Call nowBook now
Reigate taxi

01737 471099

Shared Taxis Reigate

Modeling Urban-level Impact of a Shared Taxi
Andrea Paraboschi, Paolo Santi and Carlo Ratti
Taxi systems are being challenged by alternative, emerging services like
Uber, Lyft, and Sidecar, which increasingly offer the option of ride sharing. While the enormous potential of ride sharing has been unveiled in a
number of recent papers, it also raised legitimate concerns about the potentially disruptive impact on other transportation modes. In this paper, we introduce a framework for estimating the urban-level impact of ride sharing
applied to the current taxicab service. First, we extend a representative
economic model of regulated taxi markets to include ride sharing. The
model allows predicting the interactions between demand and supply of a
shared taxi service based on a few representative parameters, and is rooted
on data analytical results. Then, we apply our model to the case study of
the New York taxi market. The analysis highlights the dramatic impact of
the pricing policy and taxi fleet management on the urban-level, systemic
outcomes of a shared taxi system.
A. Paraboschi (Corresponding author)
Senseable City Lab, Massachusetts Institute of Technology, Cambridge, MA,
USA, and Scuola Superiore Sant’Anna, Pisa, ITALY
P. Santi
Senseable City Lab, Massachusetts Institute of Technology, Cambridge, MA,
USA, and Istituto di Informatica e Telematica del CNR, Pisa, ITALY
C. Ratti
Senseable City Lab, Massachusetts Institute of Technology, Cambridge, MA,

1.1 Introduction
Taxi is a common and comfortable point-to-point transport system, introduced in Europe in the early 17-th century and playing a major role in urban transportation since then. The services offered to riders evolved following several major technological developments: the diffusion of
combustion-based engines at the end of 19-th century; adoption of twoway radios to communicate with dispatch offices in the 1940s; computeroptimized vehicle dispatching in the late 1980s, to cite the major ones. In
the last 20 years, GPS technologies and mobile data made it possible to
track the position and the availability of each vehicle (and user) in real
time, improving service performance and opening up new business opportunities.
1.1.1 Many innovations, same business model
Despite the many technological advancements and changes that occurred
in the history of this means of transportation, the taxi system business
model, the hailing experience, and the fare system have remained the
same: the final price of the ride is computed considering an initial amount
of money (flag), plus a variable amount of money derived on the basis of
time/distance travelled, and extras (such as baggage, tips, night extra fares,
highway tolls…), no matter how many people are on-board. Consistently,
in traditional taxi systems a cab can be either “vacant” or “occupied”, and
it can accept trip requests only when in “vacant” state. The result is a massive amount of taxicabs on the streets, about half of which wanders in
search of new passengers1
, which is clearly undesirable for traffic and pollution.
1.1.2 The risk of being disrupted by new entrants
The widespread diffusion of Internet-connected devices and the development of new business models leveraging on the emergent socio-economic
trend of sharing economy [12, 16] are rapidly changing the landscape of
individual, point-to-point transportation in urban environments, making the
taxi only one of the many alternative modes available to citizens. Carsharing services like car2go [2] or Zipcar [20] are gaining popularity
among urban population. People can find an available car with the phone,
ride it, pay it per minute/hour, and leave it for someone else when the trip
1 According to [11], the average occupancy rate of taxicabs in New York City is
about 50%.

is over. Bike and scooter sharing services rely on similar principles. If the
user still prefers to be transported as a passenger, a shared van can be
cheaper than a taxi while moving groups of people [1]. While these services still represent alternative ways to move in the cities, with a different
proposition with respect to taxis, cab companies are increasingly being directly challenged by the so-called TNCs (Transportation Network Companies) like Uber, Lyft, and Sidecar.
TNCs step into the extremely fragmented market of taxicab services by
acting as a centralized Internet-based multisided platform that leverages
mainly on private citizens’ car fleet, re-shapes the user experience, and innovates the pricing mechanism. TNCs take care of the verification of drivers’ licenses, provide virtual assistance, additional insurance, driver-rating
mechanisms and automatic payment processing systems, replacing the
taxicab dispatch central with advanced algorithms. By joining these platforms, the customer is being put at the center of a totally new experience
where hailing a car is as simple as pressing a button on the smartphone.
Furthermore, all payments are electronically managed, an aspect that is
convenient for both parties (passengers avoid to pay by cash, drivers avoid
unpaid rides and the risk of robbery). Thanks to their intrinsic scalability,
business agility, and a different cost structure, companies like Uber are
disrupting the established market, continuously challenging the incumbent
taxicab companies by adding new features and lowering fares. The result,
as shown in Figure 1.1, is a decreasing demand for traditional taxis, and a
consequent drop in the price of taxi medallions after a continuous and uninterrupted growth that lasted for many decades [7].
1.1.3 Ride-sharing
The innovation in the value proposition is increasingly including the notion of ride sharing as a key feature to gradually reduce the fare prices, a
key strategic move for disruptors [5]. The “UberPool” feature, recently
launched by Uber in San Francisco, Paris, New York and Los Angeles, allows a passenger to share the ride with another that is going in the same direction. This service has been described by the company as a new way “to
deliver transportation at lower and lower price points” [19], aiming to start
a virtuous cycle where demand increases, more cash-flow is generated and
money is re-invested in big-data analysis to perfect the server-side dispatch
algorithms and maximize driver utilization rates, which in turn enables further price reductions.
The fact that ride sharing is considered a key feature of innovative
transportation services should be no surprise to urban planners and policy
makers: the evolution of cities in history has been profoundly impacted by the movement of citizens and goods, and resulting emerging features such
as those summarized in the well-known Christaller’s central place theory
[4] clearly hints to the fact that a large fraction of urban trips should be
“shareable”. This intuition has been confirmed by recent studies such as
[14], which unveiled the immense potential of ride sharing in the city of
New York: more than 95% of taxi trips can be shared, with a minimal impact on passenger discomfort2
The immense potential for ride sharing has raised legitimate concerns
regarding the impact of innovative transportation services at urban level. If
not wisely implemented, these services might have undesired effects such
as reduced job opportunities for taxi drivers, lower demand for public
transportation with negative impact on carbon footprint, etc. [8, 13, 15].
Fully addressing these concerns requires performing a comprehensive
study of the impact of ride sharing at urban level, and of its integration
with other transport modes.
Making a first step in this direction is the goal of this paper. More specifically, we extend current urban economic models of regulated taxi markets to include ride sharing. Ride sharing brings a radical transformation
into the market, which becomes an instance of segmented market where
the same good (a vacant taxi) can satisfy two classes of customers: those
requesting a single trip, and those willing to share their trips. Starting from
this model, we build a framework for predicting the interactions between
demand and supply of a shared taxi service based on a few representative
parameters: the market share m of the ride sharing service, the discount
factor d applied to the price of a shared vs. a single ride, and the number N
of taxis in the market. The framework is rooted on the data analytical results of [14], which allows accurately predicting the likelihood of sharing a
taxi ride as a function of the market share m for the city of New York.

2 Passenger discomfort is measured in terms of delay in reaching the destination
vs. the case of a single ride.

Fig. 1.1 San Francisco taxi demand decline: the plot reports the average
number of monthly trips per vehicle. (source: S.F. Municipal Transportation
The application of our framework to the case study of the New York cab
market allows for the first time to quantify the dramatic impact of the pricing policy on the urban-level, systemic outcomes of a shared taxi system.
Recently raised concerns [8, 13, 15] are legitimate: with an ill-designed
pricing policy, ride sharing might actually negatively impact transportation. For instance, the fact that a single taxi can serve multiple customers
might lead to a reduction of the number of taxi drivers quantifiable in
about 16,600 units in the city of New York. On the other hand, a pricing
policy where, defining P as the per-mile price of a ride in the current taxi
market: 1) the average per-mile price of a ride in the shared taxi market is
P; 2) single ride passengers are penalized by paying a per-mile price P′ > P
for the single ride; and 3) shared ride passengers benefit of a reduced price
d · P′ < P, leads to a desirable systemic outcome where the total demand of taxi services is unchanged (implying no negative impact on public transport), the total number of miles travelled by taxis is reduced (including both vacant and occupied trips), the number of taxi drivers is unchanged, and the average income of taxi drivers is increased. The analytical framework presented in this paper can help urban planners and policy makers to better understand the transformations brought along by innovative ride sharing services, and to make informed decisions leading to desirable systemic outcomes. 1.2. Towards an upgrade of the taxicab operating model The current taxicab system needs to have many empty cabs on the street to work properly. The perceived quality of the service for the customers relies in the system’s response time, i.e., the average waiting time to “find” an empty cab. The search for a vehicle can happen in many different ways: a call to the cab company dispatch, walking to a taxi stand, by using the cab company or a third party smartphone or web app (Uber or the like), or by hailing a car on the street. In the following, we present a possible model of shared taxi system based on [6]. We start presenting how demand and supply of taxi service are modeled for a traditional, single ride system. Demand of taxicab service, denoted Q, in regulated markets depends on: where Xs! has been added to account for exogenous factors such as minimum wage. An alternative approach, more consistent with the scenario in which the taxi is owned by a company and the drivers lease the taxi to operate in shifts, is to simply assume that t is a constant, roughly corresponding to the duration of the shift. In the following, we will apply both approaches to study the shared taxi market. 1.2.1 Ride sharing models What happens if taxicabs can be shared with other people going the same way? We first observe that there exist at least two ways of operating a shared taxi system, called static and dynamic ride sharing. In the static model, the requests for a shared ride are collected by the taxi service operator for a short time interval (say, a few minutes), and only trips in the current pool of collected requests are considered for sharing. If two trips from the pool can be shared, they are matched, and a single taxi is dispatched for accommodating both trip requests. From that time on, and until the time at which the last passenger is dropped, the taxi is considered as occupied and not available for further ride sharing (even if there are still available seats onboard). Thus, similarly to traditional taxi systems in the static scenario the taxi can be in one of two possible states: vacant (no passenger onboard) or occupied (one or more passengers onboard). In the dynamic model, taxis can instead be in one of three states: vacant, when there is no passenger onboard; shareable, when there is at least one passenger onboard but seats are still available for sharing; and occupied, when all available seats are occupied. In this model, requests for a shared ride are possibly matched not only with currently unserved requests, but also with already ongoing shared trips being served by shareable taxis. In case a new trip is assigned to a shareable taxi, the driver is informed of the new passenger to pick-up, and a re-route is done to pickup the new passenger, possibly before current passengers are dropped off. Both models have pros and cons. The static model is easier to run and operate, and offers the customer a better travel experience: upon pickup, the customer knows expected travel time to destination (possibly including pickup/drop off of other passengers), and this planned route does not change after departure. On the other hand, the static model is not able to fully exploit potential sharing opportunities offered by partially occupied taxis, as it is instead done by the dynamic model. On the downside, the dynamic model is more complex to run and operate, and undoubtedly offers a lower-quality travel experience to customers, whose arrival time at destination is no longer accurately predictable at pickup time due to possible dynamic re-routing of the taxi. It is interesting to observe that the two main TNCs, Uber and Lyft, are currently operating ride sharing services adopting different approaches: while Uber is developing its algorithms around a dynamic model [18], Lyft is opting for a static one [10]. In the interest of simplicity and presentation clarity, in the following we present a possible model of a static shared taxi system, based on the assumption that no more than two trips can be combined into a shared trip (two-trips static sharing). This choice is consistent with one of the sharing scenarios analyzed in [14], and it is also supported by the results reported in [14] showing that, even in this constrained sharing model, more than 95% of the trips can potentially be shared in the city of New York. 1.2.2 Static sharing taxi system model A shared taxi system shall be analyzed as a market in which the same good (a taxi) is requested by two classes of customers: those requesting an individual ride, and those requesting a shared ride. 1.3 Case studies We now show different applications of the model derived in the previous section. The analysis is referred to the taxi market of New York City. For this market, we have the following parameters, taken from [11] and [6]: 1.3.1 Constant demand The first case study considers a situation in which the regulator is interested in keeping the total demand of taxi service unchanged in the transition from traditional to shared taxi system. This scenario finds its motivation in the fact that increasing the demand of taxi as a result of ride sharing might be considered detrimental by city authorities, since this additional demand might come at the expense of a reduced demand for public transportation services – which should be preferred due to the reduced impact on traffic and pollution. Concerns about the fact that taxi sharing services might reduce the demand for public transportation have been recently expressed in the literature This implies that the number of taxis in the shared market should be reduced with respect to the case of traditional taxi market to keep demand unchanged. The amount of this reduction as a function of the market split is reported in Figure 1.6: as the market share of shared rides increases, the number of taxis needed in the market reduces, up to close to 50% when approaches 1. Notice that reducing the number of taxis might be desirable for reducing traffic; however, reducing of a similar amount the number of taxi drivers might be undesirable from the societal perspective. However, the fact that the revenue generated by a single taxi would increase of a factor suggests a possible way of reducing the negative impact on workforce: reducing the duration of the shift. Let us clarify this point with an example. According to [11], the average duration of a shift in New York is 9.5 hours. Assuming a two-shift taxi utilization scheme Assume for the sake of this example that = 0.5, which implies n! = 1.499. If a single driver operates the taxi during a shift, his/her profit in the shared taxi market would be about 50% higher than in the traditional market. This would make the drivers that remain in the market very happy, but would have the undesirable consequence of leaving about 33% of the taxi drivers out of work, which corresponds to about 16,600 people in the city of New York [11]. Alternatively, one can think of designing a more flexible shift structure, in which taxis and drivers are pooled together. A driver can, say, have a shift composed of two mini-shifts: one on taxi for, say, 4 hours, and one on taxi for, say, 4 more hours. The notion of mini-shift allows a single taxi to be operated for 19 hours daily but this occurs being driven by 2 ∙ n! drivers on the average, instead of 2 as in the traditional taxi market. Let us elaborate more the previous example under this scenario. Suppose the goal of regulators is keeping the overall driver’s profit unchanged. In this case, thanks to the notion of mini-shift it is possible to keep the workforce unchanged, with a total duration of the per-driver shift of about 6.5 hours. Discussion. When collectively considered, the analyses reported in this section show that, independently of whether drivers follow a profit maximizing or a fixed driving time strategy, it is possible to envision scenarios in which the total workforce is not reduced, and drivers actually obtain better conditions receiver higher hourly wages. In both analyzed models, the total miles driven by the taxi fleet would be reduced of a factor with respect to the case of traditional taxi market (e.g., of about 33% when = .5), with corresponding reductions in traffic and pollution generated by the taxi fleet. Also, in order to avoid draining demand for public transportation, we assume that the average price of a shared taxi trip is at least twice as expensive as the single ride mass transport ticket price. Considering that in New York the price of mass transport ticket is 2.50$, and that the average length of a taxi trip is 2.6 miles 1.4 Conclusion In this paper, we have extended an existing economic model of demand and supply of taxicabs considering the implementation of a static ridesharing model, basing our analysis on real data from the city of New York. Our analysis clearly points out how the introduction of taxicab ride-sharing services can produce contrasting effects. This opens for a discussion at the policy making level, where an operational systemic cooperation among medallion owners (being them companies or private drivers) has to be promoted, and both existing pricing strategies and fleets operations need to evolve. The goal of this study is stimulating further analysis and discussion on this topic, highlighting the main variables and interdependences that have to be considered while envisioning the implementation of a taxicab ride sharing service, considering both the demand and supply side dynamics. New Transportation Network Companies, such as Uber, Lyft or Sidecar, are challenging the incumbents by leveraging on big-data intelligence. We believe that the adoption of similar data-intensive systems, complemented by new bold policy decisions, can help traditional yellow cabs to run for many miles more. Acknowledgements Thank you to the Enel Foundation, MIT SMART program, the Center for Complex Engineering Systems (CCES) at King Abdulaziz City for Science and Technology (KACST) and MIT, BBVA, Ericsson, Expo 2015, Ferrovial, the Regional Municipality of Wood Buffalo, Volkswagen Electronics Research Lab, and all the members of the MIT Senseable City Lab Consortium for supporting the research. The authors would like to thank Prof. Frank Levy for reading a preliminary version of the draft and giving several useful suggestions. Andrea Paraboschi would also like to thank the Institute of Management of Scuola Superiore Sant’Anna and Telecom Italia for supporting his studies at MIT. References [1] Bridj. Accessed 15 January 2015. [2] Car2go. Accessed 18 January 2015. [3] Charitou, C. & Markides, C. (2012). Responses to Disruptive Strategic Innovations. MIT Sloan Management Review, 44(2), 55-63. [4] Christaller, W. (1933). Die Zentralen Orte in Suddeutschland, Jena: Fischer Verlag. [5] Christensen, C., Scott, A. & Roth, E.A. (2013) Seeing what’s next: Using the theories of innovation to predict industry change. Boston, MA: Harvard Business School Press. [6] Flores-Guri, D. (2003). An Economic Analysis of Regulated Taxicab Markets. Review of Industrial Organization, 23(3-4), 255-266. [7] Kwong, J. (2014). Report says SF taxis suffering greatly. The Examiner. Accessed 10 January 2015. [8] Lopez, L., Domingos, T., Cadarso, M.A., Zafrilla, J.E. (2014). The Smarter, the Cleaner? Collaborative Footprint: A Further Look to Taxi Sharing. Proc. National Academy of Science, 111(51), E5488 (2014). doi:10.1073/pnas.1420242112. [9] Lyft. Accessed 25 November 2014. [10] Lyft. Lyft Line. Accessed 28 February 2015. [11] NYC Taxi and Limousine Commission (2014). 2014 Taxicab Factbook. New York Taxi and Limousine Commission. Accessed 1 December 2014. [12] Rosenberg, T. (2013). It’s Not Just Nice to Share, It’s the Future. The New York Times. Accessed 4 December 2014. [13] Rubinstein, D. (2014). Economist: Big taxi-sharing study gets it wrong. Capital. Accessed 10 January 2015. [14] Santi, P., Resta, G., Szell, M., Sobolevsky, S., Strogatz, S.H., Ratti, C. (2014). Quantifying the Benefits of Vehicle Pooling with Shareability Networks. Proc. National Academy of Science, 111(37), 13290-13294 (2014). doi:10.1073/pnas.1403657111. [15] Santi, P., Resta, G., Szell, M., Sobolevsky, S., Strogatz, S.H., Ratti, C. (2014). Reply to Lopez et al.: Sustainable Implementation of Taxi Sharing Requires Understanding Systemic Effects. Proc. National Academy of Science, 111(51), E5489 (2014). doi:10.1073/pnas.1421300112 [16] Sundararajan, A. (2013). From Zipcar to the Sharing Economy. Harvard Business Review. Accessed 15 November 2013. [17] Uber. Accessed 2 March 2015. [18] Rinkus, S. (2014). How Uberpool works. Uberblog. Accessed 2 March 2015. [19] Rinkus, S. (2015). Uber’s new BHAG: Uberpool. Uberblog. Accessed 2 March 2015. [20] Zipcar. Accessed 10 January 2015.[/et_pb_text][/et_pb_column][/et_pb_row][/et_pb_section]