Do Investors Fairly Value Private Startup Securities? Evidence from Mutual Funds, with Vikas Agarwal, Brad Barber, Si Cheng, Harshini Shankar, Ayako Yasuda, 2026
Abstract: Mutual funds overvalue their junior stakes in startups by 53% compared to fair value models that account for multi-tier capital structures. Junior securities are marked close to senior securities, despite the latter being worth 58% more per share. Mutual funds overpay in secondary purchases relative to model fair values. Overvaluation is not justified by ex-post exit outcomes and does not decline over time or vary with VC boom-and-bust cycles. Mutual funds’ valuation updates are more correlated with model fair values after a down round. The findings suggest mutual funds underestimate downside risks for junior securities while overweighting IPO exit scenarios.
Information Transmission from Stock and Bond Markets, with Sheridan Titman, Jason Wei and Huiping Zhang, 2026. Winner of Asian Finance Assn Conference Best Paper Award, 2024.
Abstract: The returns of firms’ corporate bonds predict their future stock returns. Predictability is especially strong when the bonds have low credit ratings and the stocks have lottery-like characteristics that attract retail traders. We also find that predictability is strongest when bond volume is more institutional and is high relative to stock volume. The past month’s bond returns strongly predicts stock returns in the first week of the subsequent month. The predictability in subsequent weeks is weaker, but remains economically meaningful, consistent with gradual information diffusion. Stronger predictability around earnings announcements supports the view that bond markets reflect firm-specific cash flow information rather than changes in risk premia.
The Dark Side of Geographically Dispersed Information: Evidence from Lockdown of Subsidiaries, with Massimo Massa, Zhenghui Ni, and Zhou Zhou, 2025.
Abstract: We study the impact of geographically dispersed information on the firm’s information environment using a natural experiment: the pandemic-induced localized lockdowns of subsidiaries. We find that restricting access to local subsidiary-level information improves the firm’s overall information efficiency. Specifically, the lockdowns improve forecasts of analysts located near the affected subsidiaries by reducing biases associated with local subsidiary-level information. Analysts located near subsidiaries underreact to new firm information, highlighting a key mechanism driving the proximity bias. This finding extends to enhanced accuracy of firm level analyst forecasts and improvements in stock price informativeness. Furthermore, the reduction in information asymmetry following lockdowns erodes insiders’ informational advantage, and enhance stock liquidity. Contrary to the traditional view that geographic proximity to firm operations benefits the information environment, our study demonstrates that proximity to subsidiaries can distort information processing, offering a novel perspective on the role of geographically dispersed information in financial markets.
Dividend Timing and Global Dividend Premium, with Jing Xie and Yuxiang Zhong, 2026. Winner of PBJF Best Paper Award, 2024.
Abstract: Using data from 44 international equity markets, we document a robust dividend premium of 0.58% per month after controlling for global and regional risk factors. We decompose this premium into a timing-related component and a persistent component. The timing-related component arises in predictable dividend months: dividend payers earn return run-ups around clustered ex-dividend dates that only partially reverse, leaving a residual that does not wash out. Within markets, the premium is stronger when ex-dividend dates are more concentrated. Across markets, dividend premia co-move more strongly when payout calendars overlap, consistent with synchronized cross-market demand shocks. The persistent component survives outside dividend months and is larger in weaker institutional environments (weaker investor protection, poorer disclosure and securities regulation, and lower liquidity). Together, the evidence shows that predictable payout timing generates both return premia and international return co-movement.
The Signalling Role of Earnings Consistency in Conglomerates, with Massimo Massa and Zhenghui Ni, 2026
Abstract: We study the signalling role of the internal earnings consistency (“EC”) of segment-level earnings in conglomerates. We argue EC signals managerial ability from the coherence of segment results. We document that EC amplifies the negative relationship between long-term expected earnings growth (LTG) and future stock returns: high-LTG conglomerates with high EC experience stronger underperformance, with an annual alpha of -10%. Difference-in-differences analyses around SFAS 131 confirm that EC does not affect earnings forecasts by analysts, while it impacts mutual fund portfolio investment. Our findings highlight how investor reliance on perceived signal coherence, rather than fundamentals, drives mispricing in complex firms.
Slow Trading and Stock Return Predictability, with Matthijs Lof and Matti Suominen, 2022.
Abstract: The state of market returns positively predicts the size premium (or the difference in the return on small and large firms) as small stocks adjust to market returns with a delay and large firms revert following market returns. This predictability of the size premium is strongest when aggregate asset and funding liquidity is low and is linked to institutional and informational frictions that manifest as slow institutional trading in small stocks but swift trading in large stocks. For example, slow trading by mutual funds leads to predictable small stock returns in the direction of fund flows.