Siqi (Thinkling) Li

Research

Job Market Paper

Measuring the Market Impact of Machine Learning Valuations in Auctions (The paper is available upon request.)

In this study, I analyzed 2.18 million randomized experiment auction sale records from GoDaddy.com to assess the market impact of machine learning valuations. Through comprehensive reduced-form analysis and structural modeling, I measured shifts in market behavior and improved prediction accuracy by 20% over baseline models. The research involved building a two-step structural model to estimate unobserved optimal bids and expected revenue.

Other Research Projects

• BLP Model in the Case with Large Number of Goods and Markets under Overlapping Simulation Draws
  Derived the asymptotic distribution of the BLP estimator to estimate demand curves, providing an analytical standard error formula. This new method significantly reduces computational costs by requiring fewer simulation draws than previous literature. Demonstrated the applicability of theoretical results using web-scraped large video game sales data.
• Migration Experience Has Limited Negative Effect on Satisfaction: Evidence from China
  Integrated a panel data set of 2,000 children and family members over 15 years to analyze the relationship between the life satisfaction of rural young people and their migration experiences. Enhanced prediction accuracy by 20% over baseline models through variable selection and model optimization with fixed-effect models.