 to April 2004. To measure sentiments of market participants, we employ survey data similar to the ones used in the literature. The institutional investors participate in the market for living while the individual investors’ primary line of business is outside the stock market (Brown and Cliff, 2004a). Our choice of individual investor sentiment index is based on Brown and Cliff (2004a), Fisher and Statman (2000) and DeBondt (1993) which use the survey data of American Association of Individual Investor (AAII). Beginning July 1987, AAII conducts a weekly survey asking for the likely direction of the stock market during the next six months (up, down or the same). The participants are randomly chosen from approximately 100,000 AAII members. Each week, AAII compiles the results based on survey answers and labels them as bullish, bearish or neutral. These results are published as ‘investor sentiment’ in monthly editions of AAII Journal. The sentiment index for individual investors is computed as the spread between the percentage of bullish investors and percentage of bearish investors (BullBear). Since this survey is targeted towards individual investors, it is primarily a measure of individual investor sentiments. Our choice of institutional investor sentiment index is based on Brown and Cliff (2004a, 2004b), Lee et al. (2002), Clarke and Statman (1998) and Solt and Statman (1988) which use the survey data of Investors Intelligence (II), an investment service based in Larchmont, New York. II compiles and publishes data based on a survey of investment advisory newsletters. To overcome the potential bias problem towards buy recommendation, letters from brokerage houses are excluded. Based on the future market movements the letters are labeled as bullish, bearish or correction (hold). The sentiment index for the institutional investor is found by calculating the spread between the percentage of bullish investors and percentage of bearish investors. Because authors of these newsletters are market professionals, the II series is interpreted as a proxy for institutional investor sentiments. We analyze the response of seven Pacific-Basin stock markets, Korea, Malaysia, Philippines, Singapore, Taiwan, Thailand and Korea. The market variable identified for these countries are the major indexes in their respective stock markets. The continuously compounded returns are computed from the local currency denominated stock market indexes obtained from the Datastream. We include the following variables as risk factors that have been shown to carry non-redundant information in the asset pricing literature: (i) Economic growth (Fama, 1970; Schwert, 1990) measured as the monthly changes in the industrial production index (ii) Short term interest rates (Campbell, 1991) measured as the yield on one month U.S. Treasury Bill (iii)Economic risk premia (Ferson and Harvey, 1991; Campbell, 1987) measured as the term structure of interest rates (difference in monthly yields on three month and one month Treasury bills (iv) Future economic expectations variables (Fama, 1990) measured as the term spread (yields spread on the 10 year U.S. Treasury bond and three month Treasury bill) (v) Business conditions (Fama and French, 1989; Keim and Stambaugh, 1986) measured as the default spread (difference in yields on Baa and Aaa corporate bonds)(vi) Dividend yield (Hodrick, 1992; Fama and French, 1988; Campbell and Shiller, 1988a, 1988b) measured as the dividend yield for the value weighted Center for Research in Security Prices (CRSP) index over the past 12 months (vii) Inflation (Sharpe, 2002; Fama and Schwert, 1977) measured as the monthly changes in the consumer price index (viii) Excess returns on market portfolio(Lintner, 1965; Sharpe, 1964) measured as the value-weighted returns on all NYSE, AMEX, and NASDAQ stocks minus the one-month Treasury bill rate (ix) Premium on portfolio of small stocks relative to large stocks (SMB) (Fama and French, 1993). SMB (Small minus Big) is the average return on three small portfolios minus the average return on three big portfolios (x) Premium on portfolio of high book/market stocks relative to low book/market stocks (HML) (Fama and French, 1993). This Fama/French benchmark factor is constructed from six size/book-to-market benchmark portfolios that do not include hold ranges and do not incur transaction costs. HML (High minus Low) is the average return on two value portfolios minus the average return on two growth portfolios. (xi) Momentum factor (UMD) (Jegadeesh and Titman, 1993). UMD (Up minus Down) is the average return on the two high prior return portfolios minus the average return on the two low prior return portfolios (xii) Currency fluctuation (Elton and Gruber, 1991) measured as the changes in 15-country trade weighted basket of currencies. The data on economic growth, business conditions and inflation are obtained from Datastream; short term interest rates, economic risk premium, future economic variables and currency fluctuations are obtained from Federal Reserve Bank of St. Louis; dividend yield and excess return on market portfolio from CRSP; and SMB, HML and UMD from Kenneth French Data Library at Tuck School of Business, Dartmouth College. Table 1 reports the descriptive statistics of the above mentioned variables. The mean of Sntt1 and Sntt2 are approximately 11% and 9% respectively. This suggests both individual and institutional investors have been bullish during most of the sample period. Interestingly, individual investors have been more bullish than institutional investors. The standard deviations of Pacific-Basin stock markets are very high indicating their extremely volatile nature during the sample period. Among these markets Hong Kong seems to have provided the highest return to investors, while the mean returns in case of Thailand and Korea are negative. Most of the variables relating to the risk factors have shown less variability as compared to the investor sentiments and Asian stock returns.

The variables are individual investor sentiments (Sentt1), institutional investor sentiments (Sentt2), economic growth(IIP), short term interest rates (T30), economic risk premiums (T90-T30), future economic variables (B10-T30), business conditions (Baa-Aaa), dividend yield (Div.), inflation (INF), excess returns on market portfolio (Rm), premium on portfolio of small stocks relative to large stocks (SMB), premium on portfolio of high book/market stocks relative to low book/market stocks (HML), momentum factors (UMD), currency fluctuations (USD), and stock market returns on Hong Kong (Hong Kong), Malaysia (Malaysia), Philippines (Philippines), Singapore (Singapore), Taiwan (Taiwan), Thailand (Thailand), and Korea (Korea). Mean

Median Maximum Minimum Std. Dev. Skewness Kurtosis

Sentt1

Sentt2

IIP

Baa-Aaa

Div

INF

Rm

SMB

HML UMD

USD

Hong Kong 0.0120 Malaysia 0.0005 Philippines 0.0060 Singapore 0.0048 Taiwan 0.0014 Thailand -0.0057

Korea

We choose the VAR modeling technique (Sims, 1980) as an appropriate econometric methodology to investigate the postulated relationships. The rationale for doing so lies in the arguments of Brown and Cliff (2004a & 2004b) and Lee et al. (2002) which suggest that stock market returns and investor sentiments may act as a system. Our approach is also consistent with studies such as Soydemir (2005), Kim (2003), Ratanapakorn and Sharma (2002), Janakiramanan and Lamba (1998), and Park and Fatemi (1993) which have employed the VAR models to analyze the linkages between the U.S. and PacificBasin stock markets.

The VAR specification allows the researchers to do policy simulations and integrate Monte Carlo methods to obtain confidence bands around the point estimates (Doan, 1988; Genberg et al. 1987; Hamilton, 1994). The likely response of one variable at time t, t+1, t+2 etc. to a one time unitary shock in another variable at time t can be captured by impulse response functions. As such they represent the behavior of the series in response to pure shocks while keeping the effect of other variables constant. Since, impulse responses are highly non-linear functions of the estimated parameters, confidence bands are constructed around the mean response. Responses are considered statistically significant at the 95% confidence level when the upper and lower bands carry the same sign. Thus VAR model captures the dynamic feedback effects in a relatively unconstrained fashion and is therefore a good approximation to the true data generating process. We express the VAR model as: m where, Z(t) is a column vector of variables under consideration, C is the deterministic component comprised of a constant, A(s) is a matrix of coefficients, m is the lag length and ε(t) is a vector of random error terms.

The variables are economic growth (IIP), short term interest rates (T30), economic risk premiums (T90), future economic variables (B10), business conditions (Baa), dividend yield (Div), inflation (INF), excess returns on market portfolio (Rm), premium on portfolio of small stocks relative to large stocks (SMB), premium on portfolio of high book/market stocks relative to low book/market stocks (HML), momentum factors (UMD), and currency fluctuations (USD). B10

Baa

IIP

Baa 0.00 1.00 IIP -0.15 -0.39 HML 0.06 -0.06

INF

Rm

DIV

Rm

DIV SMB T30 T90 UMD USD

SMB -0.20 -0.04 -0.05 -0.50 0.00

UMD 0.20

In this study, we investigate whether the rational expectations and irrational exuberance of the U.S. individual and institutional investors are propagated to Pacific-Basin stock markets, Hong Kong, Malaysia, Philippines, Singapore, Taiwan, Thailand, and Korea. We employ the investor sentiments data at the individual and institutional level, provided by American Association of Individual Investors and Investors Intelligence and the vector auto regression (VAR) models to investigate the postulated relationships. The results of variance decomposition and impulse response functions strongly suggest that there are distinct effects of the rational and irrational investor sentiments on these international stock market returns.

We find strong evidence of insignificant effect of irrational exuberance of the U.S. investors on these set of Pacific-Basin stock market returns. However, the rational expectations of both the U.S. individual and institutional investors have significant effect on these Asian markets with varying degrees of intensity. We find greater effects in the case of Hong Kong, Malaysia, Philippines, and Singapore while a weak linkage with the stock returns of Taiwan, Thailand, and Korea. The overall evidence suggests that institutional investor sentiments are transmitted internationally from the U.S. stock market to a greater extent than the individual investor sentiments; and that the international effects of the U.S. stock market can be attributed to the sentiments induced by fundamental trading and not to noise trading. These results are consistent with the view that besides economic fundamentals, the international effects of the U.S. stock market can be attributed to rational investor sentiments. Therefore it is important for policymakers to consider such spillover effects in their international policy corporation and for investors in their portfolio allocation decisions involving stock markets movements. Author statement: Rahul Verma, College of Business, University of HoustonDowntown. E-mail: vermar@uhd.edu. I like to record my thanks to the editors for receiving comments on the paper and making changes to the paper. The mistakes are my own responsibility.