P.S. Those in literature may like our summaries of the 56 papers we found. 12/12

🔗Find the full paper at jhall.github.io/documents/Va... 11/12

Takeaway #2: Average effects are small. General statements about the impact of ride-hailing on transit ridership, traffic congestion, and traffic fatalities should reflect that the average estimate is close to zero and not statistically significant. 10/12

Takeaway #1: Context matters. Dense NY or London ≠ mid‑sized Vancouver. But for many US/Canadian metros, long‑run ride‑hailing impacts on transit & congestion are likely modest. 9/12

Publication bias? Surprisingly little. 8/12

📚🔍To put these results in context, we then conducted a meta-analysis. We screened 2,597 papers, kept 56. Across modes & countries, lots of heterogeneity, but average estimates hover near zero; ~⅓ of papers find no significant effect (but boosts🚊and depresses 🚍). 7/12

⚠️Traffic fatalities: noisy data give wide CI [‑30%, +199%]. We can’t rule much in or out. 6/12

🚗Traffic congestion: point estimates slightly ↑ but never statistically significant; CI [‑6.1%, +20.7%]. No evidence that “Uber makes gridlock explode” 5/12

We use a weighted average of US/CA cities to build a synthetic Vancouver that has Uber to estimate how ride-hailing impacts transit ridership, traffic congestion, and traffic fatalities. What do we find? 3/12

In 2012 B.C.’s regulator blocked Uber, labeling it a “$75‑minimum limo.” For 7 years Vancouver was a big North‑American city with no ride‑hailing—an economist’s natural experiment. 2/12

Count me in!

I regularly use SQLite.

Whoa! This is really cool!