Working Paper
- “Data Preferences in Firm Learning: Evidence from an Online Auction Platform” (JMP)
- Draft available upon request
Presentations (*: Scheduled)
- 2026: BIOMS*, TPM*, Marketing Science Conference*, EMAC*
- 2025: IIOC (Rising Star Session), China-India Insights Conference (Plenary Session)
- 2024: Fordham University, Boston College, the Chinese University of Hong Kong - Shenzhen, Tongji University, Shanghai University of Finance and Economics, Shanghai Jiao Tong University, the University of Hong Kong, the University of New South Wales
Abstract
Governmental organizations and industry practitioners have promoted data sharing across firms to expedite learning to improve business decisions. However, current discussions have largely overlooked the possibility that firms may prefer their own data over others' data. This paper investigates the presence of such biased preferences among firms, focusing on used-car auction sellers on China's largest online auction platform. Auction timing is crucial on this platform as payoffs vary by hour, influenced by bidders' valuations and the number of residual bidders at different times. Despite being experienced local sellers before joining Ali Auction, these sellers face national demand and competition in the online environment, creating the scope for learning. Combining state-of-the-art auction literature, I develop a structural model of sellers' learning based on their own and others' data to optimize auction timing. The model estimates suggest that sellers' preferences for different data sources change with experience, with sellers weighing their own data at 90% compared to 10% for others' data at the average level of experience. Moreover, the data preferences account for more than half of the revenue gap between the status quo and the full information scenario. These findings have two implications for the platform. First, data sharing alone may not effectively guide sellers in selecting optimal auction timing. Second, the platform can leverage sellers' data preferences to guide new sellers to optimal timing early in their tenure, ensuring lasting benefits. Overall, the platform should play a coordinating role in helping sellers identify the best timing for their auctions. - “Shaping the Influencers: The Role of Multi-Channel Networks” [SSRN]
- with Yulin Hao
- Under 2nd Round Review at Journal of Marketing Research
Presentations (*: Scheduled)
- 2025: Marketing Science Conference
- 2023: EARIE
Abstract
Social media influencers increasingly affiliate with multi-channel networks (MCNs), also known as influencer agencies. However, little evidence exists on how these organizations affect influencers' content and monetization outcomes. Do MCNs create value merely by improving brand–influencer matching, or also by providing production resources that shape influencer content? We answer this question using a unique dataset from TikTok in China that tracks influencers' MCN affiliation changes over time. Exploiting staggered entry into and exit from MCNs in a difference-in-differences framework, we use not-yet-treated influencers (later treated) as the control group to identify the effects of MCN affiliation. Entering an MCN increases per-post views by 21%, while exiting reduces them by only 5%, consistent with influencers retaining portable production skills acquired during affiliation. Following entry, influencers shift their content toward the MCN's topic domain and toward more commercially valuable topics, and sponsorships rise gradually. Heterogeneity analyses further show that engagement gains are concentrated among influencers whose pre-affiliation content aligns with the MCN's specialty, and are larger for influencers joining smaller, more topically and geographically concentrated MCNs. These findings are difficult to reconcile with pure intermediation alone and are consistent with MCNs also providing production support.
Work in Progress
- “Spatial Matching Frictions and User Incentives in Urban Mobility Networks”
- “Discount Design of Store Credit Cards”
- with Bowen Luo and Ruiqi Wu
- Data access through collaborating with Wharton AI & Analytics for Business (AIAB)