Liyao WANG (王莉瑶)

Abstract: Post-earnings announcement drift (PEAD) displays presidential cycles: it earns 4.1% per year during Democratic presidencies but its profitability increases significantly to 15.3% during Republican presidencies. The tax component of firm earnings is significantly more volatile during Republican periods, suggesting larger tax policy uncertainty that amplifies information uncertainty and hence investor underreaction. Evidence from analyst forecasts, corporate filings, option implied volatilities and event studies helps to “rule-in” this novel tax channel. In addition, presidential cycles in PEAD are concentrated among firms with high tax policy exposures. The cycles are not driven by time-varying risk aversion or microcap firms.

Earnings Persistence Cliff and Post Earnings Announcement Drift (with Ming Zeng)

Abstract: We document a earnings persistence cliff between fiscal Q4 and the following Q1, leading investors to underestimate the persistence of Q1 earnings news and underreact to its implications. We develop a regime-switching framework in which investors infer the persistence of earnings news from past surprises. As a result, post-earnings announcement drift (PEAD) after Q1 announcements is about 83% larger than in other quarters, remains significant even after the mid-2000s, and cannot be explained by microcap firms, seasonality, earnings management, or existing PEAD theories.

Money Talks: Lending Relationships and Corporate Political Ideology (with Sumit Agarwal, Sibo Liu and Mark Yan)

CICF 2026, FMA 2023, Hawaii Accounting Research Conference (HARC) 2024

Abstract: This paper shows that banks influence borrowers' political donations by encouraging contributions to congressional banking committees that oversee bank regulators, particularly among financially constrained and high-risk borrowers. These donations are associated with more favorable loan terms and regulatory outcomes for banks, but also greater borrower covenant violations, increased bank risk-taking, and a higher likelihood of government bailouts, suggesting that bank-driven political influence contributes to regulatory capture and systemic financial risk.

Factor MAX and Predictable Factor Returns (with Ming Zeng)

Australasian Finance and Banking Conference 2025, Sydney Banking and Financial Stability Conference 2025, New Zealand Finance Meeting 2025

Abstract: This paper finds that factors with the highest recent maximum daily returns outperform those with the lowest by 0.32% per month, and this effect is distinct from factor momentum and stock-level lottery or momentum anomalies. The evidence suggests that limited investor attention leads to underreaction to extreme factor-level news, making the strategy especially profitable for low-attention factors or when extreme returns are unrelated to major macroeconomic or earnings announcements.

Equity and Bond Comovements: A Machine Learning Perspective (with Jiangyuan Li, Jinqiang Yang and Wei Zhou)

China International Risk Forum (CIRF) 2024, China Fintech Research Conference (CFTRC) 2025

Abstract: This paper uses machine learning to study stock–Treasury bond comovements, finding that machine learning models outperform traditional OLS in predicting correlations, with neural networks performing best among nonlinear methods. The results show that stock illiquidity is the primary driver of negative stock–Treasury correlations, while inflation-related measures drive positive correlations, indicating that positive comovement is linked to high inflation and negative comovement reflects cross-market hedging.

Lottery Preference and Factor Investing in the Chinese Market

CFRN Young Financial Scholars Annual Conference at Tsinghua University 2024

Abstract: This paper documents a strong factor MAX premium in China: factors with the highest recent maximum daily returns outperform those with the lowest by 0.82% per month on a risk-adjusted basis. The premium is robust across factor specifications, strongest for high-eigenvalue principal component factors and during periods of high investor sentiment, persists for up to 12 months, and is distinct from stock-level lottery anomalies while helping explain many of them.

Page updated

Google Sites

Report abuse