With a certain logic, quasi-taxi-like services such as Uber and Lyft provide a public health benefit to cities. They give the bar-hopping demographic a better way to get home at night, and, as a result, they may help cut down on DUIs.
Sommer Mathis made a great case for this idea in June at CityLab: "What if the best way to end drunk driving," she asked, "is to end driving?" Traditional cabs have long pushed this thinking, too, smartly selling their services on the inside of bathrooms stalls in popular bars (at least, this is how the marketing works in the ladies' room). Just the other night, I was reading an ad in a Georgetown loo suggesting that I leave all my risky moves on the dance floor — and call a cab.
Arguably, smartphone-enabled services like Uber and Lyft could have an even greater impact here than traditional cabs. They're particularly popular among young users with the means and interest to go out drinking. And it seems plausible — although I haven't seen great data on this — that they're not merely replacing traditional cab service; they're capturing trips that otherwise might have taken place in a personal car.
To test this theory, Nate Good, the curious CTO of an event ticketing company in Pittsburgh, recently pulled DUI data in the Philadelphia metropolitan region (hat tip to PlanPhilly for noticing). Below are his results, separating monthly DUI arrests by people over and under 30 years old:
At right, you'll notice that he marks the launch of several smartphone services in the Philadelphia region, which appear to coincide with declining DUIs. Good himself is the first to admit that correlation does not necessarily imply causation here. This data doesn't tell us, for instance, about what was happening in late 2007 that led to an earlier spike in DUIs (as the economy sours, more people drink?).
Philadelphia is also not the best test case for this argument: So-called "rideshare" services are technically banned in the city by the parking authority that regulates taxis there. Sidecar, one of the services that connects passengers to rides in a private vehicle with a non-professional driver, entered the market in 2013 but was booted out about five months later. And UberX, Uber's similar service, today operates in nearby South Jersey, but technically not in Philadelphia itself.
If widespread services like these were to change drunk driving behavior on a significant scale, Philadelphia is not the place where we would see the effect to the fullest extent. Uber itself has looked at separate DUI data in Seattle and claimed an impact there. By the company's own analysis — interpret with a grain of salt — Uber's arrival in town was followed by a decrease of about 10 percent in daily DUI arrests.
Curious about both of these claims, we turned to one more city, San Francisco, where all of these apps first launched, and where they've been on the market the longest. San Francisco also happens to have particularly good open data (thanks, San Francisco!), including crime incident data from the San Francisco Police Department dating back to 2003.
Below, Chris Ingraham and I have similarly plotted DUI arrests by month in San Francisco (this is the city proper, using SFPD data, whereas Good's analysis was metro-wide using state data). We've noted several key dates as well: The launch of Uber's original black towncar service in June of 2010, the spate of so-called "ridesharing" launches in the summer of 2012, then the passage in September of 2013 of regulation by the California Public Utilities Commission that legally recognized those services:
Again, caveats are in order. We've simply plotted arrests on a timeline here; we haven't adjusted for changes in the city's population, or bar scene, or the economy. Any number of other things may have changed in the city over the last few years affecting DUI arrests. It's possible police have changed how they conduct DUI stops and arrests, or that public pressure on them to crack down on DUIs has ebbed with time. Other changes in public transit service may have impacted alternative routes that bar-hoppers take home.
It's also striking that San Francisco and Philadelphia show a steep and parallel rise in DUIs long before these services ever came to town; on both charts, it looks as if DUI numbers may be returning to an older normal as much as they've been falling. Perhaps these services have arrived on the market just in time to ride the benefits of an improving economy? (More theories on this welcome below as well).
These data, though, do suggest that there's at least more to research here. They remind me of a comment Lyft's VP for government relations, David Estrada, made to me recently. "As a company — this might sound pollyanish — we talk about our service not being aimed at providing transportation," he said, "but at lowering crime rates in a city like Chicago."
DUIs are potentially one way this might happen. If these services, which run on credit cards, take cash out of transactions as well, they may also cut down on other kinds of crime like theft (this is an argument Uber has made in Chicago). Estrada was also simply talking about the idea that crime might decline because Lyft likes to think that it creates community — and jobs in communities that don't have enough of them. That last argument merits a lot more skepticism. DUIs, though, we might actually wrap our arms around.