ArticlePDF Available

Abstract and Figures

Studies in several cities indicate that ridesourcing (ride-hailing) may increase traffic and congestion, given the substitution of more sustainable modes and the addition of empty kilometers. On the other hand, there is little evidence if smartphone apps that target shared rides have any influence on reducing traffic levels. We study the effects of a shared-mobility service offered by a start-up in Mexico City, Jetty, which is used by travelers to book a shared ride in a car, van or bus. A large-scale user survey was conducted to study trip characteristics, reasons for using the platform and the general travel choices of Jetty users. We calculate travel distance per trip leg, for the current choices and for the modes that riders would have chosen if the platform was not available. We find that the effect of the platform on vehicle kilometers traveled (VKT) depends on the rate of empty kilometers introduced by the fleet of vehicles, the substitution of public versus private transport modes, the occupancy rate of Jetty vehicles and assumptions on the occupancy rate of substituted modes. Following a sensitivity analysis approach for variables with unavailable data, we estimate that shared rides in cars increase VKT (in the range of 7 to 10 km/passenger), shared vans are able to decrease VKT (around −0.2 to −1.1 km/passenger), whereas buses are estimated to increase VKT (0.4 to 1.1 km/passenger), in our preferred scenarios. These results stem from the tradeoff between the effects of the occupancy rates per vehicle (larger vehicles are shared by more people) and the attractiveness of the service for car users (shared vans attract more car drivers than buses booked through Jetty). Our findings point to the relevance of shared rides in bigger vehicles such as vans as competitors to low occupancy car services for the future of mobility in cities, and to the improvement of public transportation services through the inclusion of quality attributes as provided by new shared-mobility services.
Content may be subject to copyright.
The sustainability of shared mobility: can a platform for shared rides reduce
motorized traffic in cities?
Alejandro Tirachini (1,2,*)
Emmanouil Chaniotakis (3)
Mohamed Abouelela (4)
Constantinos Antoniou (4)
(1) Transport Engineering Division, Civil Engineering Department, Universidad de Chile, Santiago, Chile
(2) Instituto Sistemas Complejos de Ingeniería, Santiago, Chile
(3) MaaSLab, Energy Institute, University College London, London, UK
(4) Chair of Transportation Systems Engineering, Technical University of Munich, Munich, Germany
(*) Corresponding author:,,,
Tirachini, A., Chaniotakis, E., Abouelela, M. and Antoniou, C. (2020) The sustainability of shared mobility:
can a platform for shared rides reduce motorized traffic in cities? Transportation Research Part C:
Emerging Technologies 117, 102707.
Studies in several cities indicate that ridesourcing (ride-hailing) may increase traffic and congestion, given
the substitution of more sustainable modes and the addition of empty kilometers. On the other hand,
there is little evidence if smartphone apps that target shared rides have any influence on reducing traffic
levels. We study the effects of a shared-mobility service offered by a start-up in Mexico City, Jetty, which
is used by travelers to book a shared ride in a car, van or bus. A large-scale user survey was conducted to
study trip characteristics, reasons for using the platform and the general travel choices of Jetty users. We
calculate travel distance per trip leg, for the current choices and for the modes that riders would have
chosen if the platform was not available. We find that the effect of the platform on vehicle kilometers
traveled (VKT) depends on the rate of empty kilometers introduced by the fleet of vehicles, the
substitution of public versus private transport modes, the occupancy rate of Jetty vehicles and
assumptions on the occupancy rate of substituted modes. Following a sensitivity analysis approach for
variables with unavailable data, we estimate that shared rides in cars increase VKT (in the range of 7 to
10 km/passenger), shared vans are able to decrease VKT (around −0.2 to −1.1 km/passenger), whereas
buses are estimated to increase VKT (0.4 to 1.1 km/passenger), in our preferred scenarios. These results
stem from the tradeoff between the effects of the occupancy rates per vehicle (larger vehicles are shared
by more people) and the attractiveness of the service for car users (shared vans attract more car drivers
than buses booked through Jetty). Our findings point to the relevance of shared rides in bigger vehicles
such as vans as competitors to low occupancy car services for the future of mobility in cities, and to the
improvement of public transportation services through the inclusion of quality attributes as provided by
new shared-mobility services.
Keywords: Shared mobility; Vanpooling; ride-splitting; public transportation; Transportation Network
Companies, ride-hailing,
1. Introduction
The fast-increasing availability of information and communication technologies (ICT) and positioning tools
in smartphones has enabled a growing number of mobility innovations and services (Shaheen et al., 2016).
The spread of smartphones and mobile internet connections has facilitated the expansion of mobility
options that have existed for a longer time, such as shared bicycles and shared cars. There are new
mobility technologies that depend entirely on the real-time connectivity between drivers and passengers
to agree on a trip. One of these technologies is ridesourcing (also known as ride-hailing
), in which a
traveler is matched through a mobile application with a driver, following optimization procedures to
reduce waiting time of users (Shaheen, 2018).
In the pertinent literature, the study of new ICT-enabled technologies includes the analysis of the effects
of ridesourcing on traffic levels (Erhardt et al., 2019; Henao and Marshall, 2019; Tirachini and Gomez-
Lobo, 2020), energy consumption (Wenzel et al., 2019), social equity (Brown, 2018) and
replacement/complementarity of/to public transportation (Clewlow and Mishra, 2017; Hall et al., 2018;
Graehler et al., 2019). An extensive review of these topics is presented in Tirachini (2019). Another strand
of literature deals with the regulation of ridesourcing services, either from a policy-oriented perspective
(Beer et al., 2017; Alonso Ferreira et al., 2018) or from economic models that aim at finding optimal values
of trip fares and fleet sizes in first-best and second-best environments (Zha et al., 2016). A special type of
ridesourcing service is ride-splitting (also known as shared or pooled ridesourcing), in which users, for a
fare discount, share a ride with other passengers (unknown to them) that have different origins and
destinations, therefore detours on the shortest path may be induced. Research on ride-splitting is still
rare, a notable exception is the study from Li et al. (2019), which analyzes detailed data of Didi ExpressPool
in Chengdu, China, and estimated that, on average, ride-splitting saves 22% of vehicle-hours of traffic,
relative to the situation in which the same trips would have been made by using (unshared) ridesourcing.
Other services for shared mobility use shared shuttles or vans that operate on flexible or semi-fixed routes
and can be reserved in advance or on-demand. Specifically, vanpooling, is a transportation service first
introduced in 1970 (Kircher and Wapensky, 1978), for groups of people who live in the same area,
commute to the same directions and choose to commute as a group in a van (Ditmore and Deming, 2018).
App-based vanpool services have become somewhat popular worldwide, including Berlin (Door2door and
Allygator Shuttle), New York (Via Van), London (Ford) and Beijing (Didi Chuxing). Dynamic vanpooling can
be a competitive mode of transport, as generally, it is more affordable than driving a private car or hailing
a taxi (Dong et al., 2018). A general analysis of alternative shared-mobility modes, including vanpooling,
carpooling, and ride-splitting, is found in Shaheen and Cohen (2019).
As current research increasingly shows, ridesourcing results in increased road traffic levels, in terms of
vehicle kilometers traveled (VKT) (Henao and Marshall, 2019; Tirachini and Gomez-Lobo, 2020) and
congestion delays (Nie, 2017; Agarwal et al., 2019; Erhardt et al., 2019). This issue casts serious doubts on
the sustainability effects of ridesourcing applications for urban mobility (Tirachini, 2019). The question
that begs is if there are ways in which this type of shared-mobility innovation, could reduce or at least
keep constant traffic levels in cities (at the same time that it provides several quality attributes to riders).
Such an outcome would certainly increase social welfare and pave the way for the inclusion of these
platforms as part of the multimodal transport portfolio to fulfill the mobility needs of citizens, for example,
in complementarity to mass public transportation systems. An increase in vehicle occupancy rates by
Service providers are commonly called Transportation Netowrk Companies (TNCs).
ridesourcing, achievable through ride-splitting, could imply a reduction in VKT, as predicted by Tirachini
and Gomez-Lobo (2020) using simulation. However, the increase in vehicle occupancy rates that is
achieved with ride-splitting is not by itself enough to decrease VKT, as evidence from China and the United
States shows that shared ridesourcing users are more likely to replace public transportation trips than
(un-shared) ridesourcing users (Lavieri and Bhat, 2019; Tang et al., 2019). None of these studies attempt
to estimate the effect of shared rides on the number of vehicle kilometers traveled in a city, using actual
trip-level passenger data. The present study aims to bridge this gap in the literature, by estimating if a
platform designed for shared rides can actually reduce motorized traffic levels. We perform a large-scale
user survey that includes questions regarding socioeconomic characteristics, travel behavior, reasons to
share rides and characteristics of the shared trip performed, including modes to access and egress, total
travel time and mode substitution. The latter information is used to estimate the effects of the platform
in traffic levels. and congestion. We focus on an urban environment of massive scale, Mexico City, and
analyze the case of Jetty, an app-based platform to book shared trips in cars, vans and buses. With Jetty,
users are picked up and dropped off in designated stops and are informed in advance about the type of
vehicle, license plate of vehicle and pick-up time. Jetty tickets are more expensive than those of the
traditional public transportation (PT) operators but cheaper than ridesourcing or taxi trips.
The rest of the paper is organized as follows. Section 2 describes the data and methods used for this
research, Section 3 summarizes general findings, while Section 4 focuses on the VKT effects of Jetty.
Conclusions and directions of further research are discussed in Section 5.
2. Data and methods
2.1 User Survey Design
A user survey has been designed that explores mobility choices, socioeconomic characteristics of the
users, reasons to use a platform for shared trips and alternative modes, should the platform be
unavailable. The survey is structured in three parts:
(i) information about the user’s latest trip, such as trip purpose, modes that were used,
including parking and walk, if relevant;
(ii) information about the general travel patterns of the individual, including use of shared
ridesourcing, such as the trip rate per mode, time use information, and the factors that
affect the use of Jetty, and
(iii) demographic information, such as occupation, education, age, gender, household size
and income, whether they have a driver’s license, as well as access to car.
The survey was designed in such a way that the results obtained by each user can be linked to trip and
socio-demographic data collected by the service providers. This has been achieved with the use of tokens
that allowed the definition of a unique Uniform Resource Locator (URL) for each user. Aiming at an
independent data collection that would respect privacy without compromising research independency,
the process followed included: a) contact the service provider and establish the necessary legal basis for
data exchange; b) receive a list of anonymized IDs and trip characteristics data to be linked with the data
collected from the survey; c) create individualized URLs for each user; d) distribute the survey through the
service provider communication channels (e.g. through the mobile-phone app, or through email). In
general, it should be noted that for research integrity, the results of the survey should not be shared with
the data provider before receiving all available user data. Random respondents were offered vouchers for
free Jetty rides after completion of the survey.
2.2 Study area and description of the service provider
The Metropolitan Area of Mexico City (Zona Metropolitana del Valle de Mexico, ZMVM in Spanish) was
chosen as a prominent case for the implementation of the survey. It has a population of 21 million people,
including the municipalities of Mexico City and the conurbations State of Mexico and Tizayuca. In ZMVM,
the latest household origin-destination survey (INEGI, 2017) shows that 41% of households own a private
car (approx. 3 million cars) and that on a normal working day, 34.5 million trips are made within the
metropolitan area. Regarding modal split, 40% of trips are made by public transportation (PT), 32% by
walking, 19% by private car, and 5% by taxis (including ridesourcing). Public transportation comprises of
a Metro network (12 lines, 227 km.), Bus Rapid Transit (BRT, Metrobus) and other formal bus services
such that the Ecobus, plus thousands of poorly regulated, lower quality colectivos, which include standard
size buses, microbuses and vans or “combis”. Colectivos are the largest PT mode in terms of trips and
spatial coverage, as more than 11 million trips are made daily using their services (33% of total trips).
Jetty, launched in 2017, is an app-based platform to book a seat in a shared vehicle. Jetty does not own
any vehicle, but rather makes deals with associations of vehicle operators in Mexico City and State of
Mexico (part of the metropolitan area around Mexico City) which join Jetty. Drivers must have a work
contract, the operator needs to have proper insurances and vehicles that are relatively new, well equipped
and maintained. Jetty keeps a fraction of the fare charged and the rest is transferred to the transport
. In our sample, average car, van and bus fares in Jetty are 67, 69 and 43 Mexican Pesos (MXN),
respectively, whereas the single-ride fare for colectivos, Metro, Metrobus and Ecobus is between 5 and 7
Mexican Pesos (1 USD= 19 MXN, 1 Euro=21 MXN). In the city, Jetty is a small player, with 37000 tickets
sold in November 2018 (Flores Dewey, 2018), which is equivalent to roughly 1 in every 20,000 trips in
ZMVM. At the time of writing, Jetty only operates during peak periods mainly for commuting trips. Jetty
works with several types of vehicles, which we categorize in three groups according to vehicle size:
Cars, with a capacity to carry between 3 and 6 passengers
Vans, with a capacity to carry between 13 and 17 passengers
Buses, with a capacity to carry between 30 and 45 passengers
Figure 1 shows a map of the Jetty service area in ZMVM as per May 2019. The points in green are the
pickup stops, while the points in gray are the drop off stops. Jetty has fixed routes, i.e., passengers are
told in advance about the pickup stop, where they have to board vehicles, and they know in advance the
route that the vehicle will follow until their drop off stop. In this way, the service provided by Jetty is
different from more flexible on-demand shared services with adaptive routes, such as ride-splitting
(shared ridesourcing); Jetty is closer to a traditional fixed-route bus service, and therefore its modal
substitution might be different from that of ride-splitting. Jetty has two main destination areas in the
morning: the districts of Santa Fe and Polanco, which are characterized by being important job
agglomerations. Santa Fe is one of the main business districts of Mexico City whereas Polanco is an
A more detailed description of Jetty is found in the International Transport Forum (ITF) Discussion Paper written
by one of the founders of Jetty (Flores Dewey, 2018).
affluent neighborhood and shopping area. In the afternoon, the routes start from these zones of
employment to the points where the users started their trips in the morning, which are spread in several
districts in the north, center and south of the city, as Figure 1 shows.
Figure 1: Geographical coverage of Jetty services (source: Jetty)
2.3 User survey deployment
The authors contacted Jetty in order to launch a research collaboration that would be based on data
exchange and the deployment of a survey to Jetty users. After signing a Memorandum of Understanding
and a Data Exchange agreement, Jetty provided a list of all the riders that had used Jetty in the past 6
months prior to the survey. The number of users that were invited to participate in the survey were 14093.
The user URLs were sent back, and the survey was distributed by sending personalized emails. The survey
was conducted online using the open-source tool Limesurvey, from an account administered by the
Technical University of Munich. The instance of Limesurvey was hosted by the authors and only they had
access to the data collected. Before the full deployment, a pilot study was conducted, for which the survey
was sent to 50 users. The final set of responses was collected between May 30th and June 11th, 2019 after
sending the survey individually to participants, as well as a reminder after the first week of data collection.
Santa Fe
Distribution of the survey resulted in 3091 responses (approx. 23% response rate) out of which 2484 were
completed. After an initial screening and by joining with the data provided by Jetty, the dataset that was
used for the analysis included 2169 participants. The linked data provided by Jetty included all trips
performed by the survey respondents for a period of 6 months before the survey; showing the trips’ start
and end location (WGS84 coordinates), start time, end time, type of vehicle used, trip ID and route ID for
each user. Also, the operator provided the average capacity (number of seats) per vehicle type for the
whole month of May 2019 to be used in the VKT calculation, and the definition and length of routes.
2.4 Procedure to estimate travel distances for Jetty alternatives
For estimating the effects of shared ridesourcing operators on VKT in Mexico City (Section 4), we use
responses to the question “If Jetty was not available, how would you have made your latest Jetty trip?”,
to find the alternative mode. The choice set consists of 15 options, from which respondents could choose
up to three. It was explicitly explained that choice of more than one mode, means combination (multi-
modal trip). Options like “I would not have traveled” and “Others” were available. Respondents were also
asked to state the total travel time that they would have experienced in that case. The modes to choose
from were: Car (driver); Car (Passenger); Taxi; App-based Taxi (e-hailing and ridesourcing); Shared taxi;
metro; Suburban train; Metrobus; Ecobus; Bus; Minibus; Combi and motorcycle. A very small percentage
of responders (3%) specified the option “others”. These other modes, in most cases, were variations
within the given set of modes of transport and they were mapped into the main choice set according to
their similarities. Users also reported their home and work zip codes in the survey. This information was
geocoded using a Google Maps geocoding package
to get startpoint and endpoint coordinates for each
trip. Around 51% of the respondents (1257 users) reported valid origin and destination zip codes that
could be used in the geocoding process.
Given survey related limitations (number of questions/alternatives, fatigue), there was no question from
which we could directly extract the mode sequence or any question on the distance traveled in each
alternative mode, if more than one mode would have been used instead of Jetty. Thus, the 13 modes of
the choice set were grouped into three main groups, to calculate the total trip distance and the share of
each of the modes per trip. The choice of the modes composing each group depends mainly on the
flexibility of the route of each mode, and the method used to calculate the corresponding VKT. The three
groups of modes are specified as follows:
Group 1: Shortest Path Group: This group consists of car driving, car passenger, taxi, e-
hailing/ridesourcing, shared taxi, bus, microbus, combi, and motorcycle, and different combinations of
those modes. The main characteristic of this group in Mexico City was that they do have flexible routes
and tend to use the shortest path to travel. The Google Maps direction API service was used to calculate
the total trip distance between the origin and destination of each trip, assuming travel time shortest-path
routes. For the case of PT modes such as bus, microbus and combi, a deviation factor over the shortest
path is added to the shortest path calculation, in order to compute VKT, as explained in Section 4.
mapsapi: 'sf'-Compatible Interface to 'Google Maps' APIs. R package version 0.4.2. Available at https://CRAN.R-, accessed January 13th, 2020
Therefore, given the uncertainty on distances per mode, for the assignment of total travel distance when
the trip reported is composed by a combination of two or three of the modes of this group, the following
two scenarios are made and compared:
Scenario A: The total travel distance of PT modes is larger than those of low occupancy modes.
This assumption reflects a situation in which a low occupancy vehicle such as a private car, a taxi
or ridesourcing is used to access PT vehicles, e.g., a bus, microbus or combi. This PT vehicle covers
the largest part of the trip. We assume that when we have either one or two public transportation
vehicles and one low occupancy vehicle (cars, taxis, e-hailing, ridesourcing), of the total distance,
3/4 is made by public transportation modes and 1/4 is made by low occupancy vehicles. In case
of having two low occupancy vehicles and one public transportation vehicle, distance is divided
as 1/4 for each low occupancy vehicle and 1/2 for the public transportation vehicle.
Scenario B: The total trip distance is divided equally between the number of modes reported for
each trip. This is thought to be a less conservative scenario for VKT analysis than A, for example,
B assumes equal distances when traveling by ridesourcing and microbus, even though the former
is much more expensive than the latter. In A, the contribution of Jetty to potentially reduce VKT
is lower than in B, because in B the number of VKT that the rider would have added to the network
if Jetty is not available is larger than in A.
Group 2: Rail Group: This group includes the trips reported to be (partially) performed by metro or
suburban train. The locations of the stations of those two modes were obtained from General Transit Feed
Specifications files (GTFS)
and Google Maps. The shortest path distance from each trip’s start- and
endpoint to the nearest station was calculated using the Google Maps API. The summation of distance
from the trip’s start- and endpoint to the nearest station was divided by the remaining number of modes,
also using the two alternative scenarios A and B.
Group 3: BRT Group: This group includes trips reported to be (partially) performed by Ecobus or
Metrobus. The main characteristic of this group was that they are running on fixed routes and that, unlike
microbuses and combis, the routes of Ecobus and Metrobus do appear in Google Maps. The Google Maps
Transit API was used to calculate the travel distance for each mode. The obtained distances were
distributed to the reported number of modes on a case-by-case basis.
2.5 Latest Jetty trip
The survey’s response date and time of each user were checked against their survey completion time. The
last Jetty trip that took place before the survey completion time was considered as the latest Jetty trip for
VKT analysis. The Jetty trips performed by users consisted of three stages: (a) from origin to Jetty’s pick-
up point; (b) the Jetty ride and (c) from Jetty’s drop-off point to destination. Jetty provided the middle leg
for the survey respondents, including the pickup and drop-off locations coordinates to be used to calculate
the access and egress distance. Additionally, the respondents reported the modes used to access their
last Jetty trip pickup point and to egress from the drop-off points to their final destination, apart from
reporting estimated travel time for access and egress (walking; bicycle; car; e-hailing/ridesourcing; taxi;
Metro; minibus; combi; Metrobus). Similarly, to the Jetty alternative modes (Section 2.3), the modes were
divided into the same three main groups as for access/egress modes, based on their route flexibility and
the method used to calculate the corresponding VKT.
4, accessed July 1st, 2019
3. Descriptive analysis
3.1 Descriptive statistics
The sociodemographic characteristics of our sample are presented in Table 1. Our sample is relatively
similar to the general Mexico City sociodemographics in terms of age, gender and household size,
according to the official information published by the National Institute of Statistics, Geography and
Informatics of Mexico (INEGI, 2015), as 52% of the population in Mexico City is composed by women, the
median age is 35 years old and the median household size is 3.2 inhabitants/household (INEGI, 2015).
However, there is a strong difference in terms of income, car ownership and education level between our
sample and the general population. In our sample, middle- and high-income people are overrepresented,
compared to the general population in Mexico City, as the median monthly income of riders in our sample
is around 1000USD
and the average income is even larger, whereas in Mexico City average income is
slightly over USD 500 per month
. This is consistent with the characteristics of Jetty, which offers a higher
level of comfort than traditional PT services (combis and microbuses), but for a higher price. Additionally,
the majority of Jetty users who responded to the survey are highly educated (91.7%), as opposed to 32.1%
for Mexico City population (INEGI, 2015). Additionally, 80% of the survey respondents have at least one
vehicle available at home, almost double the rate for the population average (41%). This also shows the
potential of such a platform to replace trips that would otherwise be made by car. Table 2 shows that 35%
of Jetty users in our sample replace trips that are performed by car (as a passenger or as a driver), and
67% of the sample replace trips that would have been made, in at least one stage, in low occupancy
vehicles (personal cars, taxis, ridesourcing).
Currency in Table 1 is in USD (in the survey, currency was Mexican Pesos MXN, 1 USD= 19 MXN in July 2019)
accessed July 1st, 2019
Table 1: Statistics of the sample (n=2484)
Pct %
Pct %
Household Income
Less than 525 $
525$ - 1050$
1050$ - 1575$
1575$ - 2100$
2100$ - 2625$
More than 65
2625$ - 3150$
3150$ - 3675$
More than 3675$
Driving license
Household Size.
Cars in Household
3 or More
Education Level
Masters or Doctorate
Personal Income
Bachelor or Professional
Less than 525 $
Technical Career
525$ - 1050$
High School or Baccalaureate
1050$ - 1575$
1575$ - 2100$
2100$ - 2625$
I did not complete formal
More than 2625$
Employment Status
Part time
Full time study
Full time job and Part time study
Part time job and Part time study
Note: Income is given in USD ($)
3.2 Reasons to choose Jetty
The attributes that influence users’ choices are examined in this section. These have been identified based
on the reasons why survey respondents use Jetty, as shown in Figure 2. Various market segmentation
factors were examined with the gender of respondents to yield interesting results (see point c). Following
a descriptive analysis of the survey results, we categorize the reasons to choose Jetty in four broad
a) Conventional mode choice attributes: travel time is ranked third in relevance (66% of
mentions), travel time reliability fourth (53% of preferences) and access/egress time fifth
(42% of mentions). Fare is only seventh (36%), which might be related to the fact that Jetty
trips are more expensive than the standard PT in Mexico City, and that the Jetty user profile
has an average income larger than that of the general population in Mexico City, as shown in
Table 1.
b) Provider-related attributes: this category refers to attributes related to the booking
technology and trip management of Jetty. Reserving a seat is the number one attribute as
mentioned by users (chosen by 70% of the sample). This finding can be linked to the crowding
and standing valuation in the PT literature, where it has been estimated the increase in
willingness to pay to reduce travel time when traveling standing, as compared to sitting, or
that riders are willing to have longer travel times if they can secure a seat (e.g., Wardman and
Whelan, 2011; Bansal et al., 2019; Márquez et al., 2019). Ease of fare payment, which is done
with a credit card through the smartphone app, is eighth with 28% of mentions, and fare
transparency (with Jetty fare is known beforehand, unlike when traveling by taxi) is chosen
by 1 out of 5 respondents. In other studies on the adoption of ridesourcing, ease of payment
was the number one reason to choose ridesourcing in a sample interviewed in downtown San
Francisco (Rayle et al., 2016) and Santiago (Tirachini and del Río, 2019) and was the number
seven in a Californian sample (Circella et al., 2018). Quality of vehicles comes sixth with 36%
of preferences; Jetty vehicles are more recent models and are equipped with cameras (Flores
Dewey, 2018), which contrasts with the lower quality of combis and microbuses in Mexico
City. All in all, we clearly see the relevance of quality-of-service attributes to explain users’
preferences, as also known from the public transportation literature (e.g., Tyrinopoulos and
Antoniou, 2008; dell’Olio et al., 2011).
c) Security/safety attributes: 70% of women and 66% of men selected security, in terms of
avoiding the risk of theft, as a reason to choose Jetty. This was the second most chosen
attribute overall. Then, security against harassment was chosen by 27% of women and 4% of
men, this being the attribute with the largest difference between men and women. Lastly,
traffic safety is chosen by one out of five respondents. These findings express that a large
fraction of riders feel more comfortable in terms of personal security when traveling in Jetty,
as compared to other modes in Mexico City, where pickpocketing, theft and sexual
harassment are notorious problems for PT users (ONU Mujeres, 2018; Magaloni, 2019).
d) Car-driving related attributes: these are “Use of time while traveling” with 25% of
preferences (place 9) and “avoid parking problems” (place 13), with 8% for women and 15%
for men. As a comparison to ridesourcing studies, not having to search or pay for parking is
ranked between places 4 and 8 as a reason to prefer ridesourcing in the United States (Rayle
et al., 2016; Henao, 2017; Circella et al., 2018) and in place 12 in Santiago (Tirachini and del
Río, 2019).
Figure 2: Reasons to use Jetty
3.3 Total travel times and modal substitution by Jetty users
The answer to the question “how long would your total travel time be in the trip replaced by your current
Jetty choice”, shows that the average travel time replaced by the current trip using the shared
ridesourcing operator is 97 minutes; and that 85%, 66% and 41% of users replace trips that would take at
least 1 hour, 1.5 hours and 2 hours, respectively, if not made using Jetty, as perceived and estimated by
survey respondents. It should be highlighted that these are time perceptions by the users, which might
be overestimated. The total duration of Jetty trips includes the reported access and egress times (which
are also reported by users, and do not include waiting times) and the Jetty’s in-vehicle time (which is
extracted from the database of user trips provided by Jetty). The average in-vehicle time in Jetty is 45
minutes, while average access and egress times are 18 and 15 minutes, respectively.
Interesting insights arise concerning modal substitution. First, most respondents replace complex modal
configurations by using Jetty instead. Responding to the question “If Jetty would not be available, how
would you made your last Jetty trip”, 42% of users in three modes, 32% of users would need to travel in
two modes and only 24% of respondents state they would make the trip in only one mode. In this question
it was explicitly stated that respondents should not count walking as an access mode to motorized modes).
Regarding the specific modes replaced by the latest Jetty trip, the left half of Table 2 shows the rates of
modal substitution per mode, in specific, the rate of trips replaced that has at least one stage in the
respective mode. The most replaced individual modes are metro (51.2% of trips has at least one metro
stage), followed by buses (called camion in Mexico City, 32.1%), car as a driver (25.1%), ridesourcing/e-
hailing (22.4%) and microbus (18.3%). When combining modes in categories, we see that 74% of Jetty
trips replace trips with at least one PT stage, 32% replace trips with at least one stage by car (as driver or
passenger), 29% replace trips with at least one stage in taxi or ridesourcing, and 52% replace trips with at
least one stage in a low occupancy vehicle (private car, taxi, ridesourcing). These numbers make unclear,
at first sight, if the use of Jetty increases or decreases VKT in Mexico City, an issue that we analyze in the
next section. To further explore the modal substitution patterns of Jetty users, in the right half of Table 2
we show the combinations of modes with the largest number of preferences (respondents were allowed
to mark up to three modes, which are used in combination, in the trip replaced by Jetty). It should be
noted that the 15 combinations presented account for 77% of trips in our sample. It can also be observed
that one out of five Jetty trips replaces a trip that would have been made in microbus and metro
combined, and 11.4% of trips would have been made by car. Trips by taxi and car plus taxi amount to 10%
of replaced trips. For this selection of combined modes, an aggregation of the modes is made to reduce
the number of combinations. The new mode definition is the following:
Car: car as passenger, car as driver
Taxi: taxi, e-hailing, ridesourcing
Microbus: Camion, microbus, combi
Metro: metro, suburban train
Bus: Metrobus, Ecobus
Shared taxi
Finally, significant for the evaluation of new mobility platforms is the degree of complementarity with
other modes. That is to say, which modes are combined, to complete a trip. Figure 3 shows the modes
used for access to and egress from Jetty vehicles. Trips are separated by morning and afternoon, given
that in the morning trips are from home to work and in the afternoon travelers return home. Only the top
10 mode combinations are shown, which encompasses 92.5% (access) and 95.4% (egress) of all mode
combinations. In general, walking is the most common way to access to and egress from Jetty vehicles. In
the morning, walking is used for access in 27% of trips and for egress in 72% of trips. The fact that the
predominance of walking as egress mode is not so marked for access might be partially explained by the
spatial structure of Jetty trips, shown in Figure 1: trips origins are spread out in large areas of the city,
whereas destinations are concentrated in a few trip attractors. People are willing to take motorized
vehicles in the morning to access a pick-up point to board a Jetty vehicle, which is dropping off the traveler
at walking distance from their jobs in a majority of cases. In the evening, the large difference between
walking as access mode vs egress mode vanishes, in which other factors might be at play, such as the
performance of other activities after work (before taking the Jetty vehicle) and the fact that commuters
might have lower values of travel time savings and of travel time reliability in the afternoon, when they
go back home, than in the morning (Carrion and Levinson, 2012).
Table 2: Mode replacement by Jetty: rates of modal substitution per mode (left half) and
combinations of modes with the largest number of preferences (right half)
Disaggregated modes
Combined modes top 15 combinations
Modal substitution
Modal substitution
Camion (Bus)
Car, as driver
Shared taxi
Car, as passenger
Shared taxi
Metro+Microbus+ Car
Suburban train
I would not have
Total modes
Total trips
Trips top 15
Average modes per trip
Figure 3: Top 10 modes for access to and egress from Jetty trip leg
4. Traffic impacts of Jetty
Following the procedure described in Section 2.4, we estimate the effect of Jetty in the number of vehicle
kilometers by motorized traffic in ZMVM. The analysis is done for a subsample of 1118 users representing
89% of the valid zip codes with correct origin and destinations (the rest of survey respondents did not
provide valid zip codes for origin and/or destination zones). The main reason that the total correct zip
codes are not included in the analysis is that some of the survey respondents were not in the trip database
provided by Jetty, as the Jetty database covers only the latest six months of usage. First, we estimate
vehicle kilometers and passenger kilometers of Jetty users, as shown in Table 3. Ten Jetty vehicle types
are used by the survey respondents, with vehicle capacities as in Table 3. Estimation of the following
variables is first pursued:
Number of passengers per vehicle type in the sample. Jetty provided us with data that include
the vehicle type of the latest trip performed by each survey respondent, before responding to
the survey. In total, 4% of trips are made in cars, 37% of trips are made in vans and 59% of trips
are made in buses.
Average demand per vehicle run, defined as the average number of passengers that use a
vehicle during one run, for all vehicle types. This was provided by Jetty and it is average
demand for June 2019.
Equivalency factors of vehicles, to be translated into VKT at a common passenger car
equivalency (PCE) factors. Equivalency factors are proportional to the minimum headway
between two consecutive vehicles, therefore larger vehicles have larger equivalency factors
relative to passenger cars. PCE factors represent the number of passenger cars that would have
an equivalent effect on the traffic flow (Zhou et al., 2019). PCE equivalency factors between 1.5
and 2.5 are common for trucks and buses according to size (TRB, 2010; Pajecki et al., 2019;
Zhou et al., 2019). PCE values between 1.5 and 2.5 for minibuses, standard buses and large
(BRT) buses are adopted in Table 4.
Passenger-kilometers traveled: This information is extracted from the Jetty database on the
latest trips performed by the 1118 users in the subsample.
Concerning the trips performed in the latest six months, the number of trips and the corresponding route
lengths were reported by Jetty. From this database, it is estimated that the average trip length (i.e., the
average distance traveled by users) is 24.9 km and the average route length (i.e., the average length of
the routes defined by Jetty) is 26.3 km. An equivalent of vehicle kilometers per vehicle type is calculated
in our sample, by estimating the number of equivalent rides per vehicle type in the sample [columns
(2)/(3) in Table 3], times the average route length per vehicle type, which is estimated as the average trip
length [columns (5)/(2)], times the average ratio of route length to trip length (26.3 km/24.5 km=1.07),
times vehicle equivalency (PCE) factors [column (4)].
 
   
   
The value of VKT estimated using this procedure is shown in column (6). Columns (7) and (8) show
Passenger-kilometers traveled (PKT) and VKT per vehicle type, as a percentage of total PKT and VKT,
respectively. Finally, column (9) is simply the ratio of (8) to (7). Columns (7) to (9) show the
disproportionate effect of passengers traveling in small vehicles (“Car” type) on VKT, as compared to
passengers traveling in larger vans and buses. For example, people traveling in Jetty shared cars represent
2.3% of passenger kilometers, but 8.2% of vehicle kilometers on the sample. At the other extreme of the
spectrum, passengers traveling in buses with 45 seats, amount to 33.6% of PKT, but only to 25.0% of VKT
this shows the adverse impact of the small Jetty vehicles on the VKT. Small vehicles are used in new routes,
where there is no enough demand to operate with buses or minibuses.
Table 3: Jetty vehicle-kilometers and passenger-kilometers per vehicle type
of users
in sample
demand per
We also estimate the VKT induced by road modes that riders use to access to and/or egress from Jetty
vehicles. It should be noted that walking, which is the main access and egress mode to Jetty does not
influence motorized VKT. However, for motorized modes, the following information and assumptions are
Average occupancy rate per type of vehicle and service. In Mexico City there is only information
about the average occupancy rate for cars, which is 1.3 pax/veh for commuting trips (INEGI, 2017).
In our database, the total travel distance of car as driver is 2812 km and of car as passenger is 678
km; we choose occupancy rates of car as driver and car as passenger to be 1.08 pax/veh and 2.08
pax/veh, respectively, because the distance-weighted average of these occupancy rates is 1.3, as
measured in Mexico City. We assume the same occupancy rate for taxi and ridesourcing, while
with passengers. For PT, there is no available information, therefore, we assume base values and
then make a sensitivity analysis over these assumptions. The base values of average occupancy
rates are 10 pax/veh for combi and microbus and 40 pax/veh for Metrobus (BRT), which is within
the range of bus occupancy rates in Santiago (Tirachini and Gomez-Lobo, 2020). The alternative
scenarios tested, over these base values, are to double and to reduce by half the assumed average
occupancies of PT modes (given that, as explained, there is no information of those occupancy
rates in Mexico City). These scenarios are called A2 in Tables 4 and 5 and are referred to as low
occupancy PT (base public transportation occupancies reduced by half) and high occupancy PT
(base public transportation occupancy doubled). These average values need to take into account
the whole lengths of routes, i.e., including areas of low and high occupancy.
For modes whose distance is calculated using a shortest-path algorithm (i.e. car, combi, microbus,
taxi and ridesourcing/e-hailing), a distance deviation factor over the shortest-path was included.
For combi and microbus, average deviation is assumed to be 15%, which is the average distance
deviation for shared ridesourcing (ride-splitting) as estimated in China (Li et al., 2019) and is within
the range of values for bus deviation factors assumed in Santiago (Tirachini and Gomez-Lobo,
2020). For car and taxi, a 5% deviation is included, because car and taxi drivers do not necessarily
follow shortest paths. Finally, for ridesourcing and taxi e-hailing trips, no deviation from the
shortest path is assumed.
For taxis and ridesourcing/e-hailing, an estimation of empty kilometers as a percentage of
passenger-kilometers is included in the estimation of VKT. Because there are no data in Mexico
City about empty kilometers, we use data from a secondary source. Several authors have provided
estimations of empty kilometers by taxis and ridesourcing, of special interest is Cramer and
Krueger (2016), who for two cities (Seattle and Los Angeles) estimate empty kilometers for both
taxis and ridesourcing (UberX). In both cities, on average, taxi and ridesourcing empty kilometers
are 60% and 40% of passenger-kilometers, respectively. These two factors, 60% for taxis and 40%
for UberX, are assumed as extra empty kilometers for taxis and ridesourcing/e-hailing in our
estimations, Table 4 shows all the assumptions used in the calculation process.
In our estimation, we include empty kilometers from Jetty as well. These kilometers are not fully measured
by the company, as in the mornings, drivers take vehicles from home or from a depot nearby the origin of
routes, and then, after the morning trip, vans and buses are parked nearby to wait for the afternoon peak.
As a precise estimation is not available, we do a sensitivity analysis on the value of total VKT, for different
levels of empty kilometers introduced as percentages of actual route length (same with taxis and
ridesourcing). Total VKT is estimated based on four assumptions of average increases in VKT due to empty
kilometers: 0%, 10%, 20% and 30%. For example, for Scenario 1, total Jetty VKT estimated using this
procedure is 7089, 7509, 7929 and 8349 PCU-km for 0%, 10%, 20% and 30% of added empty kilometers,
including motorized traffic used as access or egress modes.
Finally, using the procedure explained in Section 2.4 and the assumptions outlined above for Jetty VKT
estimation, we estimate the total VKT replaced by Jetty, using the answers of survey respondents to which
modes they would have used if Jetty were not available. With regards to supply adjustment of replaced
modes, three alternative assumptions are made:
1. All modes adjust supply levels as a response to reduced demand due to Jetty, including other PT
modes. This is the scenario assumed in Tirachini and Gomez-Lobo (2020) for the estimation of VKT
impacts of Uber in Santiago. This scenario is regarded as a long term reaction and is unclear if it
will ever materialize, given that apps such as Jetty for collaborative transport have the potential
to be incorporated as part of the PT system in the future, either developed by private
entrepreneurs or by the state, if public transportation rides are possible to be booked with
technology such as a smartphone app in the official PT system.
2. Low occupancy vehicles (private cars, taxis, e-hailing and ridesourcing) reduce kilometers. This
scenario assumes that apart from private car kilometers being saved, taxi/e-hailing and
ridesourcing drivers also adjust behavior and reduce kilometers driven if demand drops, in a way
to keep constant passenger occupancy rates and empty kilometers. This is regarded as a medium-
term reaction.
3. Only kilometers by private car are saved. Therefore, all other modes such as taxi, ridesourcing and
different PT alternatives continue running at the same level, despite demand losses to Jetty. This
is regarded as a very short-term situation, in which there is no supply adjustments to demand.
Table 4: Parameters assumed for VKT calculations
Replaced modes
Vehicle occupancy
Jetty Veh.
Eq. Size
A, B
Scenario A2
Low Occ.
A2 High
Camion, Bus
Car, as driver
Car, as
Shared Taxi
Under these assumptions, we are finally able to compute the effect of Jetty on VKT, for the subsample of
1118 trips for which we have full origin-destination information. Results are shown in Table 5. In scenario
A (which assumes a larger distance traveled by public transportation vehicles than by low occupancy
vehicles in the replaced trip), Jetty saves VKT in the long term scenario in which all other modes adjust
their supply (Scenario A.1). If only low occupancy vehicles adjust their supply, then the effect of Jetty on
VKT depends on the rate of empty kilometers introduced by the platform into the road network; savings
of VKT are only achieved if empty kilometers are kept very low (lower than 10%, Scenario A2). If empty
kilometers are between 10% and 20%, there is an increase between 2% and 8% of VKT. In the short term,
if only kilometers by private cars are replaced by Jetty, then there is an unmistakable increase in VKT
(Scenario A.3). The same trend is maintained in Scenarios B (in which kilometers by shortest-paths modes
are divided equally between all vehicles involved in the replaced trips, regardless if vehicles are low-
occupancy or public transportation), but, as expected, the numbers are more favorable towards VKT
savings by Jetty, e.g., in Scenario B.2, VKT reductions are also possible with a rate of empty kilometers
between 10% and 20%.
Finally, we model an alternative scenario to A.1, in which we double or reduce by half the assumed
average occupancies of public transportation modes (given that, as explained, there is no information of
those occupancy rates in Mexico City). These scenarios are called A2 in Tables 4 and 5. Because this
assumption only affects the case in which all modes are assumed to adjust VKT, only that case is shown
for A2 in Table 5. With A2, it is shown that even though the assumed occupancy rates of public
transportation do affect the estimation of VKT changes, it only changes the conclusion of a reduction in
VKT if public transportation modes have a larger occupancy and the empty VKT by Jetty is 20% or larger
(Scenario A2.1a).
Table 5: Estimation of change in total VKT due to Jetty
Assumption on supply adjustment of
other modes
Jetty empty km (as percentage of commercial km)
A.1 All modes
A.2 Low occ
A.3 Private car only
B.1 All modes
B.2 Low occ (car+taxi+ridesourcing+
B.3 Private car only
A2.1a All Modes - high occupancy PT
A2.1b All Modes - low occupancy PT
In Figure 4 we show the effect on VKT of each survey respondent, for Scenarios A.2; a positive value
means that the user is adding VKT to the road network.
Figure 4: Difference in VKT per User
Finally, for Scenario A.2, we estimate the VKT effect per trip and per vehicle type, as shown in Table 6.
Trips in cars run by Jetty are those that add the most VKT to the network (between 6.6 to 9.5 km/trip on
average). Second, van users save VKT in all empty kilometer scenarios, with savings between -1.5 and -0.2
km/trip. Finally, bus users increase VKT between 0.1 and 1.1 km/trip. The latter result is somewhat
counterintuitive, as from Table 3 we could expect that larger vehicles would have a more favorable effect
in VKT than smaller vehicles, which is true when looking at vans but it is not true when looking at buses.
The explanation for this finding comes for the fact that the VKT effect from different vehicle types is
explained not only by the number of passengers carried by each vehicle, but also by the modes replaced.
In this respect, when examining modal substitution by Jetty vehicle type, we do find relevant differences,
as shown in Table 7, in which we show the substitution of combined modes, with the same mode
aggregation as in Table 2. Comparing van and bus users, we see that for vans the most replaced mode is
traveling by their own car (155), followed by the combination Metro+Microbus (13%) and then by two
low-occupancy alternatives: taxi (9%) and taxi+car (7%). On the other side, for bus users, the most
replaced mode is Metro+Microbus (23%), car replacement falls to 10%, and places 3 and 4 are also
combined modes that include Metro and Microbus. Therefore, we conclude that there is a larger tendency
to replace low occupancy modes in van users than in bus users, and that is the reason why van users save
VKT and bus users increase VKT in Scenario A2, even though buses carry more passengers per vehicle than
vans. It is worth mentioning, however, that higher-quality bus services as those provided by Jetty (relative
to traditional colectivos in Mexico City, as shown in Section 3.2), might be a future direction for the public
transportation system in the city, in order to help stop riders from leaving public transportation in their
daily commute. Such policy of increasing quality in public transportation should be studied in parallel with
the provision of proper fare subsidies, in order to keep affordable prices for low-income users. It is also
worth mentioning that different Jetty vehicle types may have different levels of empty kilometers added
to the road network; however, we do not have specific data to assess such hypothesis.
Table 6: Estimation of VKT difference per trip [km/user] per Jetty vehicle type, Scenario A.2
Jetty empty km (as a percentage of
commercial km)
Table 7: Substituted (combined) mode by Jetty users, per Jetty vehicle used
Car users
Van users
Bus users
Substituted mode
Substituted mode
Substituted mode
Total substitution top
15 modes
5. Conclusions and policy implications
This research focuses on the estimation of the impact that shared rides on cars, vans and buses have on
vehicle-kilometers traveled (VKT). With the use of a large-scale dataset of approx. 2500 responses (for
1118 of which we were able to obtain a full dataset including origin and destination of trips), accompanied
by some assumptions commonly met in the pertinent literature (for example, on occupancy rates of
modes substituted by Jetty), we perform sensitivity analyses and develop different scenarios for which we
evaluate possible increases or reductions of VKT. The case we examine is quite unique: Jetty is a shared-
mobility platform operating in one of the largest cities in the world; it targets a market share mainly
composed of middle to high-income travelers with a high percentage of private vehicles in the household
(double than the national average). Therefore, given the characteristics of its users, Jetty has the potential
to replace a significant number of personal car trips. Another main characteristic of Jetty is that the
operational routes serve trips whose origins and destinations are underserved by other direct connections
and means of public transportation, as 74% of the sample would need two or more modes in combination
to complete the latest trip made in Jetty, if Jetty was not available. Given that Jetty is an app for shared
trips, its rate of replacement of other shared or public transportation modes is larger than the rates
reported in ridesourcing studies.
When ranking the reasons to use Jetty, we observe that the possibility of booking a seat, security against
theft, trip duration and travel time reliability are the most valued attributed by Jetty users, whereas
security against harassment has the largest difference in valuation between male and female riders. A
large percentage of the platform users are substituting traditional modes of transport for reasons other
than travel time, which points to the relevance and valuation of quality of service attributes within the
context of disrupting mobility technologies. Regarding VKT, our results show that (i) there is an increase
in VKT if only kilometers by private cars are assumed to be ceased because of Jetty, (ii) there is a reduction
in VKT if all replaced modes (including private and public transportation) are assumed to adjust their
supply in order to keep their occupancy levels per trip constant, and that (iii) the VKT result depends on
the rate of empty kilometers that Jetty introduced to the network, in the case in which only low occupancy
vehicles (private cars plus taxis/e-hailing and ridesourcing) are assumed to adjust their kilometer supply.
Therefore, the effect of this new service on vehicle-kilometers in the city largely depends on the time-line
of the analysis being performed (short-term versus long- term effects) and on the management and
optimization of empty kilometers by the service provider.
This conclusion adds to the literature that points to the relevance of shared rides and increasing the
occupancy rates of vehicle trips, for new mobility technologies, such as ridesourcing and automated
vehicles, to reduce or at least not to increase traffic in cities (Sterling, 2018; Tirachini, 2019; Narayanan et
al., 2020). The analysis of VKT per type of Jetty vehicle shows an interesting outcome, as the effect on
potentially reducing VKT of Jetty vans is better than the effect of Jetty buses, which is explained because
van users replace more trips by low occupancy vehicles than bus users; this modal substitution effect
overcomes the fact that buses have a larger passenger occupancy rate per vehicle than vans in our
estimations. Therefore, well-tailored routes on middle-size vans seem to be attracting more car users than
Jetty buses. At this point, this negative result for buses should be qualified in a broader sense, by also
including the effect of these new on-demand bus services on retaining travelers in large occupancy
vehicles, i.e., analyzing the possible effect of these buses in stopping people from switching from low-
quality colectivos straight to driving cars. Such potential long-term effects are not possible to analyze with
our data. It is also relevant to note that higher-quality bus trips (which is the case analyzed in our study),
might be the way forward of the whole public transportation system in the future.
The fact that shared rides in vans can indeed reduce VKT is a significant finding in the strand of literature
on the sustainability effects of app-based mobility platforms. Unlike the latest studies on (unshared)
ridesourcing that estimate an increase of motorized traffic and congestion due to ridesourcing platforms
(Nie, 2017; Erhardt et al., 2019; Henao and Marshall, 2019; Wenzel et al., 2019; Tirachini and Gomez-
Lobo, 2020), our analysis shows that sharing rides in middle-size vehicles results in an opposite effect. A
reduction of VKT due to the partial replacement of trips that otherwise would be made in low occupancy
vehicles, together with an improvement in the quality of service experience by users, is good news for the
sustainability of present and future mobility patterns in cities. It is to be seen if such result holds with
services targeting a larger demand (both geographically and in terms of socioeconomic characteristics)
and in other urban environments.
The policy implications of our findings are wide. Our results point to the relevance of innovative ride-
sharing schemes in bigger vehicles, such as vans, as competitors to low occupancy car services for the
future of sustainable mobility in cities. This finding can support related policies and encourage
policymakers to steer their strategies and incentives towards such services. On the other hand, Jetty buses
may increase VKT based on our results, but they are still buses offering a new type of public transportation
services. As discussed, an integration into the public transportation system could provide synergies with
broader benefits for the overall system, that is to say, authorities could explore the adoption of booking
technologies and higher quality services as part of the public transportation system of Mexico City. An
alternative approach could also be taken by allowing (upon certain conditions) the start-up business
ecosystem to target sustainable mobility options. If this happens in cooperation with public transportation
authorities, it might substitute trips that would otherwise happen using cars, something that might also
have long term effects such as reduction of car-trips and car-ownership.
Naturally, integration and cooperation to and with the official public transportation system pose the
challenge of social equity, as the added cost of enhanced services would have to be covered by additional
subsidies to make such services affordable by the general population. Furthermore, less tech-savvy
segments of the population or citizens with limited access to smartphones might be excluded from such
services. There is also the risk of public transportation with different qualities used exclusively by specific
income groups, endangering social cohesion and deepening income inequality. These challenges should
be addressed in a balanced manner, aiming at reducing car dependency and excessive car use in
congested cities, while safeguarding public transportation as a mean to enhance social cohesion. The
equity effects of new shared and public transportation services are particularly relevant in the context of
Mexico City and other large conurbations in developing countries, characterized by several forms of social
Other take-away lessons to be learned are that Jetty users do not only choose the service because of
travel time; there are other significant factors that can encourage travelers to switch to shared-mobility
modes. These findings could be used to adapt other (including public transportation as discussed before)
services, so that they are more appealing to specific segments of the population. Of course, these
conclusions can also be useful for developing shared-mobility services similar to Jetty in other cities in the
world, but also for a potential refinement of Jetty’s services in Mexico City.
Finally, our results also highlight the relevance of the number of empty kilometers that a shared-mobility
provider adds to the road network of a city. An efficient management of empty kilometers with tactical
decisions (e.g., long-term actions such as the renting of parking slots or terminals in the proximity of route
destinations) as well as with operational decisions (e.g., day-to-day vehicle scheduling and route
optimization) is of utter importance to improve the chances of motorized shared-mobility modes to
reduce traffic in cities.
The available dataset is very rich and can support a multitude of additional analyses. Future research
directions include (i) optimizing the network structure of the Jetty operation, as well as the allocation of
vehicles to specific routes, (ii) developing policy interventions and schemes that can support the operation
of such services, in a way conducive to the reduction of overall VKT, energy consumption and emissions,
(iii) examine the impact of such services on transport equity, (iv) explore innovative and dynamic pricing
schemes and mechanisms, including fare and operational integration into the city public transportation
system, or (v) explore the impact of vehicle automation on the system operations, using the latest results
on the effects of automation for shared vehicles and public transportation services (Fielbaum, 2019;
Narayanan et al., 2020; Tirachini and Antoniou, 2020) and (vi) analyze if different vehicle types (cars, vans
and buses) have different rates of empty kilometers. The study of the optimal economic regulation of
ridesourcing (Zha et al., 2016) should be extended to the case of shared or pooled rides. Another direction
of research is the collection of similar data from other systems in the world, to analyze the results
comparatively and try to understand which might be general results and which might be context-specific.
Declarations of interest
This paper is part of a research collaboration set up in 2019 between the Chair of Transportation Systems
Engineering of the Technical University of Munich (TUM) and Jetty. We thank Jetty for allowing us to
independently conduct a survey to its users, without compromising research integrity and independence,
as the survey and its data collection process were fully managed by the authors. Part of this paper was
written while the first author was Fellow of the Humboldt Foundation at TUM. This research has been
partly supported by the German Research Foundation - Deutsche Forschungsgemeinschaft (DFG, Project
number 392047120)], the TUM International Graduate School of Science and Engineering - IGSSE (MO3
Project) and CONICYT Chile (PIA/BASAL AFB180003). We are indebted to three anonymous reviewers for
detailed comments that helped us to improve the final version of our paper.
Agarwal, S., D. Mani and R. Telang (2019). The Impact of Ride-hailing Services on Congestion: Evidence
from Indian Cities. Working paper, available at SSRN:
Alonso Ferreira, M., F. T. Salva Rocha, A. Giuli and F. de Mello Franco (2018). Politics, polity and policy of
ridesourcing regulation in São Paulo. Urban Transport in the Sharing Economy Era: Collaborative Cities.
CIPPEC, Argentina.
Bansal, P., R. Hurtubia, A. Tirachini and R. A. Daziano (2019). Flexible estimates of heterogeneity in
crowding valuation in the New York City subway. Journal of Choice Modelling 31: 124-140.
Beer, R., C. Brakewood, S. Rahman and J. Viscardi (2017). Qualitative Analysis of Ride-Hailing Regulations
in Major American Cities. Transportation Research Record 2650: 84-91.
Brown, A. (2018). Ridehail Revolution: Ridehail Travel and Equity in Los Angeles. PhD thesis, University of
California Los Angeles.
Carrion, C. and D. Levinson (2012). Value of travel time reliability: A review of current evidence.
Transportation Research Part A 46(4): 720-741.
Circella, G., F. Alemi, K. Tiedeman, S. Handy and P. Mokhtarian (2018). The Adoption of Shared Mobility
in California and Its Relationship with Other Components of Travel Behavior Report, National Center for
Sustainable Transportation, United States.
Clewlow, R. R. and G. S. Mishra (2017). Disruptive Transportation: The Adoption, Utilization, and Impacts
of Ride-Hailing in the United States. Research Report UCD-ITS-RR-17-07, UC Davis Institute of
Cramer, J. and A. B. Krueger (2016). Disruptive change in the taxi business: the case of Uber. NBER Working
Paper 22083.
dell’Olio, L., A. Ibeas and P. Cecin (2011). The quality of service desired by public transport users. Transport
Policy 18(1): 217-227.
Ditmore, C. J. and D. M. Deming (2018). Vanpooling and its effect on commuter stress. Research in
Transportation Business & Management 27: 98-106.
Dong, Y., S. Wang, L. Li and Z. Zhang (2018). An empirical study on travel patterns of internet based ride-
sharing. Transportation Research Part C: Emerging Technologies 86: 1-22.
Erhardt, G. D., S. Roy, D. Cooper, B. Sana, M. Chen and J. Castiglione (2019). Do transportation network
companies decrease or increase congestion? Science Advances 5(5): eaau2670.
Fielbaum, A. (2019). Strategic public transport design using autonomous vehicles and other new
technologies. International Journal of Intelligent Transportation Systems Research. DOI
Flores Dewey, O. (2018). App-Based Collective Transport Service in Mexico City: A Start-Up Case Study.
International Transport Forum Discussion Papers, OECD Publishing, Paris.
Graehler, M., R. A. Mucci and G. D. Erhardt (2019). Understanding the recent transit ridership decline in
major US cities: service cuts or emerging modes? 98th Annual Meeting of the Transportation Research
Board (TRB), Washington D.C.
Hall, J. D., C. Palsson and J. Price (2018). Is Uber a substitute or complement for public transit? Journal of
Urban Economics 108: 36-50.
Henao, A. (2017). Impacts of ridesourcing LYFT and UBERon transportation including VMT, Mode
replacement, parking and Travel Behavior. Ph.D. Thesis, University of Colorado.
Henao, A. and W. E. Marshall (2019). The impact of ride-hailing on vehicle miles traveled. Transportation
46: 21732194.
INEGI (2015). Banco de Indicadores, Ciudad de Mexico (Databank, Mexico City). Instituto Nacional de
Estadistica y Geografia (INEGI), available at, accessed
INEGI (2017). Encuesta Origen-Destino en Hogares de la Zona Metropolitana del Valle de Mexico (EOD
2017) (in spanish). Report, Instituto Nacional de Estadistica y Geografia (INEGI).
Kircher, D. and L. Wapensky (1978). Vanpooling: An Overview. Publication EPA-908/178- 001 (Denver,
CO: US Environmental Protection Agency, Region VIII).
Lavieri, P. S. and C. R. Bhat (2019). Investigating objective and subjective factors influencing the adoption,
frequency, and characteristics of ride-hailing trips. Transportation Research Part C: Emerging Technologies
105: 100-125.
Li, W., Z. Pu, Y. Li and X. Ban (2019). Characterization of ridesplitting based on observed data: A case study
of Chengdu, China. Transportation Research Part C: Emerging Technologies 100: 330-353.
Magaloni, B. (2019). La Victimización en el Transporte en la Ciudad de México y la Zona Metropolitana.
Report, Stanford Poverty Violence Governance Lab.
Márquez, L., J. V. Alfonso A and J. C. Poveda (2019). In-vehicle crowding: Integrating tangible attributes,
attitudes, and perceptions in a choice context between BRT and metro. Transportation Research Part A:
Policy and Practice 130: 452-465.
Narayanan, S., E. Chaniotakis and C. Antoniou (2020). Shared autonomous vehicle services: A
comprehensive review. Transportation Research Part C: Emerging Technologies 111: 255-293.
Nie, Y. (2017). How can the taxi industry survive the tide of ridesourcing? Evidence from Shenzhen, China.
Transportation Research Part C: Emerging Technologies 79: 242-256.
ONU Mujeres (2018). Encuesta sobre la violencia sexual en el transporte y otros espacios públicos en la
Ciudad de Mexico. Report (in Spanish), UN Women.
Pajecki, R., F. Ahmed, X. Qu, X. Zheng, Y. Yang and S. Easa (2019). Estimating Passenger Car Equivalent of
Heavy Vehicles at Roundabout Entry Using Micro-Traffic Simulation. Frontiers in Built Environment 5(77).
Rayle, L., D. Dai, N. Chan, R. Cervero and S. Shaheen (2016). Just a better taxi? A survey-based comparison
of taxis, transit, and ridesourcing services in San Francisco. Transport Policy 45: 168178.
Shaheen, S. (2018). Shared Mobility: The Potential of Ridehailing and Pooling. Three Revolutions. Island
Press, Washington, DC: 55-76.
Shaheen, S. and A. Cohen (2019). Shared ride services in North America: definitions, impacts, and the
future of pooling. Transport Reviews 39(4): 427-442.
Shaheen, S., A. Cohen and I. Zohdy (2016). Shared Mobility: Current Practices and Guiding Principles.
Report FHWA-HOP-16-022, Federal Highway Administration (FHWA).
Sterling, D. (2018). Three revolutions: steering automated, shared and electric vehicles to a better future.
Island Press.
Tang, B.-J., X.-Y. Li, B. Yu and Y.-M. Wei (2019). How app-based ride-hailing services influence travel
behavior: An empirical study from China. International Journal of Sustainable Transportation: 1-15.
Tirachini, A. (2019). Ride-hailing, travel behaviour and sustainable mobility: an international review.
Tirachini, A. and C. Antoniou (2020). The economics of automated public transport: Effects on operator
cost, travel time, fare and subsidy. Economics of Transportation 21: 100151.
Tirachini, A. and M. del Río (2019). Ride-hailing in Santiago de Chile: users’ characterisation and effects on
travel behaviour. Transport Policy 82: 46-57.
Tirachini, A. and A. Gomez-Lobo (2020). Does ride-hailing increase or decrease vehicle kilometers traveled
(VKT)? A simulation approach for Santiago de Chile. International Journal of Sustainable Transportation
14:3(3): 187-204.
TRB (2010). Highway Capacity Manual. Transportation Research Board, National Research Council,
Washington D.C.
Tyrinopoulos, Y. and C. Antoniou (2008). Public transit user satisfaction: Variability and policy implications.
Transport Policy 15(4): 260-272.
Wardman, M. and G. A. Whelan (2011). Twenty years of rail crowding valuation studies: evidence and
lessons from British experience. Transport Reviews 31(3): 379-398.
Wenzel, T., C. Rames, E. Kontou and A. Henao (2019). Travel and energy implications of ridesourcing
service in Austin, Texas. Transportation Research Part D: Transport and Environment 70: 18-34.
Zha, L., Y. Yin and H. Yang (2016). Economic analysis of ride-sourcing markets. Transportation Research
Part C: Emerging Technologies 71: 249-266.
Zhou, J., L. Rilett and E. Jones (2019). Estimating Passenger Car Equivalent using the HCM-6 PCE
Methodology on Four-Lane Level Freeway Segments in Western U.S. Transportation Research Record
2673(11): 529-545.
... The introduction and adoption of commercial AVs are likely to reduce the need for households to own cars by way of an increase in ride-sharing services (e.g., SAVs) (Clements & Kockelman, 2017;Krueger et al., 2016;Tirachini et al., 2020). Fagnant and Kockelman (2014) reported that each SAV can serve 31-41 passengers per day and therefore many people can do away with owning a car. ...
... Impact on travel distance/VMT (Narayanan et al., 2020) Trip length: -15% to +14%, VMT: -45% to +89% (Gelauff et al., 2019) 5 -25% increase in VMT (Fagnant & Kockelman, 2014) Up to 10% increase in travel distance (Fagnant & Kockelman, 2015) 2 -9% increase in VMT (Zhang et al., 2015) 15.3 -62.3% increase in daily VMT Median VMT increase of 26.5 miles per household, total VMT increase of 13.3% (Loeb & Kockelman, 2019) 6.05 -14.2% increase in empty VMT per SAV 2 -10% increase in VMT (Tirachini et al., 2020) VKT increase of SAV: 7 to 10 km/passenger, VKT increase of buses: 0.4 to 1.1 km/passenger (Childress et al., 2015) 11 -20% more empty VMT by SAVs (Loeb et al., 2018) SAEV on average generate 19.6 -31.5% more vacant VMT Personal AV has a 2.5% lower VMT than a personal conventional vehicle (Harper et al., 2016) 2 -14% increase in annual VMT (Ma et al., 2017) 15% increase in VMT (Carrese et al., 2019) 100% penetration of ride-sharing reduces VMT up to 19% (Auld et al., 2018) 42% increase in travel distance (Alam & Habib, 2018) 15% (20%) share of SAV increases VKT by 1.73% (14%) (Hörl, 2017) 28.01% and 30.57% empty VMT in Taxi and taxi pool, respectively for 1000 AVs on the fleet (Zhang & Guhathakurta, 2017) 5-14% VMT increase (Arbib & Seba, 2017) VMT increased by 50% in 2030 over 2015 ...
Full-text available
The article discusses the short, medium, and long-term effects of Autonomous Vehicles (AVs) on the urban transportation and environment by means of a systematic review of the extant literature on the subject matter. A corpus of 130 articles was collected from multiple sources using selected keywords. The review critically analyzes key findings of these papers in the light of a SWOT (Strength, Weakness, Opportunity, and Threat) analysis. Although the technology remains to be commercially deployed, broad consensus is found in the literature. First, AV would influence urban transportation and human mobility by reducing vehicle ownership, public and active travel, Vehicle Miles Traveled, traffic delay and congestion, travel costs, and by increasing accessibility, mobility, and revenue generation for commercial operators. Second, AVs would have long-term effects by encouraging dispersed urban development, reducing parking demand, and enhancing network capacity. Third, AVs would reduce energy consumption and protect the environment by reducing Greenhouse Gas emissions. Fourth, AVs would reduce traffic crashes involving human errors and increase the convenience and productivity of passengers by facilitating for multitasking. However, most people are very concerned about personal safety, security, and privacy. Finally, the study identifies critical research gaps and advances priority directions for further research.
... In principle, the more shared vehicles are utilized, the bigger the size of the fleet required to serve the demand, and the greater the negative impact on traffic congestion. Tirachini et al. (2020) show that by increasing the size of taxis and allowing multiple passengers to share their ride higher service efficiency can be achieved, and the total mileage can be greatly reduced. Nevertheless, this approach also poses new and challenging optimisation problems, and appropriate optimisation techniques must be developed to support decision-making processes, such as to promptly match passengers with vehicles or to efficiently update vehicle routes in a dynamic setting. ...
Full-text available
The growing number and complexity of modern megalopolises, where several millions of inhabitants request efficient transport services, pose colossal challenges to urban mobility. To capture the ever-increasing demand for mobility without further deteriorating the traffic congestion, it is essential that the resources available are used as efficiently as possible. Besides, the traditional means of transport (train, metro, bus, taxi), each responds to a particular segment of the global demand for mobility. Nowadays, transport planners can take advantage of the progress in information technologies and optimisation methods to design modern services that integrate and coordinate different means of transport. These services are potentially capable of capturing additional segments of mobility demand and, as an outcome, reduce the usage of private vehicles. Building upon these general ideas, a growing number of researchers have studied various forms of transport flexibility as well as the integration among different means of transport. This survey provides an overview of the trends emerging from contributions from the operational research literature on urban passenger transportation. We have analysed the literature according to the dimensions of flexibility and integration of the transport service studied. For each of the application areas identified, we convey the main trends studied, summarise the most relevant solution approaches and outline some open research directions that deserve particular attention.
... For example, the increased use of autonomous vehicles could lead to increased vehicle miles travelled and a rebound in car use, offsetting the potential energy and emission savings [54]. Similarly, the proliferation of ride-hailing services could increase congestion and undermine public transit [55]. Smart mobility technologies could also have social equity implications, as they may not be accessible to all, leading to a digital divide in transportation [56]. ...
Full-text available
Transportation systems globally face challenges related to congestion, decreased quality of life, limited accessibility, increased harmful emissions and costs, growing use of private cars and in some cases lack of intra and intermodal integration. Smart Mobility is believed to be a solution to some of these challenges by providing comprehensive and intelligent mobility services, decreasing transportation costs, promoting safety, and combating pollution and traffic congestion. Despite this potential, there is still uncertainty surrounding what smart mobility is and whether it is moving toward improving the quality of life and making cities more sustainable. To address this gap, this paper conducts a bibliometric review of 3223 Web of Science Core Collection-indexed documents to provide a comprehensive understanding of smart mobility research. The findings reveal a lack of multi-disciplinary approaches in previous studies with a strong emphasis on technological aspects and limited social or economic considerations in current research. The review identifies four distinct periods of smart mobility research, with recent interest sparked by advancements in big data, deep learning, artificial intelligence, and real-time technologies in transport systems. However, there is a dearth of research on smart mobility in developing countries, where urban populations are rapidly increasing. Thus, the review proposes a research agenda to address the current gaps in knowledge. Furthermore, the review provides an updated and integrated definition of smart mobility as the use of advanced technologies, such as the Internet of Things (IoT), big data analytics, and artificial intelligence, to improve transportation efficiency, mobility for all, and sustainability while safeguarding the quality of life. The primary challenge for smart mobility is the co-evolution with existing transport systems, making further research on integration with these systems and real-time technologies essential for advancing smart mobility research. The paper’s main contribution is an integrated conceptualisation of smart mobility research and novel research topics that build on this unified base.
The digital sharing economy is commonly seen as a promising circular consumption model that could potentially deliver environmental benefits through more efficient use of existing product stocks. Yet whether sharing is indeed more environmentally benign than prevalent consumption models and what features shape platforms’ sustainability remains unclear. To address this knowledge gap, we conduct a systematic literature review of empirical peer reviewed and conference proceeding publications. We screen over 2200 studies and compile a dataset of over 150 empirical studies, and consolidate reported results on the environmental impacts of the sharing economy. We find that sharing is not inherently more environmentally benign and that the type of resource shared, platforms’ size, logistic operations, as well as the ways in which sharing affects users’ consumption more broadly have an impact over the environmental outcomes. Sharing goods is generally associated with better environmental outcomes compared to shared accommodations or mobility, with shared scooters and ride-hailing emerging as particularly prone to negative environmental outcomes. Contrary to previous suggestions, resource ownership structure (centralized vs. peer-to-peer) does not seem to be a good proxy for environmental performance. As sharing becomes intertwined with urban spaces and more cities become interested in facilitating sharing, we argue that research and policy should examine how platform operations and changes in consumer behavior can be optimized for sustainability, steering the sharing economy towards more environmental paths.
Shared mobility services belong to the group of Public Transport (PT) services, and they can be divided into two categories: scheduled and on-demand. Line services are characterized by scheduled timetables, fixed routes and stops. Instead, on-demand services are characterized by variable vehicle-routing and timetables considering changes in the mobility demand, in order to satisfy users’ requests. In the last decades, innovative shared transport services, i.e., Demand Responsive Shared Transport services (DRSTs) have been implemented in addition to the traditional on-demand services, i.e., taxi and rental car with driver (NCC). The introduction of these new services is part of the legal framework linked to traditional PT, not adapted and detailed for DRSTs, making difficult to fully exploit all potentialities associated with these new transport services. The implementation should require a specific regulation able to define the operational features of the service and to avoid that DRSTs contrast or overlap with PT services. On that basis, this work proposes a Multi-Level Perspective methodology in order to provide a legal framework and analyze the criticalities linked to the introduction and implementation of DRSTs, mainly due to regulatory shortcomings. The paper focuses on the Italian legislation which is characterized by a gap for the implementation of DRSTs. At the regional level, these services have been partially defined and regulated, within the limits imposed by national legislation. Therefore, a comparative analysis is presented by analyzing several case studies (i.e., Piemonte, Valle d’Aosta, Friuli Venezia Giulia and Lombardia). The results of the research will constitute a practical tool to implement DRSTs in a permanent way and within the public transport supply.KeywordsDemand Responsive Shared Transport services (DRSTs)LegislationLocal Public Transport (LPT)Multi-Level Perspective methodologySustainable Mobility
The Paris Agreement set global energy conservation and emission reduction standards in 2016, and the ambitious goal of carbon neutrality must be achieved by 2050. However, the conflicts between environmental protection, economic development and social fairness constrain the development of sustainable transportation systems. At the same time, the automotive industry believes that large-scale application of autonomous driving technology is expected by 2050. The development and application of autonomous driving technology can provide a new solution to improve the sustainability of the transportation system. Meanwhile, autonomous vehicles can significantly enhance the mobility of people with disabilities and seniors and reduce the transportation cost for poverty people, improving the transportation fairness of the entire society. This letter discusses the impact of autonomous driving technology on sustainable transportation systems from the perspectives of social fairness and economic responsibility, as a brief summary of the third part of Distributed/Decentralized Hybrid Workshop on Sustainability for Transportation and Logistics (DHW-STL).
Full-text available
We analyse the sources of economies and diseconomies of scale in On-Demand Ridepooling (ODRP), disentangling three effects: when demand grows, average costs are reduced due to i) a larger fleet that diminishes waiting and walking times (Mohring Effect), and ii) matching users with more similar routes (Better-matching Effect). A counter-balance force (Extra-detour Effect), occurs when iii) the number of passengers per vehicle increases and users face longer detours. At low demand levels, there is little sharing and the Mohring effect prevails; as demand grows, more passengers per vehicle push for the Extra-detour Effect to dominate; eventually, vehicles run at capacity, and the Better-matching Effect prevails. The last two effects are specific to ODRP as the routes are not fixed but adapted online. Our simulations show that considering both users' and operators’ costs, scale economies prevail, and that ODRP with human-driven vehicles and walks allowed has total costs similar to door-to-door systems with driverless vehicles.
Full-text available
It is currently unknown in which city environments, automated vehicles could be deployed at reasonable speeds, given safety concerns. We analytically and numerically assess the impact of automation for optimal vehicle size, service frequency, fare, subsidy and degree of economies of scale, by developing a model that is applied for electric vehicles, with data from Chile and Germany, taken as illustrative examples of developed and developing countries. Automation scenarios include cases with partial driving cost savings and reduced running speed for automated vehicles. We find that a potential reduction in vehicle operating cost due to automation benefits operators, through a reduction of operator costs, and also benefits public transport users, through a reduction on waiting times and on the optimal fare per trip. The optimal subsidy per trip is also reduced. The benefits of vehicle automation are greater in countries where drivers' salaries are larger.
Full-text available
The actions of autonomous vehicle manufacturers and related industrial partners, as well as the interest from policy makers and researchers, point towards the likely initial deployment of autonomous vehicles as shared autonomous mobility services. Numerous studies are lately being published regarding Shared Autonomous Vehicle (SAV) applications and hence, it is imperative to have a comprehensive outlook, consolidating the existing knowledge base. This work comprehensively consolidates studies in the rapidly emerging field of SAV. The primary focus is the comprehensive review of the foreseen impacts, which are categorised into seven groups, namely (i) Traffic & Safety, (ii) Travel behaviour, (iii) Economy, (iv) Transport supply, (v) Land–use, (vi) Environment & (vii) Governance. Pertinently, an SAV typology is presented and the components involved in modelling SAV services are described. Issues relating to the expected demand patterns and a required suitable policy framework are explicitly discussed.
Full-text available
A discussion of the sustainability and travel behaviour impacts of ride-hailing is provided, based on an extensive literature review of studies from both developed and developing countries. The effects of ride-hailing on vehicle-kilometres travelled (VKT) and traffic externalities such as congestion, pollution and crashes are analysed. Modal substitution, user characterisation and induced travel outputs are also examined. A summary of findings follows. On the one hand, ride-hailing improves the comfort and security of riders for several types of trips and increases mobility for car-free households and for people with physical and cognitive limitations. Ride-hailing has the potential to be more efficient for rider-driver matching than street-hailing. Ride-hailing is expected to reduce parking requirements, shifting attention towards curb management. On the other hand, results on the degree of complementarity and substitution between ride-hailing and public transport and on the impact of ride-hailing on VKT are mixed; however, there is a tendency from studies with updated data to show that the ride-hailing substitution effect of public transport is stronger than the complementarity effect in several cities and that ride-hailing has incremented motorised traffic and congestion. Early evidence on the impact of ride-hailing on the environment and energy consumption is also concerning. A longer-term assessment must estimate the ride-hailing effect on car ownership. A social welfare analysis that accounts for both the benefits and costs of ride-hailing remains unexplored. The relevance of shared rides in a scenario with mobility-as-a-service subscription packages and automated vehicles is also highlighted.
Full-text available
We examine potential improvements to public transport systems induced by the autonomous vehicle technology (AVT). To do so, we study a feeder system that operates on-demand in an idealized local zone, and the design of a trunk system that operates over a more general city model and with traditional lines. It is shown that the AVT encourages larger fleets of smaller vehicles that follow more direct routes, when compared with the traditional technology (TT). In both sub-systems, the total savings induced by the AVT reach up to one third of TT’s costs. Congestion could increase by a marginal amount.
Full-text available
In this paper, an in-depth examination of the use of ride-hailing (ridesourcing) in Santiago de Chile is presented based on data from an intercept survey implemented across the city in 2017. First, a sociodemographic analysis of ride-hailing users, usage habits, and trip characteristics is introduced, including a discussion of the substitution and complementarity of ride-hailing with existing public transport. It is found that (i) ride-hailing is mostly used for occasional trips, (ii) the modes most substituted by ride-hailing are public transport and traditional taxis, and (iii) for every ride-hailing rider that combines with public transport, there are 11 riders that substitute public transport. Generalised ordinal logit models are estimated; these show that (iv) the probability of sharing a (non-pooled) ride-hailing trip decreases with the household income of riders and increases for leisure trips, and that (v) the monthly frequency of ride-hailing use is larger among more affluent and younger travellers. Car availability is not statistically significant to explain the frequency of ride-hailing use when age and income are controlled; this result differs from previous ride-hailing studies. We position our findings in this extant literature and discuss the policy implications of our results to the regulation of ride-hailing services in Chile.
Full-text available
Early research has documented significant growth in ride-hailing services worldwide and allied benefits. However, growing evidence of their negative externalities are leading policymakers to adopt a more conservative stance. Despite demonstrated socioeconomic benefits and consumer surplus worth billions of dollars, cities are choosing to curb these services in a bid to mitigate first order urban mobility problems. To inform these decisions, we study how the absence of ride-hailing services affects congestion levels in three major Indian cities. Using rich real-time traffic and route trajectory data from Google Maps, we show that in all the three cities, periods of ride-hailing unavailability see a discernible drop in travel time. The effects are largest for the most congested regions during the busiest hours, which see 10.1 - 14.8 percent reduction in travel times. Additionally, we provide suggestive evidence for some of the mechanisms behind the observed effects, including deadheading, substitution with public transit and inducing people to drive longer routes that avoid congested segments. These results suggest that despite their paltry modal share, ride-hailing vehicles are contributing significantly to congestion in the cities studied, and quantify the maximum travel time gains that can be expected on curbing them.
Full-text available
Passenger Car Equivalent (PCE) is a unit used to represent the impact of a large vehicle on a road by expressing it as the number of equivalent passenger vehicles. This paper focuses on estimating the PCE of various sized heavy vehicles in roundabouts with respect to different entry flow rates. A single-lane roundabout was tested under predefined mixed traffic and demand scenarios in VISSIM micro-simulation environments. The individual and group behaviour of four separate heavy-vehicle types were tested: single-unit trucks, buses, small semitrailers, and large semitrailers. The obtained PCE values were found to be on average lower than those suggested in the United States guidelines for roundabouts. The estimated PCE values for heavy vehicles in mixed traffic conditions are 1.30 for single unit trucks, 1.40 for small semitrailers, 1.60 for buses, and 1.70 for large semitrailers. Additional factors such as varying inflow (balanced, unbalanced, and congested traffic) show direct influences on the PCE values. The PCE value under these conditions ranged from 1.25 to 1.75 for smaller vehicles (single-unit trucks, buses, and small semitrailers) and 1.45 - 2.10 for larger heavy vehicles (large semitrailers). A general equation was developed based on the data to relate vehicle proportions and heavy-vehicle reduction factors that would be useful for professionals to analyze the operational performance of roundabouts with better accuracy.
Understanding in-vehicle crowding is crucial to improving public transportation service levels. Although in-vehicle crowding has usually been studied as a tangible attribute, previous research has shown that the standing passenger density cannot fully explain the experience of crowding. In the present work, we used the hybrid discrete choice modeling approach in order to provide a richer explanation of in-vehicle crowding in a choice context between Bus Rapid Transit (BRT) and Metro, using Bogotá as a case of study. Modeling results showed that perceived discomfort, perceived insecurity, waiting time, fare, travel time, and its interaction with the positive attitude towards crowding, are the main variables in explaining the choice process. We found that the positive attitude towards crowding affects the perception of the standing passengers’ density level. We demonstrated that this attitude depends on non-transportation information since people who live in larger spaces, as well as people who live in apartments, have a greater positive attitude towards crowding. We obtained crowding multipliers in regarding the users’ attitude for different levels of overcrowding and analyze several policies related to latent variables.
The passenger car equivalent (PCE) of a truck is used to account for the presence of trucks in the Highway Capacity Manual (HCM). The HCM-6 employed an equivalency capacity methodology to estimate PCE. It is hypothesized in this paper that the HCM-6 PCE values are not appropriate for the western U.S., which consistently experiences truck percentages higher than 25%. Furthermore, the HCM PCE procedure assumes that truck and passenger cars travel at the same desired free-flow speed on level terrain. However, many heavy trucks in the western U.S. are governed through the use of speed limiters so that their speeds are considerably less than the speed limit. Thirdly, the HCM-6 PCEs are based on the freeways having three lanes per direction, which might not be appropriate for the freeways with two lanes per direction that predominate in the rural sections of the western U.S. Lastly, the trucks used in the HCM-6 simulation might not be representative of the empirical trucks observed on rural freeways in western states. This paper examines these effects on PCEs using data from I-80 in western Nebraska. The PCEs were estimated using the HCM-6 equal-capacity method and VISSIM 9.0 simulation data under (1) the HCM-6 conditions and (2) the Nebraska empirical conditions. It was found that the PCEs recommended in HCM-6 underestimate the effects of trucks on four-lane level freeway segments that experience high truck percentages having large differences in free-flow speed distributions, and which have different truck lengths.
Even as ride-hailing has become ubiquitous in most urban areas, its impacts on individual travel are still unclear. This includes limited knowledge of demand characteristics (especially for pooled rides), travel modes being substituted, types of activities being accessed, as well as possible trip induction effects. The current study contributes to this knowledge gap by investigating ride-hailing experience, frequency, and trip characteristics through two multi-dimensional models estimated using data from the Dallas-Fort Worth Metropolitan Area. Ride-hailing adoption and usage are modeled as functions of unobserved lifestyle stochastic latent constructs, observed transportation-related choices, and sociodemographic variables. The results point to low residential location density and people’s privacy concerns as the main deterrents to pooled ride-hailing adoption, with non-Hispanic Whites being more privacy sensitive than individuals of other ethnicities. Further, our results suggest a need for policies that discourage the substitution of short-distance “walkable” trips by ride-hailing, and a need for low cost and well-integrated multi-modal systems to avoid substitution of transit trips by this mode.