Xin Huang
Research Subjects
Xin Huang has a total of 5 co-authors. Here are the top co-authors by number of shared publications:
- Lihong Zhang
School of Economics and Management, Tsinghua University, Beijing 100084, China
2 shared publications Chenyu Wang
1 shared publication- Jian Tang1 shared publication
Lining Huo
1 shared publicationLvxin Chen
1 shared publication
Latest Publications
- Open Access
- 11 August 2026
- Published by Elsevier in SSRN Electronic Journal
The Rational Market Reaction of AI Adoption:Evidence from Earnings Announcements and Social Media
Using 14,806 firm-year observations of Chinese A-share firms from 2017 to2024, we examine whether AI disclosure changes investors’ interpretation of earnings news. Because the economic prospects of AI applications vary substantially across firms and industries, investors face considerable ex ante uncertainty and may initially rely on broad narratives and social-media discussions to form noisy prior beliefs. Earnings announcements provide firm-specific information that allows investors to refine these beliefs. We find that AI disclosure significantly weakens the post-announcement return response to positive earnings surprises, robust to alternative return, earnings, and AI measures. This differential valuation response is weaker when firms possess stronger implementation capacity and is concentrated among firms whose fundamentals subsequently deteriorate, while disappearing among firms whose performance improves. The effect is not driven by AI washing or a reversal of prior optimism. Social-media sentiment provides an observableproxy for investors’ prior beliefs, and order-imbalance tests confirm thatthese beliefs are reflected in actual trading. Overall, the evidence is consistent with a rational, forward-looking reassessment of heterogeneous AI application prospects.
- 11 August 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessThe Rational Market Reaction of AI Adoption:Evidence from Earnings Announcements and Social Media
Using 14,806 firm-year observations of Chinese A-share firms from 2017 to2024, we examine whether AI disclosure changes investors’ interpretation of earnings news. Because the economic prospects of AI applications vary substantially across firms and industries, investors face considerable ex ante uncertainty and may initially rely on broad narratives and social-media discussions to form noisy prior beliefs. Earnings announcements provide firm-specific information that allows investors to refine these beliefs. We find that AI disclosure significantly weakens the post-announcement return response to positive earnings surprises, robust to alternative return, earnings, and AI measures. This differential valuation response is weaker when firms possess stronger implementation capacity and is concentrated among firms whose fundamentals subsequently deteriorate, while disappearing among firms whose performance improves. The effect is not driven by AI washing or a reversal of prior optimism. Social-media sentiment provides an observableproxy for investors’ prior beliefs, and order-imbalance tests confirm thatthese beliefs are reflected in actual trading. Overall, the evidence is consistent with a rational, forward-looking reassessment of heterogeneous AI application prospects.
- Open Access
- 9 June 2026
- Published by Elsevier in SSRN Electronic Journal
Empirical Performance of Long-Run Risk and Volatility Ambiguity: International Evidence
This paper empirically compares the long-run risk model and the volatility ambiguity model under Epstein-Zin preferences in international equity markets using quarterly data. Applying the dark-matter measure of model fragility by H. Chen, Dou, and Kogan (2024), we find that the volatility ambiguity model yields values between 0.045 and 0.242, while the long-run risk model yields values between 1.700 and 29.871. The volatility ambiguity model also requires more reasonable relative risk aversion coefficients to match the equity premium. Furthermore, the ambiguity model fits equity premiums, volatilities, and risk-free rates better, whereas the long-run risk model performs better in fitting logarithmic price-dividend ratios.
- 9 June 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessEmpirical Performance of Long-Run Risk and Volatility Ambiguity: International Evidence
This paper empirically compares the long-run risk model and the volatility ambiguity model under Epstein-Zin preferences in international equity markets using quarterly data. Applying the dark-matter measure of model fragility by H. Chen, Dou, and Kogan (2024), we find that the volatility ambiguity model yields values between 0.045 and 0.242, while the long-run risk model yields values between 1.700 and 29.871. The volatility ambiguity model also requires more reasonable relative risk aversion coefficients to match the equity premium. Furthermore, the ambiguity model fits equity premiums, volatilities, and risk-free rates better, whereas the long-run risk model performs better in fitting logarithmic price-dividend ratios.