Uber and Lyft once promised to replace personal cars and do away with traffic congestion. However, some municipal authorities believe that these services have only exacerbated traffic jams. Research shows that their impact on road traffic depends on the ‘design’ of cities.
November 18, 2021   |   Ivan Korolev

‘People will not own cars, they’ll have a service that takes them where they want to go, when they want to go there,’ said former Uber CEO Travis Kalanick in 2016. If 30 people go to work in 30 personal cars, those cars are in use for only 4% of the day, and they are idle 96% of the time, Kalanick compared, while ridesharing enables a single car to serve 30 people. John Zimmer, the President of Lyft, Uber’s key competitor, predicted that the practice of private car ownership in large cities would disappear by 2025.

Despite ride-hailing companies’ promises to help solve the problem of traffic congestion and improve urban transportation, they regularly come under fire from both taxi drivers and consumers. Taxi drivers think that ride-sharing companies are driving them out of business, and they have staged multiple protests against them, for example, in France, Greece and Argentina. Consumers are angry with prices. App-based ride-hailing services have recently become the focus of municipal authorities’ attention. In 2019, New York University researchers published a study of 13 cities where Uber and Lyft were operating. According to the study, the authorities of some megacities (New York, Chicago, Los Angeles, Melbourne, Mexico City, Mumbai, Moscow) were considering various regulatory options to tackle ride-hailing companies’ impact on traffic congestion, including introducing congestion taxes.

As a result, in January 2020 in Chicago, where trips using transportation network companies have grown 4.5 times since 2015, the usage fee was increased by 70% for downtown trips on Uber, Lyft, and Via. New York City introduced such a tax in 2021 for trips within a designated downtown area, extending it to limousines and standard taxi services.

Ultimately, even ride-hailing companies themselves have recognised their contribution to traffic congestion, claiming, however, that its scale does not stand comparison with the impact of personal cars. What do the existing studies say about ride-hailing companies and their impact on traffic and urban life?

Public and personal transport

In theory, access to Uber should reduce car ownership, but in practice this is not the case: after the service was rolled out in US urban areas, the number of cars per capita increased by 0.7% on average in 2010–2017, according to the researchers from Carnegie Mellon University, Stanford, and Berkeley. However, the effect is uneven and it is more pronounced in smaller cities, where the launch of the Uber service has prompted residents to purchase more cars to join the app-based services.

Nationwide, rather than citywide, the data show that the average effect turns negative, a further sign of its heterogeneous nature, the economists write. The nationwide results are consistent with the earlier findings that the number of personal transport registrations per capita decreased following the entry of Uber and Lyft into the market in 2005–2015.

However,reducing car ownership in itself may not be the ultimate goal, explains Jeremy Michalek, one of the co-authors of both papers: ‘I think the reason we might want to reduce vehicle ownership is we want to reduce the negative impact of ownership, like congestion and emissions and crashes and those things.’

Uber and its peers can both compete with public transport and complement it. The question is which prevails in reality: competition or substitution.

Yash Babar and Gordon Burtch from the University of Minnesota found that Uber’s presence in the market brought about a 1.3% drop in bus trips, but the number of rail commuting trips went up by almost 3%. The researchers analysed data from 2012–2018 for 379 urban areas in the US with populations over 50,000 people where 2,276 transport companies were operating. The authors attributed these conflicting effects of ride-hailing services on different types of public transport to the first/last kilometre problem, which is more pronounced for commuter trains. Bus stops are located in a short distance from one another (and from houses), so they are normally easily accessible, whereas the distance between commuter train stations may be quite long. By helping people get from home to the nearest station (and back), Uber supplements commuter rail transport, but it competes with bus service.

A study by Erik Nelson and Nicole Sadowsky from Bowdoin College examines the sequential market entry of the two companies in 28 of the 30 largest US cities in 2002–2016. In most cases, Uber is the first to enter, with Lyft following it after a while. When the former enters the market, the number of public transport users goes up, but when the latter subsequently enters, it drops. The economists suggest that this can be explained by competition: when there is only one ride-hailing operator in a market, it has the power to set quite high prices and therefore cannot compete with public transport over any distance; at the same time, it may reduce passenger costs (making it easier for passengers to reach the nearest public transport stops, for example), leading to an increase in the use of public transport. After the emergence of the second company, competition between Uber and Lyft triggers a decrease in prices, enabling the two operators to lure away public transport passengers.

According to the calculations by Mi Diao from Tongji University, Shanghai and his co-authors from MIT based on the number of public transport users in 174 urban agglomerations in the US in 2012–2016, the entry of app-based ride-hailing companies into the market reduces the number of public transport users by 8.9%, while average personal car ownership (the average number of cars per household) changes only slightly. On the other hand, the economists note that car ownership declined by 1% in the top ten transit agglomerations (leaders in the monthly number of public transport trips per capita). This means that ride-hailing services and public transport can be complementary to each other and act as substitutes for personal cars.

Impact on road congestion

Diao and his co-authors come to another conclusion: although, by boosting car sharing, app-based services could in theory prove a lasting solution to the problem of traffic congestion, in reality, they only worsen the situation: the travel time index (the ratio of the travel time during the peak period to the time required to make the same trip at free-flow speeds) has increased since the emergence of these services, as has the duration of trips when the traffic is busy.

However, there is another study that showed that Uber’s entrance into the market had no significant impact on traffic congestion. However, the overall impact depends on how compact the city is – how densely built-up it is and whether it has clear boundaries or is a sprawling urban area. Compact cities include, for example, New York, San Francisco and Los Angeles. Nashville and Atlanta are sprawling cities. Uber does not significantly affect traffic congestion in the most sprawling cities, but significantly increases it in the most compact ones. In exactly the same way, sprawling cities have recorded increases in the number of public transport trips with the emergence of Uber by solving the first/last kilometre problem, but has reduced the use of public transport in compact cities. It is buses which first and foremost have been affected most by this reduction. In compact cities, Uber lures users away from public transport by providing a cheaper alternative to conventional taxi services.

Consequently, the total number of car trips grows in compact cities, worsening the traffic, while in sprawling cities, ride-hailing services supplement, rather than replace, public transport and may indeed drive improvements in road congestion.

Researchers from the University of Kentucky and the San Francisco County Transportation Authority studied how the emergence of app-based ride-hailing companies had changed traffic congestion in San Francisco in 2016 compared with 2010. In 2010 ride-hailing companies had a negligible effect on traffic while in 2016 they accounted for 15% of total intracity car travel –12 times the percentage of traditional taxis. The economists used a model specifically developed for San Francisco (the San Francisco Chained Activity Modeling Process) to predict the intensity of traffic on different city roads at different times of day. Based on input data for 2016 (population, number of employees, etc.), they estimated a hypothetical situation on the roads in 2016 as if app-based services were unavailable and compared it with the actual situation. They found that the difference in travel times both during traffic jams and when the roads are clear increased by 63% compare to 2010 with ride-hailing services and by 22% without them. The average speed drops by 13% and 4% respectively. The authors conclude that app-based ride-hailing services are responsible for the more significant deterioration of the traffic situation in San Francisco over the six-year period under study.

Suvrat Dhanorkar from Pennsylvania State University and Gordon Burtch from Boston University studied the example of California to investigate the impact of ride-hailing services on traffic. They collected monthly data for the period between January 2010 and December 2015 from over 9,000 radar devices on the state’s roads and studied what happens when Uber started operations in different areas pf California. It turned out that the launch of the service significantly increased the number of cars on the roads at weekends. Moreover, ride-hailing operations resulted in stronger traffic intensity both on workdays and at weekends in densely populated areas, in areas where the initial use of public transport was high, and on local roads (as opposed to highways). This finding looks consistent with the conclusions about the relevance of cities’ compactness factor: the higher the population density and the greater the number of short-distance trips on public transport, the more public transport is supplanted by ride-hailing services. At the same time, this does not work for the long-distance trips.

A potential drawback of these studies is that they analyse the impact of app-based ride-hailing services on public transportation or traffic in isolation from the factors that Uber considers when making the decision to enter a specific market. This decision may be influenced by local factors, such as the level of demand, in which case the estimates of its impact may be incorrect. For this reason, Matthew Tarduno of the University of California, Berkeley, turns to the natural experiment that took place in Austin in December 2015, when the city passed a law introducing verification tests for drivers working with ride-hailing companies, including fingerprinting. Uber and Lyft challenged the law, but failed and shut down their operations in May 2016.

Traffic congestion declined when Uber and Lyft halted their operations, especially during the morning rush hour and between 11.00 and 14.00. Average traffic speeds increased by 0.026 minutes per mile, or 0.9%, although this result is statistically insignificant. The speed went up by more in the daytime (7.00–19.00) – by 2.3%.

Assuming that the value of one travel hour for consumers equals half the average hourly wage, Tarduno estimated the total annual cost of traffic jams in Austin caused by Uber and Lyft at $33–46 million. The annual cost of all traffic jams in the city totalled nearly $810 million, that is, ride-hailing companies accounted for a modest percentage of the total cost of traffic jams, between 4% and 6%. In addition to costs, ride-hailing companies bring benefits to consumers: in the case of Austin, they are estimated at $47–$73 million a year. Tarduno concluded that, although the withdrawal of Uber and Lyft from Austin did reduce traffic congestion, its overall implications for public welfare were hardly positive.