668. Do Taylor Swift and Bad Bunny Have Blood on Their Hands?

What It Covers

Harvard physician-economists Bapu Jena, Vishal Patel, and Chris Worsham publish an NBER working paper finding that major Spotify album releases — from artists like Taylor Swift, Bad Bunny, and Drake — correlate with a 15% spike in US traffic fatalities, roughly 18 additional deaths per release day, driven by smartphone distraction.

Key Questions Answered

• •Natural experiment methodology: Album release dates function as natural experiments because they create sudden, measurable spikes in streaming behavior — Spotify volume rises 40% on release days — without being chosen by researchers. Comparing fatalities to the same weekday one week before and after controls for the well-documented Friday traffic effect, isolating the album release signal specifically.
• •Falsification testing as credibility check: Running 1,000 iterations of randomized placebo dates produced the observed effect size only 1-2% of the time, confirming the result exceeds chance. Applying each album's calendar date to years with no release showed no fatality spike, and the effect diminished progressively as albums ranked lower in first-day stream counts.
• •Passenger presence as protective factor: Solo drivers show a larger album-release fatality effect than drivers with passengers. On album release days specifically, having a passenger reduces crash risk relative to surrounding days, suggesting a second occupant either operates the device or simply reduces the driver's impulse to interact with the phone or infotainment system.
• •CarPlay amplifies rather than reduces risk: Vehicles equipped with Apple CarPlay show a larger album-release fatality effect, not a smaller one. The hands-free interface lowers the friction of accessing music, causing drivers to interact with the system more frequently than they would without it — a compensating behavior pattern consistent with economist Sam Peltzman's model of risk adaptation.
• •Telematics data gap limits traffic safety research: Federal FARS data lags two-plus years and lacks granular driver behavior records. Automakers and platforms like Apple and Google hold real-time telematics — acceleration, braking, app usage at moment of impact — but deny researcher access. Closing this gap would allow direct causal analysis rather than inference chains through natural experiments.