The fallacy of one-way car sharing

(This all began a few months ago, when a friend in Austin and I discussed a one-way rental scheme there. Months have passed, he’s moved, and one-way car sharing still isn’t open to the public in Austin. Or anywhere else in the hemisphere. And, outside a couple isolated cases, I’m not sure it will ever work. I distilled the argument, a bit, in a comment at the Transport Politic, and promised to flesh it out. Here’s my best attempt …)

One of the most frequently asked questions for those of us in the car sharing industry is “when are you going to have one-way rentals.” There’s no good answer. Yes, car sharing is a relatively young industry: Communauto is celebrating 15 years in Montreal, most other Car Sharing Organizations (CSOs) in the western hemisphere are 10 or younger. While the technological advances during that time have been rapid, from paper log books and lock boxes to iPhone reservations and remote unlocks, everyone has been vexed by one-way rentals. Supposedly some very smart people (at MIT IIRC) tried to build an algorithm to allow for one-way rentals, and it failed miserably. So, with one minor exception, if you take out a car sharing vehicle, you have to return it to the same space.

The one exception is with Car2Go (not to be confused with a similarly named CSO in Israel), which is “operating” in Ulm, Germany and Austin, Texas. Why Ulm and Austin? Well, Car2Go is backed by Daimler, and has a fleet of Smart Cars in each city. I know very little about the operation in Ulm (or the city itself) other than it is a small, dense, European city. As far as Austin, I know two things. First, the system there is not open to the public. Second, it seems to have backing from the municipal government and the University of Texas, at least as far as parking, which is why it might not fail. Might.

There are two ways to figure out why one-way rentals do not work for car sharing. One is to take a CSO, its members, vehicles, and a defined area, and try to create an model which takes in to account car usage, parking, times of day and fees per mile or minute to see if it, well, works. That is very complicated and, even if it proved successful (so far it has not) is still only a model. The second method, however, is to take a step back and look at some of the underlying factors which would create a workable one-way car share, and how these mesh with such a program. Doing this, it becomes quite clear that one-way car sharing will never really work, no matter how many GPS-enabled cell phones there are.

So, let’s take that step back. In North America, one-way rentals would seem to be based on cities with high density, i.e. cities where, once a car was parked, it would not be long before another driver needed it. We can use a very good proxy for this: cities which are existing major car sharing markets. These are, in the United States, Boston, New York, Philadelphia, Washington, D.C., Chicago, San Francisco, and to a somewhat lesser extent, Portland and Seattle; and in Canada, Vancouver, Toronto and Montreal. (All have at least 300 shared cars on the road, except Portland and Seattle, which have about 200.)

What do all these cities have in common? They all have the soft factors which, in my opinion, are supportive of car sharing: the availability (or lack thereof) and cost of parking, the frequency, reliability and speed of a transit network, and the prevalence of urban congestion. (See this post about car sharing for a longer discussion of these three factors.)

Why won’t one-way car sharing work in these cities? One word: parking.

Imagine starting something like this in a city like Boston (a stand in for our dense car sharing cities because I am poaching most of the next few paragraphs off of an email I wrote a while back). No one would use it. The reason Zipcar is successful (as are other CSOs in other cities) is that when you return your car you are guaranteed a parking spot. If you are going from Cambridge to the South End, you take the T, because it is faster (or marginally slower) than driving, cheaper, less aggravating, and less likely to experience a traffic jam. (And if it is delayed you can walk—you’re not wedded to your car.) If time is of the essence, you take a taxi; as long as you aren’t going right there and back it’s probably cheaper than a shared car (you pay for when you use it, driving or not, so if you are going to a party, it’s cheaper to that the T there and a cab home). You use Zipcar for trips to the grocery store when you don’t want to haul groceries on the T, trips to Ikea &c., trips to the suburbs or places to which the T doesn’t run and to which a cab would be abhorrently expensive (anything within spitting distance of 128).

Now, imagine the one-way car sharing scenario. You pick up a car near your house in Cambridge, and drive to the South End. It works well because it is an easy trip. But there are two problems on two different scales. On a small scale, you get to the South End, and start looking for parking. Most likely, any time you saved driving you lose trying to find a space. Maybe a firm comes up with a way to send the location of spaces to your phone (I worked for one of these for a while, and it did not work out), but there is still a lot more demand than supply. So circling the block eats up any time savings you may have enjoyed. And there’s no way that this service, at $300+ per space per month, goes out and buys enough spaces around the city that you could always find one open and convenient. It might work if you dynamically base the price of the destination based on demand, but then high-demand locations will cost more than taxicabs.

Based on the density of cities like Boston alone, one-way car sharing would work. Except that it is impossible that the parking could be worked out. You’d almost necessarily have more parking spaces than cars, which is both an economic and social detriment: parking, especially empty, is antithetical to density and walkability. (Most CSOs have as many parking spaces as cars; some have fewer. If a CSO has 15 cars assigned to a lot and knows that 99.5% of the time no fewer than 2 cars will be checked out, and lower pricing for off hours can encourage this, they can buy fewer spaces than they have cars. At $300 a space, this is a nice thing to be able to do.)

The other problem is large scale, one of congestion. Imagine that you somehow solve the parking conundrum this takes off. Imagine that whenever you need a car, you can get one and drop it off wherever you need to, and pay a few dollars for its use. It would be great. Everyone would use it. And, there’s the problem. And all of the sudden, tens of thousands of people who used to be carried below grade on subway lines (and some buses) would be clogging the streets in their little two-seater cars. Even if they were in such eight-foot long vehicles, they’d take up a lot of space. In cities like Boston, New York and San Francisco, adding a few percent to the already at-capacity roads throws the system in to gridlock. You’d pretty quickly lose whatever time advantage you had over mass transit, trips would be more expensive than transit (or, if they were short enough, still more expensive than walking), and it’s not pleasurable to sit in stopped traffic in a city.

Finally, cars will necessarily flow to certain places at certain times of day. From residential areas to office areas, from offices to restaurants to entertainment districts. If there are too many, you have to move them. But cars aren’t like shared bikes. You can’t send one buy out with a truck, load ten of them up in one location, and ship them off somewhere else. You need a driver for each, and that gets costly.

On the other end of the spectrum, there are less dense American cities, some of which do have shared cars on the street. I’m not speaking of universities which pay for a couple of shared cars, but cities with smaller shared car fleets on their streets, like Madison, Denver/Boulder, Minneapolis/Saint Paul, Atlanta and Pittsburgh. We’ll use these are stand-ins for the less-dense American city, which may not have the aforementioned factors in place.

Here our stand-in will be Minneapolis and Saint Paul, because, uh, again, I can coöpt much of an existing email in to this post. In the Twin Cities parking is pretty easy. But for a scheme like this to work, you’d need so many cars in order for them to be within walking distance of enough people. (Add to that the fact that in the neighborhoods where it works best—Uptown, the University of Minnesota, the downtowns—parking is an issue.) So there are a few neighborhoods where it might work okay, they are not very well connected, and many people living there ride their bikes anyway. Andy pretty much every city I can think of falls in to this category.

In addition, cities such as the Twin Cities have many services which are only available in the suburbs. For instance, in Boston, Chicago and San Francisco, there are REIs and Apple Stores easily accessible by transit. In the Twin Cities, they’re in the suburbs. So for a lot of trips, you have to drive somewhere, leave the car in a suburban parking lot, and need it to get back. One-way rentals don’t really apply here.

The only cities I can think of that might be able to solve the parking issues yet have dense enough areas to support many cars are mid-sized cities with huge, urban college camuses, decent transit and middling parking issues, and the right “clientele” for the service. Basically, Austin and Madison. The colleges can be strong-armed in to giving up enough parking to make it viable on the campus, and the geography might work out otherwise. (Of course, would this program be used by broke college students when free options like biking or walking abound?) It’s definitely a might; I’d still be surprised if it works.

You may notice I have not said a word about the logistics of the whole charade yet. I am rather well-qualified to speak to them, and it would scare the hell out of me. Basically, what happens when a car goes off to never-never land—a part of the city which might be in the city limits (or the sharing limits), but where no one really wants to drive from. Either it sits there generating no revenue until someone wants it, or you have to dispatch someone a folding bike to get it. Either way costs staff time and mileage.

What happens when someone parks it and leaves it in an underground garage with no GPS reception. You know, like Whole Foods (in Austin, which has underground parking)? (I assume they’ll have sensors there, which would be somewhere I’d assume a lot of the cars would wind up, but you’d have people walking up and down the aisles in the lot, or have the cars in a special parking space. Still, now you’ve spent a lot of money wiring every garage in the city for connectivity.) No one can find the car, you have to have someone call the previous user (and good luck reaching the jet-set type anymore, many of us don’t answer calls promptly), and try to find the car. CSOs know where our cars are—they are returned to their spots 99.9+ percent of the time.

Finally, take the following scenario: you live near the outskirts of the drop-off area and work eight miles away, also near the outskirts of the drop-off area. You pick up a car one day and park it at your house, where no one else is likely to use it. Then you wake up, drive it to work (you pay for what you drive, so 15 minutes costs you $3) and leave it at work, where, likely, no one will want to use it. At the end of the day, you drive it home, another $3, and leave it there. All of the sudden, you’ve paid $6 for a 16 mile round-trip commute, gas and insurance included. If someone else uses the car, you take a bus or bike in to town, pick up another one, and do it again. If this happens every couple of weeks, it’s a minor inconvenience,  and your total commuting costs might be $120 a month. It’s not a bad deal for you, but would bankrupt anyone trying to run the thing.

A couple more tidbits:

A lot of car sharers often make reservations in advance. Like, weeks in advance. Every Tuesday evening they drive to the grocery store, or their great aunt’s house, or the climbing gym. The car is where the car is and they know it will be there. With one-way rentals you don’t have any assurance a car will be where a car will be. Thus, you throw out the segment of the market which has reserved in advance. (Without throwing around any insider information, let’s say this market is below 50%, but still significant.) You could have a dual system, where some cars are round-trippers and some are one-ways, but then you have more overhead and member confusion (we’ve found that simplicity, in car sharing, is a virtue). And the logistics—people parking in the wrong spaces; people taking one-way cars on round trips, would be an administrative nightmare. Or, you could have a system robust enough that there’d almost always be a car where you wanted it. But I think I’ve given several reasons that such a system is unlikely. And you’d still need a system whereby people could take cars for longer trips and pay for the non-driving time in between to guarantee the car would be theirs.

Okay, so what about bike sharing? It works, right? Yes, but bikes are smaller than cars. A lot smaller. You can put fifteen bikes on a sidewalk without disrupting the traffic flow of pedestrians or vehicles. Try doing that with one car, let alone a dozen.

To sum up, a one-way car sharing system only works in an area well-served by transit. However, to get from one area served by transit to another, you don’t really need car sharing. Car sharing fills a specific niche where transit is too slow or inconvenient, taxicabs too expensive, and cycling too impractical. One-way car sharing would be like taxis without drivers. Except when taxis aren’t carrying a fare, they are doing one of three things: they are either parked in an out-of-the-way location, driving to find another fare, or idling (generally at a cab stand or high-traffic area) with a driver in the seat. Taxis never have to find parking. Take away the cabbie, and the system fails. As would, in my opinion, one-way car sharing.

*****

Want to find a CSO near you? CarSharing.net‘s list a pretty good list. If you are in the Twin Cities, HOURCAR is fantastic, although, full disclosure, I do work for them.

6 thoughts on “The fallacy of one-way car sharing

  1. The city of Ulm was the natural choice for the first car2go pilot: besides an enduring commitment to technological innovations, the city of Ulm hosts one of Daimler’s leading research centers and the city has shown great willingness to cooperate with the project.

    The city of Austin is well known as a fast growing cultural and economic center of Texas. Its size, demographics, and open-minded lifestyle made it a perfect choice for car2go.

    The first pilot phase in the southern German city of Ulm has been successful beyond expectations. In October 2008, the first car2go fleet of 50 vehicles was put into operation. In just two weeks, more than 500 customers had registered with car2go – a number we had not anticipated. On March 27, 2009, the number of cars was increased to 200, and the Ulm pilot was opened to the general public. In the first 100 days, 10,000 customers had registered. Today, the car2go vehicles are rented 500 to 1,000 times per day by any of 15.000 customers. 90 percent of them take advantage of the unique service allowing members to use car2go spontaneously without booking in advance and for one-way trips.

    We want to gain additional knowledge about fleet management. As car2go provides an unparalleled high degree of customer flexibility – remember the “open end driving” and “one way driving” features – one cannot determine in advance where vehicles will be located from day to day or even hour to hour. Thus, optimizing the fleet distribution, reservation algorithms, cleaning and maintenance processes (to give some examples) is a fascinating challenge.

    Please feel free to check out the availability and distribution of cars on our website or on the iphone application:

    Website city of Ulm, Germany (started in October 2008):
    http://www.car2go.com/portal/ulm/page/home.faces

    Website city of Austin, Texas (started in November 2009):
    http://www.car2go.com/portal/austin/page/home.faces

    Information about iphone app (soon available for car2go in Austin): http://www.car2read.com/austin/en/around/lifestyle/new-iphone-app/

    Your car2go team

    More information at
    http://www.car2go.com
    http://twitter.com/car2goNA

  2. You've done a very exhaustive analysis of the challenges facing one-way carsharing (actually one-way and on-demand) but maybe overstated the issues. In Austin there will be a combination of reserved parking (in garages and lots) and floating fleet, so finding a car won't be as hopeless as you suggest.

    I think it's important to recognize car2go, like the proposed Paris Autolib program, especially given that it only offers smart cars, is not some sort of enhanced carsharing (although one can certainly use it for a round trip) but really going after a new group of users – highly convenience oriented people. And keep in mind a 10 minute trip in car2go costs only a little more than a bus ticket so if you had a meeting within the main operating zone you could drive there, be assured of parking and have a very good chance a car would be available for the return trip.

    The project in Ulm has been going for more than a year and Daimler hasn't pulled the plug on it so they must learning something from it. I've written my own 2¢ about car2go in my blog at http://www.carsharing.us

  3. Great post. You mentioned you had trouble with car insurance. Have you ever thought about how your car insurance provider works for you? Or if the one that you have chosen is the right pick? When I wanted to get car insurance for myself I had to research the best providers of car insurance in Barrington RI, which is the city where I'm from. I would definitely recommend requesting multiple quotes from different providers before making a decision.

    – Amelia

Leave a Reply