1. Discussant: Wen Wen Univ. of Texas at Austin
Review Rating-based Platform Screening and New Complementor Entry: Evidence from Natural Experiment and Machine Learning of a Sharing Economy
By Xueming Luo, Zhijie Lin, and Tedi Skiti
Discussant: Wen Wen
Univ. of Texas at Austin

2. Summary This study examines: Major findings:
A platform design that ranks incumbent complementors based on review ratings
How does such design affect new complementor entry and the overall quality of complementors?
Major findings:
The rating-based platform screening
reduces new complementor entry; has stronger effect when incumbent complementors have higher ratings
increases complementors’ overall quality

3. Key Features Undoubtedly a fascinating topic!
A variety of mechanisms are used by platforms to reduce information asymmetry and user search cost (the demand side), but little understanding on their effects on new complementors (the supply side)

4. Key Features Impressive data & interesting empirical setting!
Detailed complementor-level data on a home-cooked food delivery platform (2 million+ obs., 9k+ complementors)
A platform-wide change on complementor ranking during the sample period
Before: recommendation not based on rating
After: recommendation based on rating (higher on the top)

5. Comments and Suggestions
Central hypothesis: A rating-based ranking system reduces new complementor entry
What are the underlying mechanisms?
Makes it harder for users to find new complementors, thereby discouraging new entry
Increases competition among existing complementors, thereby discouraging new entry
Should we also consider the demand effect?
The new ranking mechanism reduces user search cost and increases demand, which may encourage new entry

6. Comments and Suggestions
A great opportunity to paint a full picture of complementor responses to a new platform screening design
Current focus: new complementor entry
Why not also look at:
Whether it has different effects on the entry of high-quality vs. low-quality new complementors
How it affects the exit rate and survival chance of incumbent complementors
How it changes the competitive strategies (e.g., price/quality differentiation) by incumbent and new complementors

7. Comments and Suggestions
The empirics
Current baseline analysis: complementor level
Alternative analysis:
At the district level (16 districts)
At the district-product category level (assuming the competition mostly happens within each product category)
# of new complementors located in complementor i’s district
platform screening change dummy

8. Comments and Suggestions
The empirics
May need to address concerns such as
It may pick up some general time trend
The platform chose to implement the new screening system because of a reduction in new complementor entry
Identify a control group for diff-in-diff estimation?
Using data from a competing food delivery platform
# of new complementors located in complementor i’s district
platform screening change dummy

9. Very Promising Research!
Very interesting research question
Unique and rich data
Exogenous platform design change offers potential for compelling identification