Behavioural & Experimental EconomicsBehavioural FinanceSustainable FinanceEconomics of AI & Technology Adoption
About
I am a PhD candidate in Finance at the
Utrecht University School of Economics,
supervised by Dr. Yilong Xu, Dr. Paul van Bruggen, and Prof. Kees Koedijk.
My research sits at the intersection of behavioural economics, finance, and emerging technology.
I study how people form beliefs and make decisions in the face of generative AI,
sustainability trade-offs, and fairness dilemmas. My work combines controlled laboratory
and online experiments with structural inference to generate evidence that speaks to both
theory and policy.
Before joining Utrecht, I completed an MSc in Behavioural Economics at the China Center for
Behavioral Economics and Finance, Southwestern University of Finance and Economics (SWUFE),
and a double bachelor's degree in Management and Economics, also at SWUFE.
Research
Working Papers
Misperceiving the Frontier: Managerial Beliefs and the Allocation of AI
Generative AI is a prominent example of a broader challenge firms face when adopting new technologies:
their benefits are often uneven across tasks and workers, exhibiting a "jagged technological frontier."
We study whether managers anticipate that the benefits of new technologies are uneven across tasks and
workers, and how the information environment affects deployment. In a linked worker–manager experiment,
workers complete real-effort tasks with and without generative AI, allowing us to estimate AI effects by
task and worker type. Managers then predict AI-assisted performance and make incentivized deployment
decisions. Without feedback, most managers mistakenly expect AI to improve performance on both sides of
the frontier. Performance feedback reduces average optimism, but substantial overestimation remains for
outside-frontier tasks. We further examine incentive-compatible deployment decisions for two new tasks
(one inside and one outside the frontier). Managers' willingness to pay (WTP) is high across all tasks,
and even higher for the outside-frontier task where AI does not add value. An information-design
intervention in the Feedback condition that makes salient which tasks are inside or outside the frontier
reduces WTP for outside-frontier deployment relative to inside-frontier deployment. The findings suggest
that realizing organizational gains from new technologies depends not only on technical capability, but
on information environments that make the limits of those capabilities visible to decision-makers.
Collective Evidence on Behavioral Interventions Targeting Carbon Pricing Support
with Yilong Xu and Juergen Huber et al.
Revise & Resubmit — Nature Human Behaviour
Carbon pricing is widely regarded as an effective and cost-efficient climate policy, yet public support
remains limited. This crowdsourced "many-design" project evaluates 55 behavioral interventions designed
by independent research teams and implemented simultaneously with nearly 20,000 U.S. residents. Across
interventions, effects on support for carbon pricing are positive but small, increasing support by
roughly two percentage points. The study also finds modest between-study heterogeneity and substantial
overconfidence among researchers regarding intervention effectiveness.
Note: As one of the participating teams, Yilong Xu and I are
interested in the following question: although carbon taxes benefit most households through rebates,
public support remains low, in part because people poorly understand how rebates are distributed. We
contribute by conducting an experiment with two treatments testing whether detailed information about
the costs and benefits of a carbon tax, especially the rebate distribution, increases stated support for
the policy and the likelihood of choosing the green product in an incentivized choice task.
Work in Progress
Beyond Returns: Probing the Resilience of Sustainability Preferences
with Peiran Jiao, Kees Koedijk, and Yilong Xu
A central debate in sustainable finance is whether investors will sacrifice financial performance to
pursue sustainability. Existing studies focus on expected performance. But what happens when the true
financial costs are actually felt? To address this, we conduct an experiment where participants allocate
between two real equity funds, one sustainable and one conventional, while the sustainable fund
persistently underperforms. Using a 2×2 design, we independently vary sustainability labeling and
information format (static description versus an interactive Demo Account). Labeling raises allocations
to the underperforming sustainable fund by 5 to 10 percentage points, and this gap persists in a
surprise investment opportunity after an extended period of realized losses. Differences in beliefs do
not explain this resilience: even though participants expect the sustainable fund to underperform, they
still allocate more to it. The Demo Account lets investors discover their true preferences before
committing real money. It reduces early overcommitment to the low-performing sustainable fund. For
investors with weak environmental values, they learn that the cost of sustainability outweighs its value
to them and pull back, whereas strongly value-driven investors may pull back at first but revert once
real stakes confirm they will bear the cost.
Sustainable Decisions Under the Veil of Ignorance
with Oliver Hauser and Yilong Xu
Global efforts to combat climate change are often hindered by conflicting national interests and economic
disparities, making fair resource allocation a persistent challenge. The Veil of Ignorance, a framework
proposed by John Rawls, elicits individuals' impartial fairness preferences by removing personal biases.
This study applies the Veil of Ignorance in an experimental setting to examine how individuals allocate
resources across generations and productivity levels, and assesses its impact on fairness in
decision-making. Results show that when participants know their assigned roles, they allocate resources
in a self-serving manner. In contrast, participants who make decisions without knowledge of their roles
allocate resources more equally across periods and positions, aligning more closely with fairness
principles. These findings suggest that differences in allocation behavior arise not from disagreements
over fairness norms but from strategic deviations when self-interest is at stake.
Conferences & Seminars
2026
Sep 2026
2026 European Meeting of the Economic Science Association (ESA) (Scheduled)
Supervised three students in 2026 and one student in 2025 on topics including board gender
diversity, green-investor sentiment, and AI-sector valuations.
