I am a PhD candidate in Finance at The University of British Columbia Sauder School of Business.

I am on the 2022-2023 academic job market.

I research human intermediation frictions in finance through AI intervention. I aim to understand the impacts of AI on information production and agency conflicts. I collaborate with a top InsurTech platform in China. Insurance advisors are the app users.

Curriculum Vitae

Research Interests:

  • Insurance, InsurTech, FinTech, Household Finance

  • Behavioral and Experimental Finance

  • Private Equity, Mergers and Acquisitions

References: Professors Jan Bena (Chair), Will Gornall, and Sabrina T. Howell

Email: xing.liu@sauder.ubc.ca

Working Papers

Picking Lemons? Algorithm-aided Human Decisions in Selection Markets: Evidence from Field Experiments on Insurance Agents (Job Market Paper)

  • Takeaway: Selling insurance is a multitasking environment where sales agents deal with both consumer demand and risk. By lowering the costs of information processing and changing the relative payoff from information acquisition around different tasks, AI-generated data on the target task can crowd out humans' attention allocation to the unassisted task, resulting in information loss.

  • Abstract: Human-intermediated sales channels are critical to finance and the broader economy. Sales agents make costly decisions about how to market and interact. One fundamental input for their decision making is information about consumer demand. I study how artificial intelligence (AI) demand predictions based on big data impact human-intermediated markets, by leveraging a large-scale randomized field experiment at a top insurance agency in China. In the experiment, the firm provided treated agents with an AI-based prediction of a consumer's demand for insurance, based on how the consumer had responded to advertising content on the largest Chinese social network platform. I show that AI demand predictions based on big data may facilitate cherry-picking for agents but fail to achieve lemon-dropping for insurers. Regarding cherry-picking, AI demand predictions shift agents’ attention to high-intent consumers, improving agents’ sales productivity by 14%. High-commission rate products are more likely to be sold to high-intent consumers, suggesting information-driven consumer discrimination. Regarding lemon-dropping, AI-based demand information reduces agents' own information acquisition and increases adverse selection, consistent with attention models and a crowding out of risk information. Supplemental survey and interview evidence confirm that agents value AI demand estimates, but information provision does not mediate agency conflicts around consumer risk. I generalize these findings to other financial products with high-touch sales channels.

  • Conferences: FSU Risk Management and Insurance Inaugural Research Symposium 2023; UGA PhD Student Symposium on Risk Management and Insurance 2023; AFBC 2022; Global AI Finance Research Conference 2022; Shanghai-Edinburgh Fintech Conference 2022; International Conference on FinTech and Digital Finance 2022; Webinar in Finance and Development (WEFIDEV) Fall 2022 (including scheduled)

Do Employees Cheer for Private Equity? The Heterogeneous Effects of Buyouts on Job Quality (with Will Gornall, Oleg Gredil, Sabrina T. Howell, and Jason Sockin), 2023.

  • Revise & Resubmit at Management Science

  • Conferences: EFA 2022; FIRS 2022; UCLA Fink 2022; PERC 2021; NBER Corporate Finance Fall 2021; PNWFC 2021

Work in Progress

Chioce Set Configuration for Complex Financial Products

Social Networks and Policy Transmission: Evidence from Insurance Agents on WeChat

Loss Aversion of Insurance Agents and Insurance Provision


Consolidating Product Lines via Mergers and Acquisitions: Evidence from the USPTO Trademark Data (with Po-Hsuan Hsu, Kai Li, and Hong Wu)

  • Journal of Financial and Quantitative Analysis, forthcoming.

The Role of Corporate Culture in Bad Times: Evidence from the COVID-19 Pandemic (with Kai Li, Feng Mai and Tengfei Zhang)

  • Journal of Financial and Quantitative Analysis, 2021, 56(7), 2545-2583.