Does ride-hailing increase or decrease vehicle kilometers traveled (VKT)? A simulation approach for Santiago de Chile

Abstract

Many authors have pointed out the importance of determining the impact of ride-hailing (ridesourcing) on vehicle kilometers traveled (VKT), and thus on transport externalities like congestion. However, to date there is scant evidence on this subject. In this paper we use survey results on Uber use by residents of Santiago, Chile, and information from other studies to parameterize a model to determine whether the advent of ride-hailing applications increases or decreases the number of VKT. Given the intrinsic uncertainty on the value of some model parameters, we use a Monte Carlo simulation for a range of possible parameter values. Our results indicate that unless ride-hailing applications substantially increase average occupancy rate of trips and become shared or pooled ride-hailing, the impact is an increase in VKT. We discuss these results in light of current empirical research in this area.

Keywords:

• Ride-hailing
• ridesourcing
• sharing economy
• taxi
• transportation network companies

Acknowledgments

We thank four anonymous referees for their insightful comments and suggestions that helped us to improve the paper.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1 On this last issue see Greenwood and Wattal (2015) who provide evidence using data from California that Uber reduces alcohol/DWI homicides. See also Dills and Mulholland (2016). For the case of Chile, see Lagos, Muñoz, and Zulehner (2018).

2 An optimization approach to integrate ride-hailing with public transportation is introduced and applied by Chen and Nie (2017).

3 City of New York (2016) does recognize that ridesourcing applications have eroded an important source of transit funding (special levy on taxi rides) in spite of the fact that ridesourcing trips are subject to an 8.875% sales tax, 0.375% of which go directly to the Metropolitan Transit Authority.

4 One cannot infer the change in patronage of transit services from these figures since the change in the number of trips by each respondent are not reported, only whether people use them more or less.

5 This is a very relevant issue as, for example, A. E. Brown (2018) shows that 32% of Lyft trips at peak periods were on the pooled service Lyft Line in Los Angeles.

6 As reviewed by Beer, Brakewood, Rahman, and Viscardi (2017), there is a great diversity in the ride-hailing regulatory frameworks applied by cities, including driver-related regulations (such that background checks, vehicle registrations, special driving licences and external vehicle displays) and company-related regulations (e.g., caps on ride-hailing vehicles in specific areas, obligation to share data with the regulator and to provide a list of drivers).

7 Rayle et al. (2016) obtain 380 completed responses and their interception survey was undertaken during two months on weekdays and Saturday evenings.

8 OECD/ITF (2015) also use scenarios or simulations to study the effects of shared self-driving cars on several variables of interest, including VKT, in an application for Lisbon, Portugal.

9 We assume that other modes such as cycling or walking do not cause congestion.

10 As traditional taxis also incorporate smartphone-based e-hailing applications to contact costumers, as it is occurring in many cities (e.g., the app Easy Taxi in Chile), these two parameters will tend to converge.

11 However, this may imply more walking to and from bus stops rather that an increase in the number of vehicle kilometers.

12 In Santiago, origin destination surveys that cover the whole metropolitan area have been conducted by the government every 10 years.

13 Ease of payment was also the number one reason to use ridesourcing apps in the San Francisco survey reported by Rayle et al. (2016).

14 “Uber detecta mayor aumento de viajes entre zonas periféricas y el centro en horas punta”, El Mercurio newspaper, March 4th, 2017, http://impresa.elmercurio.com/Pages/NewsDetail.aspx?dt=2017-03-04&dtB=04-03-2017%200:00:00&PaginaId=9&bodyid=3, accessed October 17th, 2017.

15 In this survey, respondents could mention more than one mode as being replaced by the use of ride-hailing.

16 Nie (2017) also estimates that with the arrival of ride-hailing, the rate of time taxis run without passengers increased to 50%–70% during 2015, due to the reduction on taxi ridership in Shenzhen. Interestingly, taxi ridership has stabilized in the city during 2016.

17 We used the well-known sample size formula for the estimation of a mean that is normally distributed (see, e.g., Chapter 7 in Roess, Prassas, & McShane, 2011). First, with 10,000 replications a standard deviation of 0.68 km was obtained. Then, for a 95% confidence interval and a desired margin of error of 0.01 km/trip, we obtain a sample size of 18,570 iterations.

18 This figure is the mean of all 20,000 Monte Carlo draws. If, instead, we estimate the VKT increase for the expected value of all input parameters, the result is 1.67 km.

19 Own calculation based on SECTRA (2013).

20 The effect of service frequency on reducing passenger waiting time and therefore encouraging public transportation use is known as the “Mohring effect” (Mohring, 1972).

Additional information

Funding

This research is supported by the Complex Engineering Systems Institute, Chile (CONICYT: FB0816).