Where Einstein Meets Edison

A Balancing Act: the Future of Car Sharing and Driving-as-a-Service

A Balancing Act: the Future of Car Sharing and Driving-as-a-Service

May 2, 2010

The gap left between pay-for-use transportation (public transit, cabs, and rental cars) and car ownership is being filled by car sharing services. The number of people who rely on driving as a service is growing quickly, but the companies that ultimately succeed will be the ones that provide the most convenient service while managing their fleet most effectively.

In 1999, Robin Chase and Antje Danielson decided that they could create a car sharing service by integrating wireless technology into automobiles. A production vehicle was combined with GPS and wireless communication and Zipcar, the first non-cell phone wireless company, was born. Now with 360,000 active users, 74% of the North American car sharing market, an expected eight-fold increase in sales to $1B within the decade, and a potential IPO in 2010, Zipcar has paved the way for driving-as-a-service (DaaS) to replace car ownership.

In several ways, Zipcar’s service is similar to that offered by a traditional rental car company. The customer reserves a car, picks the car up at a particular location, and returns the car to that same location. But, there are two key differences. First, cars are not grouped at distribution centers, but instead spread out across a city. Second, customers can rent the car for hourly increments or by the day.

The economics of DaaS for the consumer are compelling. Fuel accounts for only twenty percent of the total cost of driving, according to the 2010 AAA’s annual Your Driving Costs report. AAA estimates that a sedan owner’s average cost per mile driven is 56.6 cents. Given that the average driver covers 12,000 miles per year, the average cost of driving is $9/hour, which also happens to be about the hourly rate at Zipcar. Moreover, this does not account for time spent by car owners to maintain their vehicles, the cost of parking, or decreased driving due to the increased visibility of the price of driving.

Table 1: True Cost of Driving. Calculations shown in Appendix

Car sharing organizations are able to profit because the average cost per mile drops significantly as a car is driven more, and the fixed costs of ownership are diluted. For example, per mile insurance costs decrease from 10 cents to 5 cents when annual driving is doubled from 10k miles to 20k miles.

Despite these promising calculations, DaaS companies face some unique challenges. Vandalism raises a concern due to the familiar dilemma known as the tragedy of the commons. In an interview, Chase raised the case of Vélib’, a French bicycle rental program. Vandalism rates for Vélib’ in Paris were higher than expected and five times higher than those in the smaller French city of Lyon. Zipcar made the strategic decision to name each of their vehicles to personalize them and minimize abuse. In theory, people are less likely to damage “Maki” (parked in my apartment garage right now) than they are an unpersonified Toyota Matrix.

Two-way car rental, like Zipcar’s, has additional drawbacks when compared to car ownership. Dropping off your rental car at the same place where you picked it up requires round trips. This makes some trips longer and more expensive. Additionally, services like Zipcar aren’t designed to accommodate consumers that want to make short trips. Rental is done at hourly increments and reservations are often required. A system that eliminates these faults and provides drivers unrestricted transportation with more convenience than vehicle owners could revolutionize personal mobility. The concept, known as Mobility-on-Demand (MoD), is being researched by the MIT MediaLab’s Smart Cities Group, whose efforts won the 2009 Buckminster Fuller Challenge.

MoD is a vehicle sharing concept that aims to “maximize mobility and dramatically reduce congestion and pollution through energy and land-use efficiency”. The key elements of MoD are one-way, reservationless rentals, lightweight electric vehicles, and user incentives that self-balance the system. Vélib’s bicycle program follows the one-way rental model; you pick up a bike from a station, you’re charged by the time you use it, and you return it to any Vélib’ station. Paris plans to implement an automotive version of the project, Autolib’. Autolib’ will utilize charging stations as pickup and drop-off points for its electric vehicles. car2go, a Daimler subsidiary, is currently using one-way rentals with a fleet of internal combustion Smart Fortwo microcars. car2go’s model allows users to start trips spontaneously and return cars to parking spaces throughout the city. car2go has seen the user base of its one year old fleet in Ulm, Germany grow to nearly 18,000 members and their second fleet in Austin, Texas fleet is scheduled to go live to the public on May 21st.

For one-way DaaS companies to excel, they must utilize a system whereby users maintain a balanced distribution of vehicles throughout the city. According to Ryan Chin, PhD candidate in the MIT MediaLab’sSmart Cities Group, the company that outperforms competitors in MoD — always having a vehicle available within a reasonable time, using the least number of vehicles for the largest number of users — will be the one that “builds a better engine based on historic and current data than anyone else.” Vélib’ undertook fleet balancing manually, by loading bikes on a truck and relocating them to rebalance, but a more elegant approach is to incentivize users to balance the load through mechanisms like dynamic pricing.

Dynamic pricing aims to balance one-way rental fleets by charging drivers different amounts to pick-up and drop-off vehicles depending on the location and current supply of vehicles near it. Dimitris Papanikolaou, the Smart Cities Group expert on pricing, uses models to predict the effects of pricing and distribution on the balance of the system. In an interview, Papanikolaou emphasized that understanding how different pricing policies affect user behavior is a complex problem and that dynamic pricing decisions can be too risky to evaluate after implementation. Papanikolaou explains, “System Dynamics, allows us to numerically explain the effect of dynamic pricing on decision making and identify the boundaries of acceptable performance for different demand pattern scenarios by conducting a sensitivity analysis.” Essentially, simulation predicts where imbalances will occur and optimizes the system. The information obtained from simulation enables informed recommendations for fleet distribution and pricing policies that optimizes stability and profitability. The system balance challenge of MoD systems has encouraged innovative thinking, but it certainly is not a new issue. The Smart Cities Group also draws from similar issues faced by traditional rental car companies, freight and moving companies, and even airlines.

Bill Mitchell, Director of the Smart Cities Group, sums up the emerging shift in mobility as a change from the existing view of providing vehicles and physically moving people around and suggests that, “In the future, it’s much more about sophisticated management algorithms and making the best use of the resources that you have through information technology.” There are certainly challenges in balancing one-way car sharing systems, but the brightest minds of industry and academia are eager to conquer them.


This article was written by Justin Jensen, a Guest Writer for the MIT Entrepreneurship Review, with input from Robin Chase, Dimitris Papanikolaou, and Ryan Chin.


Hourly Cost of Driving Calculation

  • Average Fixed Cost Per Year = $5976/yr [1]
  • Average Variable Cost Per Mile = $0.1674/mi [1]
  • Parked Rate = 90% [2]
  • Utilization Rate = 1 – Parked Rate = 1 – .9 = .1
  • Average Operating Hours = Hours per Year * Utilization Rate = 8760h * .1 = 876h
  • Average Miles Driven = 12,000mi [3]
  • Average Speed = Average Miles Driven / Average Operating Hours = 12,000mi / 876h = 13.7mph
  • Depreciation Adjustment = +$0.0514 / Mile above 15,000 [1]

Annual Cost = Average Fixed Cost Per Year + Average Variable Cost Per Mile * Annual Miles Driven + Depreciation Adjustment

  • Example: Annual Cost = $5976/yr + $0.1674/mi * 12,000mi + $0.0514 / Mile above 15,000 * (12,000 – 15,000) = $7831

Hourly Cost = Annual Cost * Average Speed / Annual Miles Driven

  • Example: Hourly Cost = $7831 * 13.7mph / 12,000mi = $8.94


1. http://www.aaaexchange.com/Assets/Files/201048935480.Driving Costs 2010.pdf

2. http://mitpress.mit.edu/catalog/item/default.asp?ttype=2&tid=12044

3. http://www.epa.gov/OMS/climate/420f05004.htm#step3

Article image is © Flickr User: Anyhoo. MITER thanks Anyhoo for kindly allowing it to use one of his images for this article. Image source: http://www.flickr.com/photos/anyhoo/262565293/in/set-72057594109505173/.