Close
Close

Published

Ridesharing – doth one congestion study a death sentence make?

Uber’s IPO is taking up the vast majority of current headlines in the ridesharing world, but in the week before, a study appeared that claimed congestion rates in San Francisco had rocketed due to Uber and Lyft. The data seems a little dated, but the claim they were responsible for a 62% increase in congestion could prove extremely damaging to the future of ride-sharing – as one of its main collective claims was that it would help reduce congestion and associated air pollution.

The study was conducted by the University of Kentucky, in collaboration with the San Francisco County Transportation Authority, and was published in Science Advances. Broadly, it looked at a 2010 snapshot of the city’s traffic, and compared it to 2016 data, to gauge how the pre and post ridesharing world looked. What the study found is rather concerning, although more recent data would have been nice to see, especially with the San Francisco CTA being involved.

According to the report, 15% of all car trips in the city can be attributed to ridesharing (12x higher than the number of taxi trips), and because of the practice of ‘deadheading,’ where ridesharing drivers will cruise around popular areas as they wait for new passengers, the ridesharing fleets are responsible for a 62% increase in city congestion. The report estimates that around 20% of all miles driven by the ridesharing fleets can be attributed to deadheading.

But the other major contributing factor is the cost and convenience to the traveler. The report estimates that between 43% and 61% of the total ridesharing trips could have been carried out on foot, on public transport, or by bike. However, because ridesharing is so affordable and easy to access, many consumers are choosing to use these services.

The issue of price-based congestion is compounded by the continued subsidizing of trips by the likes of Uber and Lyft. Many trips are below-cost, sold as part of an attempt to gain market share and then later recouped by increasing prices. However, the ridesharing firms haven’t displaced the incumbent taxi providers, and many professional drivers are working simultaneously for multiple transport providers.

To this end, the early versions of Uber and Lyft, where a student would be able to use their car to provide laid-back trips as a way to supplement their income, have fallen completely by the wayside. Nowadays, the expectation for a ridesharing trip is a spotless car, a professional driver, and a seamless transaction – not a scruffy university student in a hand-me-down car fishing for beer money.

This evolution in expectation largely took place without an associated increase in cost per trip. In the early days, a beat-up car could be tolerated because of the relative cost of the trip in a rideshare compared to a taxi. This means that the quality of the service offered has become much higher across the board, compared to the conditions at the start of the ridesharing boom.

So, the low cost of the ride, combined with its high quality, mean that consumers are more likely to opt for rideshares now than they once were. This leads to people opting for a rideshare instead of a trip on a bus, with weather also a major component in people’s decision on whether to walk or cycle. Time is a major motivation too, and something that is difficult to model. A $10 trip in a private car that takes ten minutes might be immensely preferable to a thirty-minute trip that takes two buses, some walking, but only costs $5. The multitude of factors that go into such purchasing decisions make modeling very difficult.

The study used a six-week window in 2016, measuring how long ridesharing vehicles were deadheading, as well as the number and time of their pick-ups and drop-offs, scraping data from the APIs of the ridesharing firms. According to the model, the weekday vehicle hours of delay rose 62% between 2010 and 2016, and when the traffic is modeled in a counterfactual scenario with no ridesharing providers, that congestion has only grown by 22%.

The report says that ridesharing does have a few viable mechanisms to reduce congestion, such as concurrent usage for multiple occupants. Simulations made show that this ridesplitting practice could reduce congestion, but of course, much of the appeal of ridesharing is the private use of the vehicle – not having to share with other people. Uber Pool hasn’t exactly taken off, to this end.

Another potential solution to the congestion problem is for the ridesharing firms to focus on better first and last-mile connections to public transit, says the study. Waymo has publicly stated that this is a focus, where public transport is supported by ferrying passengers to and from hubs on ridesharing platforms.

However, the study also highlights mechanisms that would increase congestion, such as the aforementioned deadheading. The research says that deadheading is a much bigger problem in New York, at around 50% of ridesharing miles travelled compared to San Francisco’s 20%. It also points to the pick-up and drop-off process as disrupting traffic flow by blocking the lane nearest to the pavement.

The study does have a caveat. It notes that San Francisco, a dense and transit-rich city, isn’t going to be representative for many other urban areas, and that more research is needed to study (or rationalize) similar studies that have come before it – and have proven largely inconclusive. This study is pretty concrete in its claim, so there is definitely a need for more supporting data.

The average speed of travel was the main basis for the model. In 2010, this was 25.6 mph, and falls to 22.2 mph in 2016, with the vehicle hours of delay (VHD) increasing 62% in the same period. It also found that 70% of ridesharing drivers live outside San Francisco, and that 43% of the ridesharing vehicle miles travelled (VMT) occur between 1830 and 0300.

Close