题　目：Coarse Ratings on Online Platforms

嘉　宾：Si Zuo (左思), Ph.D. Candidate, Cornell University

主持人：张铄 副教授 BAT365唯一官网

时　间：2024年10月8日（周二）14:30-16:00

地　点：BAT365唯一官网 徐汇校区 安泰经济与管理学院A305

内容简介:

Ratings are widely used on online platforms, but there is huge variation in the coarseness levels of ratings displayed on platforms. For example, Airbnb displays ratings to two decimal places (granular rating), while Amazon only displays half stars (coarse rating). A natural question is: is there an optimal coarseness level of ratings on a platform? In this project, I try to answer this question using both synthetic simulations and observational data. In the first half of the project, I develop a dynamic model to illustrate the fairness-efficiency trade-off behind coarse ratings: coarse ratings could improve fairness on the platform by helping entrants stay longer, thus improving the quality of products available on the platform. On the other hand, coarse ratings hide information from consumers, naturally leading to an information loss. I show that when the fairness force dominates, coarse ratings improve total welfare and platform revenue. In addition, my results suggest there is an inverse U-shaped relation between rating coarseness level and platform revenue. I find markets with more firms or higher quality dispersion require more granular rating systems. In the second half of the project, I use Airbnb 2015-2019 NYC data for the counterfactual simulation exercise to determine the optimal coarseness of ratings on Airbnb. I find that Airbnb’s platform revenue could increase by 4.0% if Airbnb changes from the current 2-decimal place rating (4.83, 4.93) to a one-decimal place rating system (4.8, 4.9). My project provides new insights into platform rating design and offers guidance on how platforms should choose the optimal coarseness level for ratings.

演讲人简介：

Si Zuo is a 6th year PhD Candidate in Economics, SC Johnson Graduate School of Management & Economics Department, Cornell University. She is interested in empirical IO, quantitative marketing, platforms, recommendation systems, and rating algorithms. In her research, she incorporates multiple research methods, including causal inference, structural modelling, machine learning, reinforcement learning, and game theory.  Her research mainly focuses on two streams: The first stream of her research studies how ecommerce platform design affects their profits; and the other studies the impacts of AI and digitalization.

欢迎广大师生参加!