Effect of Sentiment on the Bangladesh Stock Market Returns

Transcription

1 Effect of Sentiment on the Bangladesh Stock Market Returns Shah Saeed Hassan Chowdhury * Assistant Professor, Department of Accounting and Finance, Prince Mohammad University schowdhury@pmu.edu.sa Rashida Sharmin Instructor, Department of Accounting and Finance, Prince Mohammad University rsharmin@pmu.edu.sa M. Arifur Rahman Senior Lecturer, University of Brunei Darussalam arifur.rahman@ubd.edu.bn JEL Classification: G12; G14 Keywords: Market sentiment; Bangladesh stock market; Emerging stock markets; Return predictability Abstract Research in the past two decades has shown that traditional financial asset pricing models cannot explain stock returns satisfactorily. Thus other factors that possibly capture the behavioral aspects of the market might help explain the returns of stocks. The recent research on developed markets has shown significant relationship between stock returns and the sentiment of the market. As emerging markets are dominated by individual investors and lack availability of quality information and professional financial analysts services, it is not a surprise that the performances of these markets are subject to sentiment of general investors. Literature survey on Bangladesh stock market reveals that academicians in the past have used many factors to explain returns, but the relationship between factors and return is found to be weak. Even, stock beta fails to explain the cross-section of expected returns of the Dhaka Stock Exchange. Under such backdrop, this paper looks at Bangladesh stock market completely from behavioral perspective by introducing behavioral factors in the empirical asset pricing models. Lack of data of behavioral proxies constrains authors to use only few factors, namely, TRIN index, trade volume, number of IPOs per month and change in four-month moving average. Since these variables are not available from data vendors, they are created from raw data. Due to low correlation among sentiment measures, a unified sentiment index is not necessary. Since macroeconomic factors are sources of systematic risk, residual returns series are created from regressing market returns on macroeconomic factors. Monthly macroeconomic and stock market data are collected from International Financial Statistics and Thomson Reuters Datastream, respectively. We have found that TRIN and moving average significantly affect the residual market returns. This paper checks the impact of sentiment factors by considering firm size. Results show that the impact of TRIN and trade volume is strong for large and medium size firms. However, the effect of TRIN is either low or insignificant for small size firms, indicating less interest of investors for neglected stocks. Results also show that high sentiment leads to high return followed by negative correction in the next period. Overall, we can conclude that sentiment plays a big role in determining Bangladesh stock market performance. * Dr. Shah Saeed Hassan Chowdhury is the corresponding author. He can also be reached at hchowdh@yahoo.com. 1

2 1. Introduction Investor sentiment may be defined as the propensity of investors to speculate. This attitude is related to the psychological state of mind of investors. In positive states, investors may follow their routine investment strategies whereas in negative states they take investment decisions with more consideration and analyses (Schwarz, 2002). Investors become less attentive in good time, which leads to mispricing of stocks (Dellavigna and Pollet, 2009). Brown and Cliff (2004) describe sentiment as the expectations of market participants to a norm. The average investor is a zero sentiment person whereas a bullish (bearish) investor expects higher (lower) returns than an average investor does. High sentiment can push stock prices away from their fundamental values. In 1996, Federal Reserve system Chairman Alan Greenspan used the term irrational exuberance to warn about the bubble in technology stock prices in the 1990s, although his speech could not stop the market from further rise of stock prices, resulting in bubble burst in Shiller (2000) also blames the presence of sentiment as irrational exuberance which drove the prices of U.S. stocks well above their fundamental value. Thus, market sentiment could be a source of systematic risk, which should be priced. In traditional finance theory, stock prices are equal to the present value of all expected future cash flows. It also suggests that cross-section of expected returns is supposed to be explained by only cross-section of systematic risks and any mispricing should be driven away by the activities of arbitrageurs. Therefore, there is no role of investor sentiment on stock returns. However, there is long debate on whether or not sentiment influences stock prices. Black (1986) explains how noise traders, being influenced by sentiment, behave irrationally in the market. De Long et al. (1990) provide evidence that noise trading influences equilibrium stock prices. In their model, arbitrageurs cannot drive away such disequilibrium and consequently sentiment is priced. Lee et al. (2002) also find that sentiment is a systematic risk that is priced. Swaminathan (1996) and Lee et al. (1991) find predictive power of individual investor sentiment on excess expected returns of small firms. Brown and Cliff (1999) find weak relationship between sentiment and stock returns although there is strong relationship between sentiment and small stock returns. Elton et al. (1998) provide the evidence that investor sentiment does not influence returns of closed-end funds. The recent research on developed markets has shown significant relationship between stock returns and the sentiment of the market (Barberis et al., 1998; Brown and Cliff, 2005; 2

3 Kumar and Lee, 2006; Baker and Wurgler, 2006). Beaumont et al. (2008) report that sentiment has significant and asymmetric effect on stock return volatility. Finter et al. (2010) show that market sentiment index explains the spread between the returns of sentiment stocks and sentiment insensitive stocks. Moreover, they find that stocks that are difficult to arbitrage and value are more sensitive to market sentiment. Baker et al. (2011) consider both local and global sentiment factors to investigate presence of sentiment on six major markets. They report that high sentiment causes low future returns for relatively difficult to arbitrage and difficult to value stocks. Antoniou et al. (2012) show that higher momentum of returns occurs in higher sentiment periods and momentum is absent during pessimistic periods. Literature survey on Bangladesh stock market reveals that academicians in the past have used many factors to explain returns, but the relationship between systematic risk factors and return is found to be weak (Chowdhury and Sharmin, 2013). Even, stock beta fails to explain the cross-section of expected returns of the Dhaka Stock Exchange (DSE), the largest stock exchange in Bangladesh. Some studies also show negative relationship between beta and returns (Basher et al., 2007). Baker and Wurgler (2006) argue that in case of young firms, extreme growth firms, small firms, and non-dividend-paying firms there is higher possibility of subjectivity of valuations, which leads to susceptibility to market sentiment. Thus, if the market itself is young, there is high probability of presence of sentiment. As emerging markets are dominated by individual investors and lack availability of quality information and professional financial analysts services, it is not a surprise that the performances of these markets are subject to sentiment of general investors. Above discussion has also shown that even in the developed markets there is strong possibility of effect of sentiment when stocks are new and difficult to arbitrage. Thus, this paper looks at Bangladesh stock market completely from behavioral perspective by introducing behavioral factors in the empirical asset pricing models. 2. Sentiment in Emerging Markets Aitken (1998) finds that emerging stock markets experienced a sharp increase in autocorrelation in total returns at a time when institutional investors began to invest heavily in these markets. He suggests that sentiment of institutional investor toward emerging markets probably plays an important role in determining asset prices, resulting in asset price overshooting. Doukas and 3

4 Milonas (2004) report that there is no impact of investor sentiment on stock returns in Greek stock market. Their finding is valid across different size portfolios and mutual funds. Thus, investor sentiment is not a source systematic risk. Kling and Gao (2008) show that sentiment influences stock returns in the Chinese stock market. More precisely, although the mood of investors follows a positive feedback in the shortrun, stock price and investor sentiment do not have a long-run relationship. Institutional investors are optimistic (pessimistic) when previous market returns are positive (negative). Moreover, pessimistic sentiment affects negative returns more than optimistic sentiment does on positive returns. Canbas and Kandir (2009) use several indirect sentiment measures to examine their effect on future stock returns for the stocks listed in the ISE (Istanbul Stock Exchange) and find that only turnover ratio, a proxy for sentiment, has the potential to predict stock returns. Zouaoui et al. (2010) use panel data of 15 European plus U.S. stock markets and find that sentiment has more effect on stock returns in those countries where investors are prone to herd behavior and institutional investment is relatively low. Grigaliuniene and Cibulskiene (2010) examine the effect of sentiment in Scandinavian markets and report negative relationship between sentiment and future stock returns. Liu et al. (2011) consider the direct and indirect effects of sentiment factors on Taiwan stock market returns. Results indicate that extreme sentiment indicator plays a critical role in determining changes in market returns. Anusakumar et al. (2012) report that local as well as global sentiment influences stock returns of a sample of 13 Asian countries. Their findings indicate the presence of sentiment in the market and their result is robust to firm size, trading volume, sample period and alternative proxies. Zhuang and Song (2012) examine the relationship between investor sentiment and stock returns and volatility in the Chinese stock market. They report that investor sentiment has a strong impact on stock returns in the Chinese stock market. Rehman (2013) gives empirical evidence that stock returns of Karachi Stock Exchange, a market which primarily consists of retail investors, are also influenced by investor sentiments. Liston et al. (2012) investigate Dow Jones Islamic Equity indices to find that bullish shifts in sentiment in current period lower conditional future volatility. 4

5 3. Data and Methodology There are two types of sentiment measures: direct and indirect. Investor surveys provide the direct measure of sentiment of the market. In the literature, several individual sentiment proxies are used. The most important or popular proxies for indirect sentiment measures are: aggregate net flows of equity mutual funds, put-call ratio, aggregate trading volume, IPO returns, number of IPOs, short sales to total sales ratio, close-end fund discounts, bull-bear spread, and TRIN (TRading INdex, also known as Arms Index) index. There is no survey-based sentiment measure for Bangladesh stock market. Lack of data of behavioral proxies constrains us to use only few factors, namely, TRIN index, trade volume, number of IPOs per month, number of BO (Beneficiary Owner) account changes and moving average. Since these variables are not readily available from data vendors, they are created from raw data. Usually, sentiment proxies contain common information. Consequently, researchers employ tools such as Kalman filter or principal component analysis in order to construct a single sentiment measure. Due to low correlation among sentiment measures for Bangladesh market, a unified sentiment index is not necessary. Monthly macroeconomic and stock market data are collected from International Financial Statistics and Thomson Reuters Datastream, respectively. BO account and IPO data are collected from Dhaka Stock Exchange, the largest exchange in Bangladesh. Since macroeconomic factors are sources of systematic risk, residual returns series are created by regressing market returns on macroeconomic factors. The regression model is R t = α 0 + α 1 IP t + α 2 IP t 1 + α 3 INF t + α 4 INF t 1 + α 5 INT t + α 6 INT t 1 + ε t, (1) where IP, INF, and INT indicate industrial production, Inflation and interest rates, respectively. Thus, error term represents the return, which is not explained by macroeconomic factors. We then regress ε on sentiment factors. The regression model to investigate the effect of sentiment on return is R ε,t = β 0 + β 1 TRIN t + β 2 TRIN t 1 + β 3 VOL t + β 4 VOL t 1 + β 5 IPO t + β 6 IPO t 1 + β 7 MA t + β 8 MA t 1 + β 9 OPCL t + β 10 OPCL t 1 + ε t, (2) where TRIN is the TRIN (Arms) index, VOL is the aggregate number of shares traded divided by total number of shares outstanding, IPO is the number of IPOs in month t, MA is the number of stocks having positive change in four-month moving average divided by number of firms having negative change in their four-month moving average at time t, OPCL is the number of BO accounts opening divided by number of BO account closing at time t. 5

6 To estimate TRIN index we use the following formula: TRIN t = DECVOL t #DECt ADVVOL t #ADVt (3) Bullish period indicates that average advance volume is greater than the average decline volume, that is, TRIN is less than 1. Likewise, if TRIN is larger than 1, it indicates bearish period. Thus, traditional TRIN Index was originally defined in such a way that it is counter intuitive. To solve this issue we invert each TRIN value and then take its logarithm. After this transformation, positive change in TRIN indicates higher sentiment. 4. Analyses of Results Table 1 exhibits the effect of sentiment on the market return. Results show that TRIN and MA (moving average) significantly affect the residual market returns. When only MA is used, adjusted R 2 is only 0.09, but when both MA and TRIN are used, adjusted R 2 jumps to In all the models where TRIN is used, it is significant at 1%, which shows its strong impact on stock returns. VOL (turnover) weakly explains market returns as its coefficients are significant at 10% level. Coefficients of TRIN and VOL are significantly positive concurrently whereas these are negative in the lagged (next) month. Table 1 about here The effects of TRIN and VOL indicate the possible dynamics of the behavior of stock returns with respect to market sentiment. Initially, high sentiment leads to higher stock prices and higher return. In the next period, investors realize the overreaction of the past period, resulting in negative correction of stock prices. For MA, contemporaneous and lagged coefficients are significantly negative and positive, respectively. The explanation could be that this sentiment measure is constructed based on past performance, which has an inbuilt lagged market sentiment effect. That is, the relationship between TRIN, VOL and MA is well-aligned and there exists no contradiction among them. Interestingly, IPO (number of monthly IPOs) and OPCL (ratio of BO account opening to BO account closing) do not explain market returns. Although not reported 6

7 here due to brevity, same regressions model using residual equally-weighted market returns as the dependent variable provide similar finding. Table 2, 3, and 4 examine the impact of sentiment factors by considering firm size because previous research shows relationship between sentiment and stock returns is particularly strong for small firms. The reason is the fact that small stocks largely represent hard to value and hard to arbitrage stocks. Table 2 provides the findings for large firms. Model 2 uses all variables and the adjusted R 2 is Effect of TRIN is strong as before. However, despite TRIN and MA, VOL becomes significant at 1% level for almost all models. Model 4, which uses all these three sentiment measures, has found all of them significant and the adjusted R 2 is The effects of TRIN, VOL and MA indicate the dynamics of the behavior of stock returns with respect to market sentiment almost the same manner that we have found in Table 1. Only difference is that the effects are more pronounced for large firms. Also as before, initially higher sentiment leads to higher stock prices and higher return, which is followed by subsequent corrections. Table 2 about here Table 3 provides the results for medium size firms. TRIN, VOL and MA again are the key sentiment variables to affect medium size firm returns. Table 4 gives similar results for small firms. In firm size analyses, one observation strikes the most as we move from large firms to small firms, adjusted R 2 in general goes down. That is, the relationship between firm size and sentiment is positive. It contradicts with previous finding that small firms are more sensitive to market sentiment. For Bangladesh stock market, the reason for this is probably the fact that investors, in the absence of dependable information, demand large stocks and neglect less informative medium and small stocks. In such a situation, investors in high market sentiment period will mainly focus on large stocks, contributing more sentiment to these stocks. Table 3 and 4 about here 7

8 5. Conclusion This paper looks at Bangladesh stock market completely from behavioral perspective by using five indirect sentiment proxies in the empirical asset pricing models. Findings show that sentiment affects Bangladesh stock market performance. The results are robust to firm sizes. As emerging markets are dominated by individual investors and lack availability of quality information and professional financial analysts services, it is not a surprise that the performances of these markets are subject to sentiment of general investors. Interestingly, large firms are more prone to sentiment than small firms are. Even in the developed markets there is higher possibility of effect of sentiment when stocks are new and difficult to arbitrage. In the Bangladesh stock market, sentiment is a source of market risk, which cannot be diversified away and hence it is priced. Consequently, behavioral factors should be considered in empirical asset pricing models for emerging stock markets. Results also show the possible dynamic relationship between the behavior of stock returns and market sentiment. Higher sentiment contemporaneously leads to higher stock prices. In the next period, investors realize the overreaction of the past period, resulting in negative correction of stock prices. Thus, it seems that the role of fundamentals is rather limited in the Bangladesh stock market. In the absence of dependable earnings information, sentiment-based trading is indeed a good option for active participation in the market. References Aitken, B., Have institutional investors destabilized emerging markets? Contemporary Economic Policy 16 (2), Antoniou, C., Doukas, J. A., Subrahmanyam, A., Cognitive dissonance, sentiment and momentum. Journal of Financial and Quantitative Analysis 48(1), forthcoming. Anusakumar, S., Ali, R., Wooi, H. C., Momentum and investor sentiment: Evidence from Asian stock markets, Seventh Annual Conference on Asia-Pacific Financial Markets, South Korea. Baker, M., Wurgler, J., Investor sentiment and the cross-section of stock returns. Journal of Finance 55, Baker, M., Wurgler, J., Yuan, Y., Global, local, and contagious investor sentiment, Working Paper, Harvard University. 8

9 Barberis, N., Shleifer, A., Vishny, R., A model of investor sentiment. Journal of Financial Economics 49 (3), Basher, S. A., Hassan, M. K., Islam, A. M., Time-varying Volatility and Equity Returns in Bangladesh Stock Market, Applied Financial Economics 17, Beaumont, R., Daele, M, Frijns, B., Lehnert, T., Muller, A., Investor sentiment, mutual fund flows and its impact on returns and volatility. Managerial Finance 34(11), Black, F., Noise, Journal of Finance 41(3), Brown, G. W. and Cliff, M. T., Sentiment and the stock market. Working Paper, University of North Carolina, Chapel Hill. Brown, G. W., Cliff, M. T., Investor sentiment and the near-term stock market. Journal of Empirical Finance 11, Brown, G. W., Cliff, M. T., Investor sentiment and asset valuation. Journal of Business 78, Canbas, S., Kandir, S. Y., Investor sentiment and stock returns: Evidence from Turkey. Emerging Finance and Trade 45(4), Chowdhury, S. S. H., Sharmin, R., 2013, Relevant Factors to Explain Cross-section of Expected Returns of the Firms Listed in the Dhaka Stock Exchange. International Business Research 6(3), Dellavigna, S., Pollet, J. M., Investor inattention and Friday earnings announcements. Journal of Finance 64 (2), De Long, J. B., Shleifer, A., Summers, L. G., Waldman, R. J., Noise trader risk in financial markets. Journal of Political Economy 98(4), Doukas, J. A., Milonas, N. T., Investor sentiment and the closed-end fund puzzle: Out-ofsample evidence. European Financial Management 10(2), Elton, E., Gruber, M., Busse, J. A., Do investors care about sentiment? Journal of Business 71(4), Finter, P., Niessen-Ruenzi, A., Ruenzi, S., The impact of investor sentiment on the German stock market. Working Paper, University of Cologne, Germany. Grigaliuniene, Z., Cibulskiene, D., Investor sentiment effect on stock returns in Scandinavian stock market. Economics and Management 15,

10 Kling, G., Gao, L., Chinese institutional investors sentiment. Journal of International Financial Markets, Institutions and Money 18(4), Kumar, A., Lee, C. M. C., Retail investor sentiment and return comovements. Journal of Finance 61, Lee, C. M. C., Shleifer, A., Thaler, R. H., Investor sentiment and the closed-end fund puzzle. Journal of Finance 64, Lee, W. Y., Jiang, C. X., Indro, D. C., Stock market volatility, excess returns, and the role of investor sentiment. Journal of Banking and Finance 26, Liston, D. P., Haq, S., Huerta, D., Does investor sentiment risk impact the returns and volatility of Islamic equities? Working Paper, Prairie View A&M University. Liu, H. H., Wu, C.C., Su, Y.K., The role of extreme investor sentiment on stock and futures market returns and volatilities in Taiwan. British Journal of Politics and International Relations 11(1), Rehman, U. M., Investor s sentiment and stock market volatility: An empirical evidence from emerging stock market 7(1), Schwarz, N., Situated cognition and the wisdom of feelings: Cognitive tuning. In B. Feldman and P. Salovey (Eds.), The Wisdom of Feelings (pp ). New York: Guilford Press. Swaminathan, B Time-varying expected small returns and closed-end fund discounts. Review of Financial Studies 9, Zhuang, X. C. L., Song, D., 2012, Investor sentiments in the Chinese stock market: An empirical analysis. Applied Economics Letters 19(4), Zouaoui, M., Nouyrigat, G., Beer, F., 2010, How does investor sentiment affect stock market prices? Evidence from panel data. Working Paper, CERAG, France. 10

11 Table 1. Impact of Sentiment on Market Index Returns Model Constant TRIN TRIN VOL VOL IPO IPO MA MA OPCL OPCL Obs. Adj. R (-1.24) (6.07) *** (0.45) (-0.41) (-1.49) (-0.19) (6.55) *** (-1.58) (1.17) (1.24) (-0.05) (-0.50) (-1.66) * (1.39) (0.21) (-0.52) (-1.56) (6.42) *** (-0.81) (1.71) * (-1.71) * (0.27) (-0.03) (-1.68) * (2.14) ** (-1.61) (6.51) *** (-0.85) (1.74) * (-1.72) * (-1.69) * (2.18) ** (-2.14) ** (7.06) *** (-1.01) (1.79) * (1.69) * (-2.64) ** (7.27) *** (-1.46) (-2.04) ** (6.70) *** (-1.28) (-1.70) * (2.20) ** (-0.24) (-2.47) ** (3.21) *** (-0.04) (-0.51) (0.59) (-0.45) (-0.26) (0.90) (-2.55) ** (3.12) *** (-0.28) (2.27) ** (-1.99) * Note: ***, ** and * indicate significance at 1%, 5% and 10% level, respectively. 11

12 Table 2. Impact of Sentiment on Equally Weighted Large Stock Portfolio Returns Model Constant TRIN TRIN VOL VOL IPO IPO MA MA OPCL OPCL Obs. Adj. R (-1.49) (6.73) *** (1.45) (-1.35) (-1.35) (-0.97) (0.06) (7.28) *** (-3.18) *** (2.40) ** (-2.08) ** (-0.97) (-0.63) (-1.82) * (1.10) (-0.44) (0.16) (-1.69) * (6.94) *** (-2.17) ** (3.13) *** (-1.91) (-0.50) (-0.49) (-1.94) * (2.28) ** (-2.02) ** (7.01) *** (-2.16) ** (3.12) *** (-2.04) ** (-1.96) * (2.25) ** (-2.57) ** (7.71) *** (-2.38) ** (3.11) *** (-2.04) ** (-1.77) * (7.84) *** (-2.99) *** (-1.12) (7.12) *** (-2.82) *** (-2.14) ** (2.02) ** (-0.37) (-2.60) ** (3.47) *** (0.33) (-1.07) (0.45) (-0.15) (-0.86) (0.66) (-2.61) ** (3.38) *** (-0.91) (3.49) *** (-2.38) ** Note: ***, ** and * indicate significance at 1%, 5% and 10% level, respectively. 12

13 Table 3. Impact of Sentiment on Equally Weighted Medium Stock Portfolio Returns Model Constant TRIN TRIN VOL VOL IPO IPO MA MA OPCL OPCL Obs. Adj. R (-1.98) * (3.64) *** (2.95) *** (-0.89) (-1.67) * (-0.55) (-0.54) (3.57) *** (-1.81) * (3.14) *** (-2.05) ** (-0.97) (0.38) (-1.92) * (0.75) (0.14) (-0.48) (-1.79) * (3.84) *** (-1.24) (3.98) *** (-2.28) ** (-0.82) (0.51) (-2.19) ** (1.64) (-1.96) * (3.81) *** (-1.18) (4.03) *** (-2.37) ** (-2.28) ** (1.67) * (-2.69) *** (4.54) *** (-1.37) (3.98) *** (-2.35) ** (-1.09) (4.71) *** (-2.09) ** (-0.25) (3.98) *** (-1.96) * (-2.48) ** (1.31) (0.07) (-2.98) *** (2.71) *** (-0.29) (-0.75) (1.31) (-0.40) (-0.44) (1.53) (-3.15) *** (2.59) ** (-1.90) * (4.46) *** (-2.73) *** Note: ***, ** and * indicate significance at 1%, 5% and 10% level, respectively. 13

14 Table 4. Impact of Sentiment on Equally Weighted Small Stock Portfolio Returns Model Constant TRIN TRIN VOL VOL IPO IPO MA MA OPCL OPCL Obs. Adj. R (-1.51) (1.39) (3.06) *** (0.09) (-1.75) * ( ) (-0.31) (1.43) (-1.52) (3.42) *** (-2.23) ** (-0.09) (0.27) (-1.70) * (0.36) (-0.07) (-0.27) (-1.40) (1.77) * (-1.27) (4.34) *** (-2.57) ** (0.03) (0.36) (1.80) * (1.04) (-1.39) (1.77) * (-1.28) (4.43) *** (-2.58) ** (-1.84) * (1.09) (-2.30) ** (2.24) ** (-1.32) (4.52) *** (-2.53) ** (-0.10) (2.55) ** (-2.08) ** (0.70) (2.08) ** (-2.11) ** (-2.08) ** (0.73) (0.30) (-2.55) ** (1.91) * (-0.61) (0.47) (0.63) (-0.56) (0.66) (1.11) (-2.73) *** (1.84) * 11 ( ) (-2.23) ** (5.05) *** (-2.97) *** Note: ***, ** and * indicate significance at 1%, 5% and 10% level, respectively. 14