Elsevier

Finance Research Letters

Volume 26, September 2018, Pages 32-39
Finance Research Letters

Does sentiment matter for stock returns? Evidence from Indian stock market using wavelet approach

https://doi.org/10.1016/j.frl.2017.11.008Get rights and content

Abstract

This article examines the relationship between investor sentiment and stock returns using the data from Indian stock market. We investigate the relationship using a broad set of implicit sentiment proxies and value-weighted market indices. The wavelet method has been used to decompose sentiment variables and stock returns into different timescale frequencies. We find a strong effect of sentiment on return both in the short-and long-run by employing decomposed returns and sentiment proxies at different time-scale frequencies, The study lends support to the fact that whether investors are short-term or long-term traders, their investments activities cannot be delinked from sentiment.

Introduction

Behavioural finance literature suggests that investors are not entirely rational but normal (Statman, 1999) and cognitive bias induces them to make suboptimal investment decisions (Barberis and Thaler, 2003). Cognitive bias of investors with some combinations of limited arbitrage (Shleifer and Vishney, 1997) and short-sale constraint (Miller, 1977) succumb their ability to differentiate between noise and information (Black, 1986). Therefore, when investors are unusually bullish or bearish due to cognitive bias, irrational or noise traders influence equity prices in equilibrium and generate systematic sentiment risk (De Long et al., 1990). Empirical literature supports a positive (negative) relationship between investor sentiment and contemporaneous (expected) stock returns (Baker and Wurgler, 2006, Baker et al., 2011, Schmeling, 2009).

This paper attempts to revisit the sentiment and equity returns relationship using the data from Indian stock market. Our motivation follows three distinctive arguments. First, whether sentiment affects market return or return influences sentiment is open for discourse due to lack of consensus in the empirical literature. Early research by Lee et al. (2002) suggests that excess returns are contemporaneously positively correlated with the shifts in sentiment. However, Brown and Cliff (2004), Solt and Statman (1988), Jansen and Nahuis, 2003, Wang et al., 2006 and Bekiros et al. (2016) document that returns cause sentiment rather than vice versa. Since the findings of existing literature are far from consensus, further validation of sentiment and return relationship is an important enquiry. Second, the impact of noise trading on financial stability is negative (Shleifer and Summers, 1990) and systematic mispricing due to sentiment induced stock valuation can cause substantial resource misallocation (Benhabib et al., 2016, Daniel et al., 2002, Morck et al., 1990). Therefore, the existence of a uni-directional or bi-directional causal relationship between sentiment and stock returns is a pertinent issue from the policy perspective. Third, the influence of investment horizon or time-frequency is expected to have economic and statistical significance for the sentiment and stock return causal relationship (Chu et al., 2016, Marczak and Beissinger, 2016). Trading activity of investors is subject to varied expectations based on fundamental or utilitarian characteristics (i.e., risk appetite, expected return, risk premium, available information) and psychological or value-expressive characteristics such as sentiment (Statman, 1999). Consequently, investment horizon or trading frequency changes from one group of traders to the other making the market very heterogeneous. Heterogeneity created by market participants expected to produce divergent responses to the information shocks appear in the market (Tiwari et al., 2013, Dacorogna et al., 2001). Hence, short investment horizon (high frequency) or long investment horizon (low frequency) may have different implications for sentiment and return relationship. This time-frequency domain approach has been paid limited attention in the existing literature. The time series decomposition of sentiment and return data into high and low frequency using wavelet approach enables us to examine the casual relationship of sentiment and stock in different time horizons.

We extend the related literature in two aspects. First, to our knowledge using single emerging market data this paper perhaps the most extensive study regarding the number of sentiment proxies, alternative equity indices, and wavelet-based approach. The wavelet method (Reboredo and Rivera-Castro, 2014) allows us to examine the causal relationship between stock returns and sentiment in different time horizons. Our analysis also extends to uncover the effect of non-linear causality based on Diks and Panchenko (2006) approach. Second, related literature mostly focused in the context of developed markets, and regardless of the importance of emerging stock market (ESM hereafter) for international portfolio diversification, studies on the role of sentiment and stock returns in these markets are sparse. Due to its different market structure, lower level of market integration and investors’ behaviour, the ESM provides an ideal ground to revisit sentiment and return relationship (Bekaert and Harvey, 2003, Bekaert et al., 2007, Chu et al., 2016, Kim and Nofsinger, 2008). Given the lack of consensus in the existing literature and paucity of research in the context of ESM, our findings help to shed more light on this important issue.

Our results do not provide encouraging evidence on the impact of sentiment on stock returns based on linear causality in the non-decomposed series. However, employing decomposed return and sentiment proxies at different time series frequencies, we observe a strong effect of sentiment on return both in the short and long run.

The remainder of the paper proceeds as follows. Section 2 presents data. Section 3 elaborates methodology. Section 4 discusses results. Section 5 provides robustness tests, and section 6 concludes the paper.

Section snippets

Data

We use weekly data from 1st April 2002 until 30th May 2014. Choice of the sample period is constrained on data availability for sentiment variables. We use value weighted index returns of National Stock Exchange (NSE) of India Nifty 50 (Nifty), NSE Mid Cap (Midcap), and NSE Small Cap (Smallcap). Since, a certain category of stocks are disproportionately sensitive to sentiment effect (Baker and Wurgler, 2006), we focus on large size (Nifty), medium size (Midcap) and small size (Smallcap) indices

Regression analysis on original series

We have employed regression analysis of sentiment on returns of original series and decomposed series (discussed in Section 3.2)Rit=αit+βit(Sentt)+uitDjRt=αjt+βit(DiSentt)+uitRit, Sentt are index returns and sentiment proxies, respectively. Dj represents decomposed series of returns and sentiment at different frequency timescales with j = 1,2,…J.

Wavelet decomposition

The basic wavelets in any family of wavelets consist of father wavelets ϕand mother waveletsψ. Father wavelets integrate to one ∫ϕ(t)dt = 1, and Mother

Investor sentiment and stock returns: original undecomposed series

In order to test the impact of sentiment on returns in original series, individual sentiment proxies are regressed on Nifty, Midcap, and Smallcap returns. Reported results in Table 2 show that sentiment proxies like PCI and TOV have a significant impact on the returns across all three indices. The results are in agreement with Baker and Wurgler (2006), and Baker et al. (2011), where the positive relationship between sentiment and stock returns is observed. Linear Granger causality test reported

Robustness tests

We also perform robustness checks with sentiment index (Sent index) constructed by us to investigate whether the pattern of explanation changes as we move from individual proxy to a composite index. We find evidence of sentiment index Granger causes Smallcap and Midcap returns in the short-run (2–4 months) and long-run (16–32 months), respective while the sentiments index causes Nifty returns both in the short-and long-run (see Table 7). Bi-directional causality is observed between sentiment

Conclusion

This paper attempts to examine the impact of sentiment on stock returns in the emerging Indian stock market. The study reveals that stocks with higher returns like Smallcap and Midcap are more influenced by investor sentiment than the stocks with lower returns (Ni et al., 2015). We find that turnover, Put-call ratio and VIX are consistently good sentiment indicators of stock returns of all sizes. Consistent with the behavioral argument that fear is a strong motivator than confidence; we notice

Acknowledgement

We are thankful to Dr. Lucy M. Brian, the editor of the journal and referees for their insightful and scholastic comments which guided us to improve the article.

References (38)

  • J.C. Reboredo et al.

    Wavelet-based evidence of the impact of oil prices on stock returns

    Int. Rev. Econ. Finance

    (2014)
  • M. Schmeling

    Sentiment and stock returns: some international evidence

    J. Empir. Finance

    (2009)
  • A.k. Tiwari et al.

    Oil price and exchange rates: a wavelet based analysis for India

    Econ. Model.

    (2013)
  • Y.H. Wang et al.

    The relationships between sentiment, returns and volatility

    Int. J. Forecast.

    (2006)
  • M. Baker et al.

    Sentiment and the cross-section of stock returns

    J. Finance

    (2006)
  • M. Baker et al.

    Global, local, and contagious sentiment

    J. Financ. Econ.

    (2011)
  • N.R. Barberis et al.

    A survey of behavioral finance

    Handbook of the Economics of Finance

    (2003)
  • G. Bekaert et al.

    Liquidity and expected returns: lessons from emerging markets

    Rev. Financ. Stud.

    (2007)
  • S. Bekiros et al.

    A non-linear approach for predicting stock returns and volatility with the use of sentiment indices

    Appl. Econ.

    (2016)
  • Cited by (45)

    • Economic sanctions sentiment and global stock markets

      2024, Journal of International Financial Markets, Institutions and Money
    • The role of trust, investor sentiment, and uncertainty on bank stock return performance: Evidence from the MENA region

      2022, Journal of Economic Asymmetries
      Citation Excerpt :

      While Baker and Wurgler (2007) delineated that investor sentiment supported investment risk and future cash flow. Several studies have revealed that investor sentiment had a statistically significant positive effect on stock returns (Gao & Yang, 2018; Lee et al., 2002; Wang et al., 2020) and synchronous stock returns (Baker & Wurgler, 2006; Dash & Maitra, 2018). Lee et al. (2002) found that changes in investor mood positively impacted stock returns in the stock market but negatively affected market volatility.

    View all citing articles on Scopus
    View full text