Working Papers

Using data from a Twitter-like investor social platform, we document evidence consistent with self-enhancing transmission bias. We find investors are more likely to post about their better-performing stocks. Their followers are more likely to buy the posted stocks than others, and postings’ effect on follow-up purchases is related to postings’ perceived credibility. The performance-postings relationship is stronger among more volatile stocks and the relationship between postings and follow-up purchases is stronger among stocks with higher recent returns, shedding light on the spread of high-variance and extrapolative strategies. We also document that the social network features influential nodes.

Large language models (LLMs) can emulate human perspectives. Leveraging this idea, we study how investor disagreement emerges in response to firm news. We endow an LLM with 216 representative investor personas and elicit buy, hold, or sell responses to S&P 500 firm news headlines from 2010-2025. Dispersion in responses yields article-level disagreement for 5.5 million headlines and sheds light on its sources. Disagreement is highest for socially-oriented news and lowest for fundamentals. Persona responses reflect non-pecuniary rationales and align with human-survey benchmarks. Disagreement predicts elevated same- and next-day abnormal trading volume, especially retail, and results persist beyond the LLM’s training cutoff.

We study how celebrity influencers affect financial decision-making and investor behavior. Using data from social media platforms, we document that celebrity endorsements significantly influence retail investors' trading activity and portfolio choices, with effects persisting beyond the initial exposure.

This paper studies the investing behavior and asset pricing implications of the new classof retail traders exemplified by the relatively young, small, and inexperienced individualinvestors on the Robinhood platform. I first document an absence of investing in lotterystocks, value stocks, and small-cap stocks, in contrast to prior literature, and show thatthese differences arise from a combination of investor behavior and market design changessuch as the introduction of fractional share trading. I then identify three key drivers ofRobinhood investment: a novel “buy-the-dip” effect, event-based trading in response toearnings announcements and analyst recommendation revisions, and trading connectedto popularity and sentiment on the WallStreetBets platform. I develop a model to shedlight on the buy-the-dip phenomenon and introduce a novel financial dictionary basedon WallStreetBets sentiment. Finally, I analyze performance and show that Robinhoodinvestment predicts returns and improves price discovery up to a one month horizon.

We find that consumption spreads through social networks via a "visibility bias" channel, consistent with the model of Han et al. (2023). Using county-level Facebook connectedness and exogenous fracking shocks, we show that a 1% increase in a closely linked county's consumption raises local spending by 0.35% next year. Heterogeneous effects-stronger among more socially connected households and for socially visible goods-indicate transmission through social observation instead of economic linkages. Without their own income gains, households respond by buying cheaper goods, and peer-induced spending increases delinquency rates, underscoring that biased social observation, rather than economic connectedness, undermines financial stability.

We examine whether improved access to firm disclosures weakens social media echo chambers, using the introduction of livestreamed earnings calls on StockTwits as a shock. We find that call streaming is associated with lower selective following, reduced exposure to confirmatory content, and belief revision over the following 30 days. Both call content and real-time chats contribute, with the strongest effects when calls surface disconfirming information and chats are active and diverse. The findings suggest that reducing frictions to accessing primary information and increasing exposure to diverse views can help mitigate the consequences of echo chambers in financial markets.

We revisit the puzzle of limited stock market participation using qualitative methods common in other social sciences but rare in economics. Through in-depth interviews with investors and non-investors in Germany—a high-income country with low market participation—we elicit openended reflections on money without mentioning investing upfront. This allows beliefs and barriers to emerge naturally. We analyze these interviews using traditional human-led content analysis, complemented with a large language model (LLM)-based approach. We validate our findings using a representative survey of more than 7,000 individuals. While many known factors appear, we uncover a pervasive misconception: participation is believed to require selecting “safe” stocks, avoiding “bad” ones, and timing the market through monitoring and frequent trading. This inflates perceived costs and deters participation. Some investors overcome these barriers with support from family, friends, or trusted advisors. Notably, even active investors hold these beliefs, suggesting the misconception influences both entry and behavior in the market.

This paper examines how investors react to CEOs' personal use of social media and how these reactions affect firm valuation. We theorize two mechanisms: a corporate communication channel, in which work-related tweets enhance transparency and increase firm value, and a CEO distraction channel, in which non-work-related tweets reduce value by signaling distraction or self-promotion. Using a panel of S&P 1500 firms from 2006-2020, covering 248 CEOs and over 250,000 tweets, we find that Tobin's Q rises with work-related tweeting volume but falls with non-work-related tweeting volume. These effects are moderated by information environment, CEO celebrity, and investor horizons. A supplementary survey of professional investors corroborates these mechanisms. Our results highlight social media's dual role as both a communication tool and an agency cost.

During the 2020-2021 "AAPI Hate," mutual funds led by female managers perceived as East Asian underperformed relative to other female managers. This effect is stronger in states with higher anti-Asian animus, among more actively-managed funds, and when these managers hold sole or senior roles. The effect cannot be explained by factors concurrent with COVID-19, including childcare challenges, concerns for overseas families, marketplace and workplace discrimination, and exposure to the Chinese economy. Underperformance is traceable to poor stock picking due to impaired ability in generating private information. Racial-ethnic animosity, even outside workplace or marketplace, hinders productivity and decision-making in high-skill professions.

We study how social connections influence financial advice seeking and investment decisions. Using comprehensive administrative data on household portfolios and social network information, we find that individuals with banker connections in their social network are more likely to seek financial advice and hold more diversified portfolios.

Based on data gathered from 27,000 real-world meetings between financial advisors and clients of a large German bank, we show that advisors offer more self-serving advice to women, while men are more likely to receive sales fee rebates and less likely to be recommended expensive bank-owned funds. Additional client and advisor surveys provide evidence consistent with statistical discrimination based on gender as a proxy for client financial sophistication with female clients exhibiting lower literacy, confidence, and price sensitivity. Moreover, female advisors report less confidence in their own professional skills and engage in less discrimination than their male colleagues.

While theory suggests that firms should remain silent on divisive sociopolitical issues due to uncertain investor reactions (Bond and Zeng 2022), we document that nearly 30% of S&P 1500 firms publicly express support for the Black Lives Matter (BLM) movement on Twitter. This disclosure is positively associated with proxies for the management team’s liberal ideology. We provide evidence of ideologically driven investor responses to BLM disclosures: liberal-leaning mutual and hedge fund managers exhibit abnormal purchases of BLM-supporting firms, whereas conservative-leaning managers exhibit abnormal sales. Similarly, liberal-leaning depositors increase their holdings of riskless insured deposits at BLM-supporting banks, while conservative-leaning depositors reduce theirs. These shifts occur despite minimal changes in risk or return, highlighting the role of ideological alignment in investment decisions. Our findings imply that managers derive ideological utility from value-consistent disclosure and are willing to bear the cost of investor polarization.