Social media and finance

Social media has transformed how investors access information, form beliefs, and interact with markets. This session establishes the research landscape: how to think about social media data, where to find it, and what kinds of questions it can answer. The course organizes the literature along two axes — whether social media is used as a lens on investors and markets, or whether the focus is on social media's direct effects — and, within effects, whether the channel runs through the production, consumption, or distribution of financial information. Three landmark papers anchor the session. Chen et al. (2014, RFS) show that the textual content of Seeking Alpha articles predicts future stock returns, establishing that social media carries genuine signal. Bailey et al. (2018, JPE) use Facebook friendship networks to show that social connections shape real economic decisions, including house purchases and prices. Müller, Pan and Schwarz use the 2007 South-by-Southwest festival as an instrument for Twitter adoption and find that social media exposure drives stock market participation. Together, the three papers span the breadth of the field.

From the slides

Social media is not merely a source of financial noise — it is a window into investors' deeper beliefs, identities, and reasoning styles. This session develops the "social media as a lens" approach, in which the abundant, high-frequency data generated by investors' online activity is used to study questions that were previously unanswerable. Three original papers illustrate the approach. Cookson and Niessner (2020, JF) exploit self-labeled bullish and bearish posts on StockTwits to construct a daily measure of investor disagreement, showing that differences in investment philosophy — fundamental versus technical — account for a large share of trading volume. Cookson, Engelberg and Mullins (2020, RAPS) show that partisan identity shapes investment beliefs even when the stakes are high: partisan Republicans maintained substantially more optimism than others throughout the COVID-19 pandemic. Cookson, Dim and Niessner use CAPS, the prediction platform of The Motley Fool, to measure investor horizons directly, finding that short- and long-horizon investors update their beliefs differently around earnings announcements, acquisition rumors, and market shocks — with horizon disagreement independently predicting trading volume.

From the slides

Social media does not merely reflect investor beliefs — it also shapes them, through selective exposure, information cascades, and the formation of echo chambers. This session examines the social transmission of information and its consequences for market efficiency, then turns to the problem of extracting reliable signals from social media data. The first half focuses on echo chambers: Cookson, Engelberg and Mullins (2023, RFS) show that investors on StockTwits are five times more likely to follow users who share their views, that the resulting information silos generate worse stock picks, and that echo chambers independently account for a large share of aggregate trading volume — suggesting that selective exposure sustains the very disagreement that drives markets. The second half examines cross-platform signal extraction. "The Social Signal" (Cookson, Lu, Mullins and Niessner, 2024, JFE) shows that investor attention is common across platforms while sentiment is platform-specific, with sharp implications for return predictability. At the market level, Cookson, Lu, Mullins and Niessner (WP) show that aggregate sentiment and attention both predict negative future returns — through distinct dynamics.

From the slides

The consequences of social media extend beyond individual investor behavior into the structure of information production, corporate decision-making, and financial stability. This session examines three channels through which social media produces real effects. Dessaint, Foucault and Frésard (2024, JF) show that the entry of social media into a firm's information environment shifts analyst forecasts: short-horizon forecasts become more informative while long-horizon forecasts deteriorate, consistent with social media being a short-horizon information source. Cookson, Niessner and Schiller (2026, JF) ask whether social media feeds back into corporate decisions, using M&A deal withdrawals as a setting; they find that post-announcement social media sentiment predicts withdrawal at a magnitude comparable to the market's own price signal, and the effect is strongest when managers have the most to learn from external information. Finally, Cookson, Fox, Schiller, Gil-Bazo and Imbet (2026, JFE) study the collapse of Silicon Valley Bank — the first social media bank run — showing that pre-existing Twitter exposure amplified classical bank run risk factors, with top-tercile exposure predicting larger stock losses than a one-standard-deviation increase in uninsured deposits.

From the slides