Michelangelo Rossi

🚨Free data alert!! 🚨 Please share.

Large new dataset of Amazon product reviews, including full text and photos and product characteristics, with individual *reviews labeled as fake reviews*.

I believe this is the first publicly available data of this kind.

github.com/bretthollenb...

YouTube Turns 20—How Have Its 20 Billion Videos Changed Us? BU faculty experts: the social media platform has impacted our mental health, helped small businesses, artists, and musicians, birthed a do-it-yourself generation, and exposed us to dangerous misinfor...

YouTube at 20! Very interesting post from Boston University #econsky
www.bu.edu/articles/202...

📢 Su Lavoce.info con miei coautori @kdbtran.bsky.social @micherossi.bsky.social and Mark Tremblay approfondiamo un tema cruciale per il mercato digitale (partendo da un nostro recente studio empirico): come la trasparenza nelle piattaforme influisce sui prezzi e sull’efficienza del mercato.

EC 2025 will be held at Stanford from July 7-12. Itai Ashlagi and I are the chairs. The abstract deadline is February 3, and the paper deadline is February 10. The scope is inclusive of many topics across CS, economics, and operations research. Submit your best work!

My first post on Bluesky! Excited to share my paper with Yixing Chen and Xiaoxia Lei "Trade-offs in Leveraging External Data Capabilities: Evidence from a Field Experiment in an Online Search Market" has been accepted for publication in Management Science! papers.ssrn.com/sol3/papers....

If only someone had recently published a WP on the impacts of price transparency (at least on peer-to-peer platforms)! @leonardomadio.bsky.social @micherossi.bsky.social
papers.ssrn.com/sol3/papers....

Holding the supply side fixed, transparency helps consumers and could also reduce overall prices, but if sellers change prices due to the policy, things can get tricky ...

Digital Economics Paris - Conference 2025 Paris Conference on Digital Economics

⏳ Only 3 days left to submit your paper for the Paris Conference. 🗼✨ www.digitaleconomics-paris.fr/conference-2...

Thanks! I agree: price increases are not necessarily welfare decreasing (especially for sellers). Moreover, it is hard to measure the reduction of guests’ search costs with more transparency… so any welfare analysis is very complex in this framework

Curious to learn more?☝️

Check out the full paper: papers.ssrn.com/sol3/papers....

We uncover a fascinating mechanism linking transparency to pricing strategies in digital markets. Feedback and thoughts are welcome!

Policy implications:

Price transparency isn’t universally good or bad. While it reduces search costs and obfuscation, it can also lead to price increases in some cases. Regulation needs to consider both demand-side and supply-side effects carefully.

The core insight: transparency changes how hosts (not just guests) look at prices. When rivals’ total prices are clearer, hosts adjust their strategies. This is especially impactful in peer-to-peer platforms where pricing frictions exist and some hosts and guests might be naive or have search costs.

Key results:

• Listings with high cleaning fees reduced them by 2-4% post-policy.⬇️

• But listings without cleaning fees raised their nightly prices by 5-6%.⬆️

Why? Greater transparency let some hosts realize their prices were too low, prompting increases.

Airbnb hosts set nightly prices and cleaning fees. Before 2019, EU guests only saw cleaning fees at checkout. After a regulatory push, Airbnb made these fees visible upfront. Using a difference-in-differences approach, we studied the impact on prices and fees.

How does price transparency affect a market?🤔

In our new paper, Kevin Tran, @leonardomadio.bsky.social, Mark J. Tremblay, and I analyze Airbnb’s policy change in the EU, where cleaning fees became fully transparent. The surprising finding: transparency doesn’t lead to lower prices. Here’s why:

Thanks a lot, Brett!

Here is the link: pubsonline.informs.org/doi/10.1287/... (15/15)

📚 Overall, our work sheds new light on how consumer heterogeneity shapes online ratings and offers practical solutions to improve rating systems. We’re excited to see it published in Management Science! (14/15)
#Ecommerce #Ratings #ConsumerBehavior #ManagementScience #DataScience #IMDb #MovieLens

📊 Our findings have important implications for platform design. By understanding these biases and applying corrections, platforms can deliver more reliable ratings, benefiting consumers and (high-quality) producers alike. (13/15)

💡 Conversely, simply overweighting the ratings of experienced users, a common practice on several platforms, can actually backfire, further penalizing high-quality movies. (12/15)

Yes! Once debiasing the ratings, this movie’s rating goes up! In particular, this movie is one of the biggest “winners” of our debiasing procedure. (11/15)

🎞️ A notable case of this bias? The French movie "The Unknown Girl", selected for the Palme d’Or Cannes in 2016, is rated 6.5 on IMDb. That’s relatively low… but is it due to the fact that most of the raters were experienced, stringent users? (10/15)

After applying it, the corrected ratings better align with external measures of quality, such as the Oscars and Metacritic scores. It also helped fix those ranking reversals! (9/15)

🔄 However pervasive, this bias can be undone. We developed an algorithm to de-bias ratings by adjusting for user stringency. Our approach doesn’t require us to take a stance on users’ expertise. Rather, we let ratings and individual stringencies iterate until they converge. (8/15)

📉 It gets worse. Ratings need not just be compressed, they can actually lead to ranking reversals: in about 8% of cases, lower-quality movies get higher ratings than better ones due to this biased. This further skews future consumer choices! (7/15)

And since experienced users’ ratings represent a higher share of ratings for higher quality products… IMDb ratings are compressed, that is, they penalize high-quality films compared to their lower quality alternatives. That’s the exact opposite of what we’d like IMDb to do! (6/15)

A 7 out of 10 from a user with 10000 ratings is harder to obtain than one from a user with 5 ratings! Absent a normalization, we’re comparing apples with oranges. (5/15)

In other words: experienced users choose better movies on average ➡️ they get used to higher quality, and form higher reference points ➡️ they rate more harshly, for any quality level. (4/15)

How about ratings' differences between experienced and novice users? This is where it gets interesting. Experience users rate virtually ALL movies more harshly, independent of genre, year, quality, actors, director, and more. (3/15)