Metro’s 29 hour shutdown led to an Uber boom, with a 70 percent increase over the previous week in people signing up for the app as the shutdown drew near. While many worried about sky-high fares, they rarely materialized due to a combination of ride-splitting (the company expanded its UberPool ride-sharing platform to cover the entire region, not just DC), more drivers on the road, and many people in the area just staying home. Uber’s policy of “surge pricing” is, at this point, as ubiquitous as the company itself. The car-hailing service increases its rates when demand for rides outpaces the number of available drivers. In theory, this surge in pricing should encourage more Uber drivers to get on the road bringing prices closer to the standard rate.
New data show in practice where, when, and how much price surging occurs throughout the District. Select a neighborhood on the map below to show the average rate for an UberX any hour across the week within the selected neighborhood from February 3 to March 2.
Nearly all of DC experiences a price surge during morning commutes. Surging is most common and most extreme downtown, while least common along the outer ring of the city. As Wonkblog reports, neighborhoods with more people of color have fewer periods of surge pricing, but also wait longer for drivers to arrive.
And while there are some citywide trends—like weekdays about 8 AM, when fares shoot up just about everywhere—weekend surging is more variable by neighborhood. Areas with more nightlife, like Logan Circle and Shaw, see Uber prices surge about 2 AM Sunday as the bars empty. Quieter neighborhoods like Eastern Market see a similarly increases, but hours earlier. Upper Northwest has an Uber surge when most people start their evening, about 6 PM. For many parts of the city, Uber riders consider the late ride home just another part of the bar tab.
Technical notes: Data shown are from the Uber API. The data were collected by Jennifer A. Stark and Nick Diakopoulos over a month long period from 276 DC locations every three minutes. You can read more about the data, and their findings, here. You can find complete code for this post on my GitHub page.
