By LARRY SWEDROE
The increased popularity of social media as a forum for market participants to post and exchange opinions has been accompanied by heightened interest from academic researchers who have sought to determine if there is valuable information in the postings.
For example, the June 2020 study Do Individual Investors Trade on Investment-related Internet Postings? investigated whether social media postings help individual investors identify investment strategies that deliver superior performance in the future. The authors found that “it is mainly unsophisticated individuals who rely on investment-related Internet postings when making investment decisions, but this does not help them identify traders with superior skills.” These findings are consistent with those of the authors of the November 2020 study Attention Induced Trading and Returns: Evidence from Robinhood Users, who found: “Large increases in Robinhood users are often accompanied by large price spikes and are followed by reliably negative returns.”
The findings are also consistent with those of the authors of the March 2021 paper The Rise of Reddit: How Social Media Affects Retail Investors and Short-sellers’ Roles in Price Discovery, who found that “Reddit social media activity encourages retail buying behavior, and deters shorting.” They added: “Social media activity and retail flows cultivate price bubbles, while the short-sellers correct the bubbles created by social media activity and retail order flows.” And the author of the August 2020 study Investor Emotions and Earnings Announcements had a particularly interesting finding—investors are typically excited about firms that do end up exceeding expectations, but their enthusiasm was excessive and resulted in negative post-announcement returns.
Unfortunately, the body of evidence demonstrates that naive retail investors can be easily convinced they have an edge — they know something the market hasn’t yet incorporated into prices. Sadly, the evidence also shows that while these less sophisticated investors can be convinced they “know” something by finding an “expert” on a social media platform, the results of trading activities based on following “experts” show negative outcomes. As usual, the ones benefiting are the platforms (like Robinhood), not the investors who use them.
Sell-side analysts and social media
Ann Marie Hibbert, Qiang Kang, Alok Kumar and Suchi Mishra took a different approach to the issue of how social media impacts markets. In their February 2022 study Twitter Information, Analyst Behavior, and Market Efficiency, they examined whether sell-side equity analysts are able to effectively extract information from social media to improve their earnings forecasting performance. They used Bloomberg’s daily Twitter sentiment data on S&P 500 firms over the period 2015-2019 to determine if that was the case.
The authors began by noting: “One strand of psychology literature shows that, across a broad range of contexts, negative information is processed more thoroughly than positive information. … Consequently, negative information would be more influential than comparable positive information.” They added: “The same psychology literature also demonstrates that negative information elicits more thorough and careful information processing than positive information. Therefore, negative information may capture more attention and receive more conscious processing.” And finally, they noted: “Due to either conflict of interest or economic incentives, analysts issue overly optimistic earnings forecasts.
Innovation in information technology such as the advent and prevalence of social media have intensified competition in information production, which may induce analysts to make less biased forecasts.” And in fact, the authors of the 2016 study The Value of Crowdsourced Earnings Forecasts found such behavior among firms on Estimize, an open platform that crowdsources short-term earnings forecasts.
Hibbert, Kang, Kumar and Mishra found that while analysts’ forecasts are too optimistic on average, Twitter information tends to be relatively more pessimistic than traditional news.
They also found that while positive Twitter information had little or no impact on analyst forecasts, more negative Twitter information was associated with more pessimistic (less optimistic) and more accurate earnings forecasts—Twitter information reduces forecast optimism and improves forecast accuracy of equity analysts. The effect was also greater for smaller firms with greater information asymmetry. The bottom line was that, in aggregate, Twitter-sensitive firms have smaller earnings surprises and consequently weaker stock market reaction—the post-earnings-announcement drift (PEAD) anomaly (the tendency for a stock’s cumulative abnormal returns to drift in the direction of an earnings surprise for several weeks, or even several months, following an earnings announcement) was reduced, especially for negative earnings surprises.
Their findings led Hibbert, Kang, Kumar and Mishra to conclude: “Collectively, these results suggest that financial analysts extract useful information from Twitter, improving their overall forecasting performance and market efficiency. Investors recognize this relation and respond to this phenomenon accordingly.”
All of these studies focused on equity markets. And there are major differences between equity and corporate bond markets. For example, equity markets are generally much more liquid because liquidity in the corporate bond market is limited, as most bonds do not trade.
In addition, while retail investors play a significant role in equity markets, corporate bonds are almost exclusively the realm of institutional investors, considered to be more sophisticated and less subject to emotion-driven biases. Thanks to Eli Bartov, Lucile Faurel and Partha Mohanram, authors of the April 2022 study The Role of Social Media in the Corporate Bond Market: Evidence from Twitter, we now have evidence on the role of social media in the corporate bond market as well.
Evidence from the corporate bond market
Bartov, Faurel and Mohanram examined the role of social media information in the corporate bond market by testing the ability of Twitter opinions (OPI) to predict bond returns and changes in credit default swap (CDS) contracts’ spreads around the upcoming quarterly earnings announcements—prior research showed that bond-trading volume increases sharply around earnings announcements. Their test variable was the aggregate OPI during the window <-10;-3>, where day 0 was the earnings announcement date. They ended their OPI measurement window on day -3 because some of their tests involved the window <-2;+2>. They focused on the window <10;-3> because prior research documented a concentration of Twitter activity in the period just prior to quarterly earnings announcements.
They measured OPI using textual-analysis methodologies focusing on the words that comprised each individual tweet. Their tests controlled for a variety of firm characteristics: size, value, profitability and leverage.
Their data sample covered 2,692,185 tweets (9,404 firm-quarters; 1,158 unique firms) over the period December 17, 2008-December 31, 2012 and considered all stocks with traded corporate bonds that were ever included in the Russell 3000 Index during the period. Following is a summary of their findings:
Controlling for earnings surprises and announcement stock returns, OPI was significantly positively associated with bond returns and significantly negatively associated with changes in CDS spreads around quarterly earnings announcements.
Bond prices are more sensitive to bad earnings news than good earnings news, consistent with the nonlinear payoff structure of bonds—the association between OPI and announcement bond returns (changes in CDS spreads) was more positive (more negative) for bad news compared to good news.
The association between OPI and announcement bond returns/changes in CDS spreads was stronger for tweets containing information directly related to bonds and credit risk and when information uncertainty was high.
OPI was significantly negatively associated with the probability of a future credit rating downgrade but insignificantly related to the probability of a future upgrade.
OPI was strongly negatively associated with future changes in implied default probability for three measures of default risk: the Altman Z-score, the Ohlson O-score and the Black-Scholes-Merton models (which view a company’s equity as a call option on its assets). In addition, the association was stronger for speculative-grade bonds, confirming findings in prior studies that speculative-grade bond prices show a greater sensitivity to news compared to investment-grade bonds. Further, this association was stronger for firms with greater information uncertainty, where Twitter information is more likely to be incrementally meaningful.
Their findings were robust to various tests of the Twitter data, such as level of Twitter activity and original versus dissemination tweets.
Their findings led Bartov, Faurel and Mohanram to conclude: “We provide novel evidence that social media appears to convey economically important information to even the presumably sophisticated investors who dominate the corporate bond market, an important financial market that differs significantly from the stock market, as well as to credit rating agencies.”
Investor takeaways
In our book The Incredible Shrinking Alpha, Andrew Berkin and I provided the evidence demonstrating that it is persistently more difficult for active managers to add alpha (outperform appropriate risk-adjusted benchmarks). We also demonstrated that there are four main themes that explain the shrinking alpha:
Academic research has been converting what was once alpha into beta.
The pool of victims that can be exploited has been shrinking.
The competition has been getting tougher.
The supply of dollars chasing the shrinking pool of alpha has increased.
The latest research on the impact of social media provides us with yet another explanation: Social media provides analysts with information that reduces their forecasting errors — another example of the benefits of the wisdom of crowds (at least when they act independently, not in herds). The result is that the ability to generate alpha continues to be under assault — trying to outperform the market by stock selection is becoming even more of a loser’s game.
Picture; Alexander Shatov via Unsplash
© The Evidence-Based Investor MMXXIV. All rights reserved. Unauthorised use and/ or duplication of this material without express and written permission is strictly prohibited.