My research
I use unique data and settings to study relevant, real-world financial market phenomena with an emphasis on the use of financial market information by everyday people. My ongoing research studies how individuals interact with financial market information in various settings such as mobile phone trading apps, fake financial news, ESG portfolio concerns, politics, and sports gambling markets.
The Irrelevance of Environmental, Social, and Governance Disclosure to Retail Investors
Co-authors: Jim Naughton and Clare Wang
Forthcoming at Management Science
Media: altsforall.com, Investor Relations Magazine
Abstract: Using an hourly data set on retail investor individual security positions from Robinhood Markets, we find no evidence that environmental, social, and governance (ESG) disclosures inform retail investors’ buy and sell decisions. The response on ESG press release days by retail investors is indistinguishable from nonevent days. In contrast, these same investors make economically meaningful changes to their portfolios in response to non-ESG press releases, especially those that pertain to earnings announcements. We use stock return tests to show that there is economic content in ESG press releases, and we conduct subsample analyses showing that retail investors do not respond to the most salient and economically transparent ESG disclosures. Collectively, these tests suggest that a lack of economic content, a lack of visibility, and difficulty with investment integration are unlikely to explain our findings.
“The Role of Accounting Information in an Era of Fake News.”
Co-author: Betty Liu
Under Fourth Round Review at the Journal of Accounting & Economics
Abstract: We investigate the role of accounting information in deterring the production of financial fake news and attenuating its market impact. Specifically, for firms that have issued more management forecasts, provided a 10-K with higher readability, or released major accounting information more recently, we find an 8-14% reduction in the probability of being targeted by a fake article on Seeking Alpha and an 8-18% reduction in the number of fake articles. We also find a reduction in abnormal trade volume and idiosyncratic return volatility following the publication of a fake article for these firms. Furthermore, analyses using a bunching identification strategy find that fake news production peaks prior to earnings announcements and drops upon announcement, providing evidence that impending accounting information releases induce fake news authors to avoid publishing post-disclosure when the accounting information environment is relatively stronger. Overall, our results are consistent with accounting information reducing the production of and the market reaction to fake news, providing evidence of an ex ante and ex post role of accounting information in safeguarding firms from financial misinformation.
“How Do Brokerages’ Digital Engagement Practices Affect Retail Investor Information Processing and Trading.”
Job Market Paper
Abstract: I investigate how retail brokerages’ digital engagement practices (e.g., push notifications and content curation) impact retail investor information processing and trading. My identification strategy exploits a unique institutional feature of Robinhood in which it automatically sends push notifications to its customers when the intraday return of a stock in their portfolio reaches +/- 5%. Using an intraday event study design, I document that push notifications significantly increase retail investor trading by at least 25% in the fifteen minutes following notifications relative to non-retail investor trading. I then exploit the discontinuous increase in the proportion of retail investors trading on Robinhood with similar information sets to examine whether Robinhood’s content curation practices induce retail investors to incorporate earnings information in their trades. I find that retail investors trading on Robinhood after push notifications use earnings surprise information in their trades. Notably, Robinhood displays earnings information in a way that an investor’s visual perception of earnings surprise displayed on Robinhood differs from how accounting academics have typically transformed it into a value relevant information signal (i.e., scaling unexpected earnings by stock price). When I examine whether retail investors use the academic earnings surprise in their investment decisions, I consistently find that they do not. A difference-in-differences analysis around the introduction of earnings information on Robinhood as well as placebo analyses using Robinhood outages and an alternative notification threshold support these inferences. Lastly, I find that the influence of these digital engagement practices does not have a meaningful impact—positive nor negative—on aggregate retail investor informativeness.
“The Impact of Earnings Season on Election Day.”
Co-author: Clare Wang
Presentations: University of Iowa
Abstract: Due to the statutory timeline for general elections, U.S. elections always fall in the middle of the third-quarter earnings season. Our analyses exploit voter-level stock market ownership to provide evidence on the impact corporate earnings news has on voter perceptions and behavior. For the five most recent U.S. presidential elections, we show that earnings news released by S&P 500 firms prior to Election Day has a statistically and economically significant impact on the likelihood that stock-owning voters will cast an incumbent vote. We conclude that corporate earnings announcements serve as a timely and relevant signal influencing voter perceptions of economic performance at a critical time in the election cycle. In addition, our findings suggest that firms with a September 30th fiscal quarter-end may have a disproportionate impact on voters’ behavior on Election Day.
“A Corrupt NBA Referee and the Sports Betting Market: Quasi-Random Tests of Informed Trading and the Price Formation Process.”
Co-author: Dennis Hamilton
Abstract: We study price formation in the presence of market participants who hold a significant private information advantage using betting contracts on National Basketball Association (NBA) games. During the 2006-07 season, an NBA referee (Tim Donaghy) sold information about his games to a big-time bet broker who subsequently traded on the private information. Our identification strategy exploits that a referees’ game assignments are effectively random, resulting in the random assignment of an information-advantaged trader to the games Donaghy refereed. Using measures of betting market quality with analogous stock market counterparts, we find strong evidence that informed trading causes significant increases in long- and short-term price movements as well as decreases in measures of market liquidity. Our price informativeness evidence is less conclusive, suggesting either a slight increase in informativeness or no effect. Lastly, we find evidence consistent with the informed trader engaging in strategic price bluffing. The unique details of our setting allow us to: (1) directly observe informed trading and assets’ true, exogenous terminal values, (2) identify random variation in private information trading, and (3) overcome the joint hypothesis problem that plagues asset pricing research in other markets.