Can Social Media Inform Corporate Decisions? Evidence from Merger Withdrawals

Financial social media has transformed the way investors share opinions, access information, and interact with markets. While there is extensive evidence that social media affects investor behavior and asset prices, a fundamental question remains: does social media also influence corporate decisions? In our paper, Can Social Media Inform Corporate Decisions? Evidence from Merger Withdrawals, forthcoming in the Journal of Finance, we provide evidence that it does.

We study one of the most consequential corporate investment decisions, mergers and acquisitions (M&As), and show that the social media sentiment reaction to a merger announcement is a strong predictor of whether the deal will ultimately be completed or withdrawn. Our core finding is striking: a one standard deviation decrease in abnormal social media sentiment after a merger announcement is associated with a 0.64 percentage point greater likelihood of merger withdrawal, representing roughly 16.6% of the baseline withdrawal rate in our sample. Importantly, this effect is not explained by the stock market’s reaction to the deal, traditional news media coverage, or analyst recommendations.

Measuring Social Media Sentiment

Our analysis uses a sample of U.S. M&A announcements from 2010 through 2021 matched to firm-day sentiment measures from a major financial social media platform, StockTwits. StockTwits is well suited to this question because users explicitly discuss firms using “cashtags” that indicate the firm’s ticker (for example, $AAPL for Apple), and the investor discussion on the platform is both in real time and high frequency. Our dataset includes over 250 million StockTwits messages and over 6,438 announced deals. We measure abnormal sentiment for each merger announcement by comparing the average sentiment of messages in the days immediately following the announcement (days 0 through 3) relative to a benchmark period before the announcement became public.

Social Media Contains Unique Information

Do social media sentiment reactions aggregate other sources of information, or does social media contain new information? We find social media information is distinct from other sources. Conditional on a positive stock market reaction to the deal, the StockTwits signal is only slightly more likely to be positive than negative: 55% of positive CAR deals have positive abnormal sentiment versus 45% with negative abnormal sentiment. Thus, a positive social media sentiment reaction to the merger announcement is not simply a reflection of the market reaction. The same broad conclusion holds for traditional news sentiment and analyst recommendations: there is significant dispersion in abnormal social media sentiment within both positive and negative values of these other signals, implying unique variation in the social media measure.

Evidence for Manager Learning

These findings raise a natural question: do managers learn from the social media signal or does sentiment simply correlate with other factors that independently lead to withdrawal? Several pieces of evidence point toward the learning channel. First, the relationship between the withdrawal likelihood and abnormal sentiment increases after acquirers establish corporate Twitter accounts, especially for firms with active Twitter accounts. This is consistent with firms putting more weight on social media signals once they maintain a presence on these platforms. Second, the relationship weakens when we restrict attention to withdrawals driven by regulators or target decisions, suggesting that a sizable portion of the main effect reflects managerial learning rather than pure prediction of outcomes outside managerial control. Third, the relationship between withdrawals and abnormal sentiment survives controls for CEO network measures, acquirer fixed effects, and alternative sentiment constructions including Twitter-based measures, which reduces the likelihood that an unobserved correlated channel explains the pattern. Taken together, these results are most consistent with managers learning from substantive investor messages.

Which Social Media Content Carries Information?

We next examine which aspects of social media content carry information for managers. StockTwits users differ in their investment approach, so we leverage the content of the tweets to classify users by whether their posts are consistent with fundamental analysis or technical trading. We also separate messages by length. Consistent with managers learning from context, the predictive power of social media is greater for longer, more detailed messages and from     posts from users who tend to write about fundamental analysis, while short posts and technically oriented messages are not predictive. Digging deeper, we apply a topic model to identify themes in discussions around merger announcements. Sentiment from topics focused on company fundamentals, disclosures, and deal terms predicts withdrawals, whereas sentiment from meme-related or technical-topic clusters does not. Together, these content-based tests indicate that the informative component of social media sentiment resides in longer, information-rich posts that discuss fundamentals and deal-relevant issues, consistent with managers learning from substantive, deal-relevant analysis rather than noise.

We further study how changes in the broader social media environment influence informativeness. Beginning in early 2021, investor activity on several platforms shifted sharply toward attention-driven discussions during the GameStop trading episode and the broader rise of meme stocks. Consistent with this change in content composition, the predictive power of social media sentiment for merger withdrawals declines after this period. This pattern aligns with the idea that social media is informative when discussions concentrate on fundamentals and deal-relevant issues, but less useful when platforms are dominated by noise or attention-driven dynamics.

Ruling Out the Governance Channel

Although our evidence supports the idea that managers learn from social media, an alternative theory is that social media offers a channel for governance, disciplining managers who pursue value-destroying mergers  If this governance channel drives the connection between sentiment and withdrawals, social media sentiment should matter more for deals that generate negative announcement returns and for firms with weaker governance structures. In contrast, we find a similar relationship between sentiment and withdrawals for deals with positive and negative announcement returns, and the magnitude does not strengthen for firms with weaker governance. These findings indicate that a governance-based channel is unlikely to account for the main result.

Implications

To summarize, our findings show that social media influences real economic decisions, not just financial markets, by serving as a meaningful source of information for corporate decision-making. These results contribute to the broader debate about whether social media benefits or harms financial markets. While concerns about social media amplifying behavioral biases and enabling trading frenzies are well-founded, our evidence suggests that social media can also serve as an important channel for information transmission with important consequences for economic outcomes, complementing traditional sources like market prices and news media. Around merger announcements, investors and social media participants generate substantial M&A-relevant discussion, and managers appear to learn from these discussions when evaluating whether to proceed with an acquisition.  We also find that withdrawals of deals with negative social media sentiment are associated with positive long-run returns for acquirers, suggesting that social media may help identify deals that should not proceed. As firms continue to engage with investors through digital platforms, understanding how these platforms influence corporate outcomes will become increasingly important.