Associate Professor of Finance (without tenure) at Saïd Business School
University of Oxford
Publications and Accepted Papers
Journal of Financial Economics, Accepted
​
JFI/FIRS Best Doctoral Paper Award (2022)
AsianFA Best Doctoral Paper Award (2022)
EEA/UniCredit Econ Job Market Best Paper Runner-Up Award (2022)
PIIRS Graduate Student Research Award (2022)
​
Abstract: This paper investigates how cashless payment affects credit access for underserved populations using data from Alipay, a leading Chinese BigTech platform with over 1 billion users that offers a wide range of financial services. By exploiting the staggered rollout of Alipay-bundled shared bikes across cities as a natural experiment and analyzing a representative Alipay user sample, I find that cashless payment adoption increases credit access by 56.3% and that a 1% rise in payment flow increases credit lines by 0.41%. These effects are stronger for less educated and older individuals, who have traditionally faced greater barriers to accessing financial services.
Working Papers
with Long Chen, Yadong Huang and Wei Xiong
​
Media Coverage: VoxEU, VoxChina
Abstract: We integrate survey and behavioral data from Alipay to analyze how users’ data sharing with third-party mini-programs relates to their privacy concerns, digital engagement, and credit access. Paradoxically, users with stronger privacy concerns do not share less data, exemplifying the data privacy paradox. We resolve this paradox as an omitted variable issue: users with stronger privacy concerns also exhibit a higher demand for digital services offered by mini-programs. Furthermore, our analysis reveals that increased engagement with digital services amplifies privacy concerns. Nevertheless, data sharing not only facilitates access to these services but also enhances credit access.
AI as Decision-Maker: Ethics and Risk Preferences of LLMs [SSRN Link] [arXiv Link] [pdf]
with Hayong Yun and Xingjian Zheng
Abstract: Large Language Models (LLMs) exhibit diverse and stable risk preferences in economic decision tasks, yet the drivers of this variation are unclear. Studying 50 LLMs, we show that alignment tuning for harmlessness, helpfulness and honesty systematically increases risk aversion. A ten percent increase in ethics scores reduces risk appetite by two to eight percent. This induced caution persists against prompts and affects economic forecasts. Alignment therefore promotes safety but can dampen valuable risk taking, revealing a tradeoff risking suboptimal economic outcomes. Our framework provides an adaptable and enduring benchmark for tracking model risk preferences and this emerging tradeoff.
with Jing Huang
Abstract: Designing effective prompts is challenging when seeking advice from large language models (LLMs) on tasks involving users’ soft traits. We introduce preference uncertainty—capturing soft information—into a cheap talk framework (Crawford and Sobel, 1982) and model soft information communication with AI as the investor’s optimal stopping problem with Brownian information flow, which we solve in closed form. Although LLMs are not subject to misaligned incentives, soft information communication is inefficient due to inevitable losses from digitization and LLMs’ limited memory. The model predicts that an investor generally prefers LLMs trained to be more “opinionated” than her own prior, except when she is most confused and prefers an aligned and equally confused LLM. We validate model predictions through LLM-driven simulations: investor profiles are simulated based on the Survey of Consumer Finances, and multi-round LLM advising simulations, benchmarked against standard portfolio questionnaires, confirm our theoretical predictions.
The Fixed Disposition Effect [SSRN Link] [pdf] [Google: fixed disposition effect]
with Qinglin Ouyang
Abstract: We revisit the disposition effect and argue that it is best understood not as a primitive behavioral bias, but as a reduced-form outcome of stable investment styles. Using a unique inter-linked dataset that combines a large-scale experiment with real-world mutual fund transactions, we document strong within-investor persistence in disposition behavior across time and contexts. This persistence is largely driven by a fixed investment style: contrarian investors exhibit a substantially stronger disposition effect, while it is minimal for momentum investors. Investment style explains far more variation in the disposition effect than standard demographic and socioeconomic characteristics. By contrast, realization preference is generally shared. We provide some of the first field evidence that it accounts for roughly 10% of the bias via a sharp jump at the zero-return threshold. Overall, our findings suggest that the disposition effect often emerges as a structural outcome of price-based trading rules, rather than a generic behavioral bias.
How Conversational Generative AI Reshapes User Decision Processes [SSRN Link]
with Yue Guo, Siliang Tong, Tao Chen, and Subodha Kumar
​​
Information Systems Research, R&R
​
Abstract: As generative AI increasingly enables conversational interaction, fintech platforms are rapidly adopting conversational generative AI agents to support complex financial decisions. Despite growing practical interest and rapid deployment, existing research provides limited insight into how and why users respond to conversational generative AI in high-stakes decision contexts, how such responses unfold through concrete decision processes, and whether such systems fundamentally reshape users’ underlying decision processes rather than merely improving observable economic outcomes. Drawing on a large-scale field study conducted in collaboration with Ant Fortune, China’s leading fintech platform, we examine how the use of a conversational generative AI–based decision-support system (GAIC) reshapes users’ decision processes. Integrating the exploration–exploitation framework with behavioral bias theory, we develop a mechanism-based perspective that conceptualizes GAIC usage as a decision-support technology that alters how users search for information, reassess existing choices, and act under uncertainty.
with Ravi Jagannathan and Jiaheng Yu
​​
Review of Finance, R&R
​
Abstract: We measure the aggregate lifetime return to equity investments in venture companies. We consider the aggregate portfolio of all equity investments made in various funding rounds of 17,242 ventures that had their first funding round between 1980-2006. We track these ventures till their exits or till 2018 to get a complete picture of their lifetime return. The Kaplan and Schoar (2005) Public Market Equivalent (PME) measure of performance is 1.42, i.e., each dol- lar invested returned 1.42 dollars after adjusting for risk and time value. We use an imputation model following Hall and Woodward (2010) to address missing data issues.
​​
Abstract: This paper examines how individual investors respond to the market price fluctuations, using unique individual-level transaction data from a trading experiment and the same individuals’ real trading history on the Alipay app. We find that, in response to exogenous price movements in the experiment, investors tend to be contrarian traders. The magnitude of investors’ response is asymmetric in downturn and upturn markets. Sophisticated investors tend to be more contrarian than the less sophisticated ones. We further document that investors’ contrarian styles are persistent in the experiment and real transactions. The results imply that investors use simple heuristics from the price movement when they make investment decisions in the real world.
Invited Articles and Non-Refereed Publications
(Generative) AI in Financial Economics [SSRN Link] [JCEBS Link] [pdf] [Google: ai fin econ]
with Hongwei Mo
​​
This review is updated monthly with the latest high-quality research on AI and finance.
This version: February 2026. First version: May 2025.
​
Abstract: This review article synthesizes the burgeoning literature on the intersection of (generative) artificial intelligence (AI) and finance. We organize our review around six key areas: (1) the emergent role of generative AI, especially large language models (LLMs), as analytic tools, external shocks to the economy, and autonomous economic agents; (2) corporate finance, focusing on how firms respond to and benefit from AI; (3) asset pricing, examining how AI brings novel methodologies for return predictability, stochastic discount factor estimation, and investment; (4) household finance, investigating how AI promotes financial inclusion and improves financial services; (5) labor economics, analyzing AI’s impact on labor market dynamics; and (6) the risks and challenges associated with AI in financial markets. We conclude by identifying unanswered questions and discussing promising avenues for future research.
Venture Investment Returns [Link]
with Ravi Jagannathan and Honghao Wang
The Palgrave Encyclopedia of Private Equity, 2023
Editors: Douglas Cumming, Benjamin Hammer