Kriston Capps is a staff writer for CityLab covering housing, architecture, and politics. He previously worked as a senior editor for Architect magazine.
More hired drivers should mean fewer drunk drivers. But a new study proves conventional wisdom wrong.
When Uber and Lyft pulled out of Austin earlier this summer after voters passed a ballot measure regulating their drivers, many of the services’ supporters argued that their exit would put more drunk drivers on the road. After the ride-sharing services left, Austin recorded 359 arrests for DWI in May 2016, an uptick from 334 arrests for the same period one year prior.
Austin is one of a few major metro areas where Uber and Lyft opened up shop only to quickly close down operations less than two years later. With such small figures and over such a short time, it’s hard to say whether a 7.5-percent increase in arrests represents a spike in drunk driving, a surge in police vigilance, a statistical blip, or something else. Still, given the popularity of these services (especially Uber), it should be possible to show where ride-hailing has cut down on drunk-driving fatalities—somewhere, if not in Austin.
In fact, Uber hasn’t had much of an impact at all. A new study released by the American Journal of Epidemiology finds that Uber’s deployment in the nation’s 100 most-populated metro areas has no association with traffic fatalities—not in aggregate, not those caused by drunk driving, and not during weekends or holidays.
The study, by Noli Brazil of the Spatial Sciences Institute at the University of Southern California and David S. Kirk at the University of Oxford, assessed traffic fatalities before and after the advent of Uber in 100 different metro areas (by county) between 2005 and 2014. The researchers controlled for various factors that could affect risk, including state laws legalizing or decriminalizing marijuana, changes to driver licensing, bans on texting, and so on. The tax rate on alcohol, too, was another factor accounted for by Brazil and Kirk, as were monthly unemployment rates and the number of taxi drivers working in a given metropolitan county. Looking exclusively at Uber’s impact on drunk-driving fatalities, they found no impact.
“If would-be drunk drivers were rational, then lowering the difficulty of finding alternate transportation options and the cost of those options would, in theory, reduce the number of drunk-driving occurrences and fatalities,” the study reads. “This is the promise of Uber and other [transportation network companies], particularly in terms of increasing the supply of transportation options.”
Using three different models, Brazil and Kirk found that Uber did not reduce traffic fatalities in a statistically significant way. That runs counter to the conventional wisdom that more hired drivers would seem to mean fewer drunk drivers. The conventional wisdom appears to be wrong, and the researchers have some theories on why.
Would-be drunk drivers aren’t rational
For almost any driver, the potential cost of driving drunk outweighs the real cost of an Uber fare. You could kill someone (or yourself), wreck your car, accrue massive penalties, or wind up in jail when driving under the influence. Uber on the other hand is not so bad (except when it’s surging).
Drunk drivers don’t necessarily weigh these potential costs rationally. As the researchers explain, intoxicated drivers may be more likely to think exclusively about the odds. Ultimately, drivers are unlikely to get caught when they drink and drive—whereas they are certain to be on the hook for an Uber fare, a parking ticket, a car tow, or personal inconvenience if they decide not to drive. This is a bad bet! That’s the point: Drunk drivers make bad bets.
Uber is big, but not drunk-driving big
The number of Uber drivers across the U.S. is growing astronomically. The figure for monthly Uber drivers on the road has grown from a few thousand drivers in January 2013 to 450,000 monthly drivers in April 2016. If the service continues to expand at exponential rates, it could some day make a dent in the number of drunk driving-related crashes and fatalities. (Note that the study also accounts for Lyft, but focuses on Uber because it has vastly more drivers.)
But for all the sound and fury over Uber and Lyft, drivers for these services represent a tiny, tiny share of drivers on the road. There are 210 million licensed drivers in the U.S. Of these, some 4.2 million adults decide to drive while intoxicated in any given month. Uber simply doesn’t compare.
For drunk drivers, Uber is no substitute for driving
The Austinite argument that drunk-driving incidents would skyrocket once Uber and Lyft left town was one part genuine concern, one part concern-trolling. After all, no one making that argument ever said, “If I can’t take Uber or Lyft, then I plan to drive drunk.” Instead, the worry was framed as a problem facing other people. If other people cannot take Uber or Lyft, then they will drive drunk. This concern may be misguided, according to Brazil and Kirk.
“Uber may be a substitute for taxis and other forms of public transportation but not a substitute for drunk driving,” they write. “Accordingly, Uber passengers may have formerly been taxi and public transit users, and thus the number of at-risk drivers on the road would not substantially change. Prior evidence has suggested this substitutability.” (Links to footnotes added.)
Uber riders may not be your typical drivers
In the most populous metro areas, passengers have other options beyond Uber. Even before Uber came around, many would-be Uber passengers likely didn’t drive. They took taxis, buses, bikes, or rail (and still do). In major metro areas, Uber works more like a very expensive form of public transit than an alternative to driving. People living near public transit or people earning low incomes may not consider Uber to be an option. Neither might drivers who simply depend on their cars instead of any public transportation.
Where Uber’s impact on drunk-driving fatalities needs further study is in smaller towns, the researchers concede. “Uber may have a greater association with the number of traffic fatalities in smaller areas, where transportation options are limited,” the study reads. “In future research, investigators should examine whether the association between Uber’s presence and traffic fatalities depends upon the availability of alternative transportation options.”
On this point, Austin is something of a special case. For a city as populous as it is and growing as fast as it is, Austin’s public transportation network is lagging behind. Austin combines big-city growth with small-town infrastructure. One consequence may be that the relatively wealthy residents who, for a brief spell, relied on Uber and Lyft as a form of transit—the profligate California refugees, the high-earner-not-rich-yet Silicon Hills types, the Torchy’s Millennials with disposable income to burn—may be returning to their cars.
This very visible, very voluble cohort is small, though. The greater concern for Austin is the growing number of residents who don’t consider Uber or any kind of transit to be a viable option.