Twitter and the Stock News Echo Chamber That Whips Up Volatilityby
Most talked-about stocks on social media see more price swings
Traditional news media sees more subdued volatility, volume
Anyone watching the stock market has seen this: a post hits Twitter containing old news, and investors react as if it were new.
A working paper by researchers at Oxford University suggests the phenomenon is prevalent, creating a source of excess volatility for investors. It found evidence that the unmediated flood of news and opinion that pours over the social media transom results in some stocks getting whipped around for weeks by the same facts.
Like many academic studies into equities, the Oxford one examined theories calling into question the market’s ability to process information with perfect efficiency, in this case something called the “stale news” hypothesis. It says that investors probably aren’t the automatons of rationality depicted in classical market models when they can’t even recognize when a story has been reported already.
“Twitter almost by definition stirs up repetition,” Ansgar Walther, the postdoctoral research fellow who co-authored the report, said in a phone interview. “People perceive old information as fresh. If everyone knew it was repeated stuff, the market wouldn’t react as much.”
Volatility climbs appreciably when a stock is mentioned more often on Twitter, and not just for a day -- the researchers noted an effect that is detectable a month later. The most talked-about stocks on social media exhibited volatility that was 50 percent greater than average, as well as a 25 percent increase in average volume over the following 30 days, data compiled in the study show.
The study, titled “Social Media, News Media and the Stock Market,” co-authored by Walther and his Oxford colleagues Peiran Jiao and Andre Veiga, uses a metric they refer to as “buzz” to quantify news coverage of specific stocks. The measure is defined as the number of articles in conventional news outlets or social media posts devoted to a given stock, relative to other companies.
Social media’s role in the stock market is growing. The authors estimated that Twitter and Facebook have about 1.8 billion active users and that 500 million tweets are sent each day. Dozens of companies, including Bloomberg LP, the parent company of Bloomberg News, have products aimed at analyzing social media sentiment toward equities.
Stale news as a theoretical concept was addressed in a 2007 paper by Paul C. Tetlock, who found evidence that investor overreaction to stale information put temporary pressure on asset prices.
“The stale news effect is a very good example of markets not processing information with perfect efficiency,” Juhani Linnainmaa, associate professor of finance at the University of Chicago’s Booth School of Business, said in a phone interview. “It would be very difficult to find any rational explanation for what’s going on here.”
It isn’t just repetition that drives up volatility in the Oxford study. The authors said their findings are consistent with a theory that investors simply get things wrong when they read them on social media. “A model of strong random sentiments can explain our results by assuming that investors’ interpretation of social media content are subject to significant and volatile misperceptions,” they wrote.
As an extension of their research, the study authors also found that stocks receiving the most buzz from traditional news sources actually saw subdued price swings and volume. There’s a behavioral bias for that, too: overconfidence, where traders “believe the precision of the signals they observe is higher than their true precision.” Conventional news stories provide “shocks to disagreement,” damping trading driven by discord.
Could there be other reasons that mentions on social media correspond to higher volatility while traditional news doesn’t? The authors consider several possibilities, among them that people are more likely to post when they disagree, or that large investors try to move markets by getting their opinions out after taking big positions. But neither would explain why the effect lingered for the following month.
The study is a work in progress, according to Walther, who says there are many other ways to explore the behavioral biases shown by investors as they digest different types of content. The next step is to conclude an assessment of how retail and institutional ownership affects their initial findings.
“The main message we want to get across is that news and social media are really different beasts,” said Walther by phone. “But we’ve by no means answered every question.”