Confirmation bias on social media is nothing new in politics and news. New research shows it's a challenge in investing, too.
As ongoing revelations about Facebook’s algorithm are showing, social media’s ability to shape our attitudes by steering users into echo chambers is raising questions about our national discourse and drawing increasing attention from regulators.
Echo chambers—formed when users are directed to media that confirm their existing beliefs, without providing a balanced view—are pervasive in media and politics. What’s less understood is whether these echo chambers affect our daily lives, including our personal investing decisions.
Now, new research from the Leeds School of Business at the University of Colorado Boulder is shedding light on the pervasiveness of echo chambers in investor beliefs and financial decisions.
“A financial context is a surprising place to see echo chambers,” said Tony Cookson, associate professor of finance at Leeds. “In politics, you can be wrong, but you don't have to pay a price for it. In finance, if you’re wrong, you pay for it through worse returns—yet we still find echo chambers are pervasive.”
In his research, Cookson and his coauthors relied on data from StockTwits—a platform similar to Twitter, but devoted to conversations about investing. Another important difference is that, unlike Facebook or Twitter, StockTwits does not have a strong recommendation engine to suggest followers or posts based on the user’s activity.
As it turns out, people don’t need an algorithm to put themselves in an echo chamber.
“If I declare in my posts that I’m bullish on Tesla, and then subsequently start following other Tesla bulls, that looks like I’m like tilting my information sourcing toward people who agree with me, versus just tilting it toward the topic itself,” Cookson said. “And this is actually pervasive across all kinds of stocks that people discuss on these platforms.”
A troubling trading trend?
What’s more, Cookson and coauthors found users who put themselves in echo chambers were more likely to be active traders—perhaps to their own peril, given the limits of their information. StockTwits is not a trading platform, so the researchers can’t see actual trades, but based on their posting activity, “we see these really strident disagreements by people who get their information from different sources, and this leads to persistent and sustainable trade in a really publicly observable way,” he said.
StockTwits users represent a good cross-section of investors and include both amateur traders as well as professionals. If you think being a pro trader insulates you from echo chambers, “it doesn’t—it’s to a lesser degree than novices, but professionals aren't immune to wanting to confirm their existing beliefs,” Cookson said.
“In politics, you can be wrong, but you don't have to pay a price for it. In finance, if you're wrong, you pay for it through worse returns.”
His findings about investment echo chambers represent the latest research thread for Cookson, a prolific publisher who also serves as an associate editor at the Journal of Financial Economics. His work has explored a wide variety of topics from bank mergers to fracking royalties to casinos. At the heart of this new research agenda, Cookson is trying to understand the role of information in financial markets and how it drives investor behavior.
He has been working with StockTwits data since about 2015, but interest this latest work on social platforms has intensified following some of the interesting retail investing climate, including outages on Robinhood and the GameStop short squeeze—the latter also driven by social media, in this case, Reddit. Cookson’s working paper won the NASDAQ Award for best paper on asset pricing at the 2021 Western Finance Association meetings; it also won best paper awards from the Chicago Quantitative Alliance and Midwest Finance Association.
“We were always getting interest from people who work in investor behavior and social media, but then this became front page news, with Reddit,” he said. “That doesn’t happen too often in academic research, so it has been exciting to see this play out.”
This research is a working paper available on the Social Sciences Research Network. Cookson’s co-authors included Joey Engelberg and Will Mullins, both of the University of California San Diego.