Elsevier

Pacific-Basin Finance Journal

Volume 44, September 2017, Pages 160-172
Pacific-Basin Finance Journal

Media sentiment and trading strategies of different types of traders

https://doi.org/10.1016/j.pacfin.2017.07.001Get rights and content

Highlights

  • How does media sentiment influence retail and institutional trader behaviours?

  • Media sentiment leads institutional traders to overreacting to earnings news.

  • Retail traders are indifferent to earnings news in presence of media sentiment.

  • Retail traders' trading decisions are affected by media sentiment alone.

  • Media sentiment complements CSI to provide timelier sentiment measure to investors.

Abstract

This paper investigates how the prevailing sentiment conveyed by the media influences trading by retail and institutional traders around earnings announcements made by ASX200 constituent firms. We find that media sentiment influences institutional traders to overreact to earnings news measured by analyst forecast errors. In contrast, retail traders are indifferent to either positive or negative unexpected earnings in the presence of media sentiment. Further analysis reveals that their trading decisions are affected by media sentiment alone, independent of the earnings surprise. We also find that firm-specific media sentiment complements market-wide sentiment by providing timelier information to investors in shaping their perceptions of unexpected earnings in the Australian market.

Introduction

This paper examines the effects of media sentiment on the trading decisions of market participants, and if the effects differ between different trader types. We provide further evidence to support studies that challenge the traditional market efficiency view, and find that investor sentiment plays a role in shaping trading behaviour (Kumar and Lee, 2006, Hirshleifer, 2001). In particular, we extend the literature by identifying whether sentiment drives traders' behaviour by examining periods around company announcements containing earnings news (Hribar and McInnis, 2012). Further, we examine whether firm-specific sentiment, derived from examining media articles (as in Cahan et al., 2013), has any explanatory power for retail and institutional investors around the earnings announcements, respectively.

The classic assumption of market efficiency posits that asset prices are only affected by fundamental information, and any directional trading behaviour driven by noise will be offset by arbitrageurs. This assumption has been challenged by studies documenting that noise traders' collective trading on market sentiment induces co-movements in prices and shifts in returns (see for example, Barberis et al., 2005, Hirshleifer, 2001, Kumar and Lee, 2006). For instance, Kumar and Lee (2006) find a common directional component in retail investors' trading activity that explains stock returns beyond fundamental factors.

Using the systematic trading by retail investor as a sentiment measure, Kumar and Lee (2006) did not identify the driver of this sentiment in retail investors' trades. Our study aims to investigate the trading behaviour of retail and institutional investors with a particular focus on the source of their directional trading. We do so by investigating the net purchases of these investors around earnings announcements. The role investor sentiment has in the market's reaction to price-sensitive information has been established in the literature, with several studies examining behavioural biases around earnings releases. For example, Mian and Sankaraguruswamy (2012) and Hribar and McInnis (2012) use a market-wide measure for investor sentiment to show that stock prices are systematically affected by the prevailing mood in equity markets. If sentiment affects investors' beliefs of future stock performance, then it should also affect the way they trade. The current paper therefore directly investigates the impact of our sentiment measure on trader behaviour for individuals and institutions.

Although the focal point of a number of studies is on market sentiment, Baker and Wurgler (2006) proposed that a firm-specific level of sentiment exists on the notion that market-wide sentiment measures are only able to reflect the general mood in equity markets. The advantage of using a firm-specific sentiment measure is that it captures more granular and timely changes in sentiment about individual stocks while macroeconomic conditions can be slow moving. In view of this, Cahan et al. (2013) constructed a firm-specific sentiment measure using the optimistic and pessimistic language embedded in the media coverage about the firm. They found that stock prices have a tendency to overreact to positive (negative) unexpected earnings when sentiment is high (low), establishing a role for the firm-specific media sentiment measure that is incremental to market-wide investor sentiment. We adopt Cahan et al.'s (2013) approach by using the prevailing sentiment presented by media coverage as a measure of firm-specific sentiment in our study. This paper contributes to the growing literature on the effects of investor sentiment on trader behaviour by investigating the influence of media sentiment as a firm-specific sentiment incremental to that of market-wide sentiment around earnings announcements.

Greater accessibility to information through media is typically considered advantageous for market participants as it reduces the asymmetric advantage of informed as opposed to uninformed investors. The manner in which news is delivered and conveyed (i.e., tone of the news) can also have a profound influence on the way it is interpreted by traders – that is, media sentiment is also likely to influence investors' decisions. For instance, Busse and Green (2002) found that within 15 seconds of CNBC1 providing a positive report, the firm's referenced within the report experienced significant increases in their stock price. Interestingly, the authors found that the market's reaction is not justified by the content in the news report.2 Tetlock (2007) constructed a sentiment index from the ‘Abreast of the Market’ column in the Wall Street Journal and found that media pessimism can affect the market's trading behaviour. He contends that economists who read the column in the Wall Street Journal consider it entertainment and that no value-relevant information exists.

In addition, we extend the existing literature by examining the influence of sentiment on different types of traders. As media outlets act as information intermediaries between the firms and market participants, it is important to investigate how information is disseminated by the media (i.e., content and sentiment) influences the trading decisions of different market participant types. Higher levels of press coverage have been found to improve the information environment of stocks, resulting in lower levels of information asymmetry around earnings announcements (Bushee et al., 2010). In isolation, these findings might imply that greater media coverage is entirely beneficial. However, Barber and Odean (2008) argue that access to excessive volumes of information creates a ‘search problem’ for investors. They find that individual traders are particularly susceptible to cognitive limitations and, as a result, make trading decisions based on what captures their attention. Barber and Odean also argue that institutional traders such as funds and brokerage firms utilise computers and algorithms to interpret the vast array of information produced by the media and, therefore, are less susceptible to similar cognitive biases. This speaks to the representative heuristic introduced by Tversky and Kahneman (1973), suggesting that compared to institutional investors, retail traders have a tendency to disregard older news contained in analyst forecast errors and accept attention grabbing information, which leads to trading decisions based on information that is expected to represent state of the world. Tan et al. (2014) argued that institutional traders are superior at interpreting media coverage and disclosures provided by management and therefore are less prone to the effects of sentiment. Furthermore, Hendershott et al. (2015) found empirical evidence that shows institutional traders are informed about the occurrence as well as the level of sentiment in news coverage. On the contrary, Kumar and Lee (2006) report that sentiment measured by retail investors' directional trading alone drives the expectations of retail traders regarding earnings figures.

Prior literature that have used trade size to differentiate informed (institutional) from uninformed (retail) traders include Barclay and Warner (1993), Battalio and Mendenhall (2005), Cready et al. (2014). However, Wee and Yang's (2016) recent finding that trade sizes used by all traders have dramatically decreased over the years makes this classification less accurate. For this reason, our study uses a sample of Australian listed stocks as the data allows us to separate the trades made by institutional and retail traders using the broker classification (Fong et al., 2014).

Our findings show that an interaction between media sentiment and unexpected earnings exists only for institutional traders. This implies that media sentiment produces either an upward or a downward bias in institutional trading behaviour. The insignificance in the interaction between unexpected earnings and media sentiment for retail traders suggests that these variables operate independently. Further analysis reveals that retail traders are more biased towards the prevailing media sentiment, rather than the magnitude of unexpected earnings.

The remaining sections are organised as follows: Section 2 provides a review of prior research relating to sentiment and discusses our contribution to existing literature, Section 3 describes the data and method, 4 Results, 5 Robustness tests report the results and robustness tests, and Section 6 summarises and discusses the findings.

Section snippets

Literature review and contribution

The behavioural finance literature has challenged the traditional assumption that individual irrationalities do not lead to directional trading behaviour, by documenting that the moods and sentiments of investors affect price movements (Hirshleifer, 2001, Kumar and Lee, 2006). Sentiment refers to the beliefs of market participants regarding the future performance of a company that are not justified by fundamentals.3 Theoretical support for this

Data and method

The initial sample, collected from Morningstar's DatAnalysis database, comprise all annual earnings announcements by individual companies constituting the ASX200 index from 2009 to 2013. To be able to compute analyst forecast errors, we excluded firm-year observations where no analyst forecasts were available on the I/B/E/S database or when the forecasts were not reported in Australian dollars.8

Descriptive statistics

Table 1 provides the descriptive statistics for all variables. The mean and the median values of the variable NetBuy for institutional and retail traders indicate that the former exhibit a negative skewness distribution, and the latter a positive skewness distribution. This means that, on average, retail traders are more likely to be net buyers with more abnormal trading activity during the earnings announcements, and the opposite was concluded for institutional investors. This finding is

An alternative media sentiment measure

In this section we employ an alternative measure for media sentiment and investigate whether our measure for sentiment captures value relevant information. As Tetlock (2007) suggested, media sentiment may represent a firm's fundamentals to some degree. For this reason, we adopted Cahan et al.'s (2013) method to purge the sentiment scores by first regressing the scores against an array of firm fundamentals and using the residual as a media sentiment measure.

Specifically, the firm fundamental

Conclusion

This paper investigated how the prevailing sentiment portrayed by the media influences the trading decisions of retail and institutional traders around earnings announcements. We found that the two types of traders respond very differently to unexpected earnings news in the presence of media sentiment. In particular, media sentiment influences institutional traders to overreact to both good and bad earnings news. When sentiment is high, good earnings news leads to even stronger net purchases,

References (47)

  • Pi-Hsia Hung

    Investor sentiment, order submission, and investment performance on the taiwan stock exchange

    Pac. Basin Financ. J.

    (2016)
  • C. Lee

    Earnings news and small traders: an intraday analysis

    J. Account. Econ.

    (1992)
  • C. Lee

    Market efficiency and accounting research: a discussion of capital market research in accounting by SP Kothari

    J. Account. Econ.

    (2001)
  • A. Tversky et al.

    Availability: a heuristic ror judging frequency and probability

    Cogn. Psychol.

    (1973)
  • M. Baker et al.

    A catering theory of dividends

    J. Financ.

    (2004)
  • M. Baker et al.

    Investor sentiment and the cross-section of stock returns

    J. Financ.

    (2006)
  • B. Barber et al.

    All that glitters: the effect of attention and news on the buying behavior of individual and institutional investors

    Rev. Financ. Stud.

    (2008)
  • N. Bergman et al.

    Investor sentiment and corporate disclosure

    J. Account. Res.

    (2008)
  • N. Bhattacharya

    Investors' trade size and trading responses around earnings announcements: an empirical investigation

    Journal of Account. Rev.

    (2001)
  • F. Black

    Noise

    J. Financ.

    (1986)
  • B. Bushee et al.

    The role of the business press as an information intermediary

    J. Account. Res.

    (2010)
  • S. Cahan et al.

    Media Sentiment, Investor Sentiment, and Stock Price Sensitivity to Earnings

    (2013)
  • M. Chang et al.

    Sentiment, Market Order Choice, and Returns

    (2013)
  • Cited by (8)

    View all citing articles on Scopus
    View full text