Data, et cetera

Riding the subway: Uber & the future of American metro

March 2, 2019

Photo credit: John St. John

Email Facebook Print Twitter

Much is written about the weaknesses (or, some might say, decline) of the United States public transportation system. Every person that uses public transportation is likely to have stories of waiting an hour for a bus, or using a ride-sharing app because the bus stop was a 30 minute walk from home, or being crammed into an overly-full bus at rush hour. And anyone that travels to major Europe or Asia cities is likely to come away a bit surprised at the efficiency and convenience of the subway system.

There are a number of hypothesized reasons for the failures of American public transit. In a piece in CityLab, professor of urban affairs Ralph Buehler described a series of government failures including the subsidisation of driving through public spending on roadways and lower tax rates for fuel (which are often earmarked specifically for more roads), failure of public funding to save or support struggling publicly-owned transit systems, and large residentially-zoned neighborhoods that make transportation without a car difficult. A piece in The Economist has attributed the decline in transit ridership to ridesharing, low-cost car loans, and the growth of people working remotely from home.

The decline of American public transit is not new; indeed, ridership was beginning to drop in the 1940s. Federal funding for transit systems began to grow in the 1960s, and today the United States continues to invest in urban rail projects that are almost universally more expensive than comparable systems in Europe (according to a piece in The New York Times, due to a combination of trade unions, construction companies, and consulting firms). Yet despite the high costs of investing in transit, not investing in the $90 billion needed to modernize US public transit could cost the U.S. economy nearly $60 billion in lost business revenue every year.

Who is using public transportation?

So we decided to dig into the data. We wanted to see how people's use of public transportation has changed over time and how funding practices vary by city. To do so, we measured monthly ridership of rail, bus, trolley, and 'on demand' transit systems from 2002-2016.

We find that every state, including Washington DC, has a bus system and some sort of demand-responsive transit. There are 32 rail systems (including commuter rail, light rail, and heavy rail), 21 trolley systems, and 20 other systems that include aerial tramway, ferryboat, and inclined plane.

By analyzing ridership (measured as unlinked passenger trips), we can observe trends in public transportation use between 2002 and 2017. Bus systems, on average, had their peak ridership in 2010, whereas rail systems peaked in 2012. There wasn't a single bus system that was at its historical maximum ridership in 2017. By contrast, rail ridership was at its peak in 2017 in Alaska, Colorado, Minnesota, Texas, and Washington. The largest declines in ridership (from peak year) for bus were in New Hampshire, Puerto Rico, and Louisiana (more than 50% decline). For rail, the largest drops were in Tennessee, Puerto Rico, and Missouri (more than 40% decline).

Ridership, by transit type (2002-2017)

Select a state:

Transportation types
Source: Monthly Module Adjusted Data Release. Federal Transit Authority. September 2018. 'Bus' includes bus, bus rapid transit, commuter bus, and publico. Rail includes Alaska railroad, automated guideway, commuter rail, heavy rail, hybrid rail, light rail, and monorail/automated gateway. Trolley includes cable car, streetcar, and trolley bus. On demand includes demand response, demand response - taxi, and vanpool. Other includes aerial tramway, ferryboat, and inclined plane.

Which subway systems have the most riders?

There are fifteen heavy rail systems in the United States, known colloquially as metro, subway, or rapid transit systems. Despite only serving a handful of states, these receive a tremendous amount of media attention and are often a critical component of the transit infrastructure in large cities. We decided to analyze how ridership has changed over time, and what may be driving those changes.

In the United States, the fifteen transportation agencies with heavy rail systems are based in just twelve cities: Atlanta, Baltimore, Boston, Chicago, Cleveland, Los Angeles, Miami, New York, Philadelphia, San Francisco, San Juan, and Washington, DC. By far the largest heavy rail system in the country is the MTA New York City Transit, reaching across New York, New Jersey, and Connecticut and providing nearly 2.7 million unlinked passenger trips in 2017.

The Chicago Transit Authority (Illinois) and the Washington Metropolitan Area Transit Authority (Washington, DC, Maryland, and Virginia) had the second and third largest ridership, providing approximately 230 and 227 million rides in 2017, respectively. The smallest systems were in Staten Island Rapid Transit Operating Authority (New York, New Jersey, Connecticut), the Alternativa de Transporte Integrado (Puerto Rico), and the Greater Cleveland Regional Transit Authority (Ohio), providing fewer than ten million trips per year.

How has Uber impacted ridership?

Many writers and analysts have speculated that the popularity of ride-sharing apps and services like Uber and Lyft have contributed to declines in public transportation ridership. To analyze this, we looked at the change in heavy rail ridership from the time when the first ride-sharing app, Uber, was rolled out to 2017. Uber was first rolled out in San Francisco, CA, in 2010 and by 2016 was rolled out in every city with a subway system.

We find that the impact of Uber has varied tremendously between cities. In Chicago, San Juan, New York City, and San Francisco, subway ridership has increased since the rollout of Uber. By contrast, the other eleven heavy rail systems have experienced declines in ridership since Uber was introduced. Baltimore's metro system experienced the largest declines, with ridership going down by nearly 47% since February 2013, followed by Washington, DC's decline of 17%.

Change in metro ridership: Uber launch to September 2018

Metro System Location Date Uber launched Change in ridership
Baltimore Metro Subway Baltimore, MD February 2013 -46.59%
Washington Metro Washington, DC December 2011 -16.96%
Metro Rail Los Angeles, CA March 2012 -14.47%
Metrorail Miami, FL June 2014 -9.28%
MBTA Subway Boston, MA September 2012 -8.32%
MARTA Atlanta, GA August 2012 -7.60%
RTA Rapid Transit: Red Line Cleveland, OH April 2014 -4.91%
Staten Island Railway New York, NY May 2011 -4.53%
SEPTA Philadelphia, PA June 2012 -4.20%
New York City Subway New York, NY May 2011 -1.95%
PATCO Speedline Philadelphia, PA June 2012 -1.49%
Chicago "L" Chicago, IL September 2011 0.12%
Tren Urbano San Juan, PR July 2016 3.41%
PATH New York, NY May 2011 5.05%
BART San Francisco, CA July 2010 12.90%
Source: Monthly Module Adjusted Data Release. Federal Transit Authority. September 2018.

Are subway systems paying for themselves?

To see how subway systems are being paid for, we looked at the percent of each system's operating expenses that come from ridership fares. Described another way, this calculates how much each system is being subsidized by other forms of government revenue, as opposed to it paying for itself.

On a whole, American subway systems are heavily dependent on general government revenue. On average, subway riders are only paying for an average of 41% of subway operating expenses. We find astounding variation in revenue coming from ridership. The BART system in California and the New York metro are mostly rider-paid, with around three-quarters of costs paid by ridership fees. On the other end of the spectrum, smaller metro systems like Staten Island, San Juan, Miami, and Cleveland were more than 80% subsidized by other forms of government revenue.

This could be due to variation in the cost of riding or variation in the cost of operating. The total cost to the subway system for one ride ranges from more than $8 in Staten Island and San Juan to less than $3 in Atlanta, Chicago, Boston, Philadelphia, and New York City. The average fare of a trip is highest (more than $2) in New York City, Washington DC, Philadelphia, and San Francisco/Oakland. Fares are lowest (less than $1) in Miami and Los Angeles.

Percent of operating costs paid for by fares (2018)

Source: Monthly Module Adjusted Data Release. Federal Transit Authority. September 2018.

Dense cities or urban sprawl: do city characteristics matter?

Comparing geographic coverage, we find that the Massachusetts Bay Transportation Authority (Massachusetts, New Hampshire, and Rhode Island) covers the largest geographic area (3,244 square miles), followed by the Baltimore and Los Angeles systems. The BART in San Francisco, CA, the Staten Island Transit Operating Authorit, and the Philadelphia Port Authority Transit each cover less than 100 square miles.

By population, the subway systems in New York City, Los Angeles, and Baltimore each cover a population of greater than seven million. By contrast, the systems in San Francisco, San Juan, and the port authorities in New York and Philadelphia each serve less than one million.

Overall, the New York City systems (the city subway and the Port Authority) serve the densest area, with more than 26 and 14 thousand people per square mile, respectively. Chicago and Philadelphia's subway systems are also dense (both around ten thousand people per square mile). The most spread out systems were in Boston (less than one thousand people per square mile), Atlanta (two thousand), and Baltimore (three thousand).

What's the big picture for U.S. metro systems?

By comparing ridership, costs, and city characteristics we find very different subway systems in the United States. As a result of economies of scale, subway systems with large ridership like New York City and Chicago tend to have the lowest per-customer operating costs. The largest American subway systems are also more likely to pay for themselves out of fares. Subway systems with fewer riders, like San Juan or Cleveland, tend to be heavily subsidized from other government resources. Generally subway systems covering a more dense coverage area, like Chicago or New York City, declined less than others.

Why some subway systems have lost ridership due to Uber while others have not is unclear. Cities with higher fares had, paradoxically, much smaller declines in ridership after Uber began operating. However, the BART, which has the highest fares and has seen ridership increase since Uber appeared, has the highest average trip length, indicating that riders may be less likely to switch to cars or Uber if they are taking the subway long distances. This relationship doesn't hold for other cities; Miami, for example, provides long trips yet has a low and declining ridership.

Understanding why subway systems succeed or fail is dependent on many factors, many of which are not described here. But understanding and evaluating the drivers of ridership changes is critically important as cities try to reduce traffic, mitigate climate change, and promote less congested cities.

Email Facebook Print Twitter

Copyright © 2019 Data, Et cetera