Friday, February 24, 2012

The cost of new riders

Some nice person Reddited my post on park-and-rides. So far it's only gotten one comment, but that comment is extremely insightful:

This article leaves out the obvious reason that has long been the reason park and rides are added to projects I work on in the AA phase - goosing up ridership. Federal funding rating has long been based on cost per new rider, and if you can't get costs down, you can always get ridership up (in the models, at least) by adding park and rides.

FTA has now published new guidelines (PDF) that emphasize transit supportive land use at the same level as cost effectiveness (which is also no longer based on cost per new rider, as well) so hopefully we'll start to see a shift in the next few years as projects conceived and developed under these new guidelines move into engineering and construction.

This may be more important than the congestion-fighting mentality I've described in recent posts in accounting for the proliferation of park-and-ride lots and garages in transit systems new and old. Both ideas flow from, and reinforce, the Kotkinist idea that the ideal commute is a solo drive in a personal car from the driveway of a single-family home to a parking lot.

I've heard the complaints over the years about the cost-benefit requirements for various Federal transit capital funding programs. In principle I'm in favor of cost-benefit analysis. Who wouldn't want to know whether they're getting their money's worth?

The main complaint I heard was that there was a double standard: transit projects were being asked to justify their funding in ways that most road projects would have completely flopped. And yeah, it's hypocritical *and* idiotic to care about one set of costs but not another.

I had thought that that was the only real problem with these cost-benefit analyses, and it seemed like it had been fairly well covered by other bloggers. I assumed that the "benefit" part of the formula represented something that more or less corresponded to an actual benefit. Boy was I wrong.

The metric was referred to as the "cost per new rider," but that's a bit vague. The official term is "incremental cost per incremental rider." To estimate it, a transit planner comes up with ridership estimates for the entire system if the project is built and for a no-build scenario, and divides the difference by the amortized cost of the project.

Does this make sense to anyone but a bureaucrat? When would you ever do a return on investment analysis where the "return" was simply providing a service? If you're deciding whether to build a factory, would you base your decision on the cost per unit? No, you'd look at the cost per dollar of profit. If you're a sane government official deciding whether to build a soup kitchen, would you look at the cost of providing a meal? No, you'd look at the cost of keeping people from going hungry.

Similarly, if you're deciding whether to build a transit facility, why would you give a shit how many people use it? Your ultimate goal is not to get people to ride the thing. It's to reduce pollution, or carnage, or to increase efficiency, improve society or provide access. If your project can do all those things without anyone actually riding it, that's better!

In general, yes it's true that a commuter who's on a train isn't driving, and therefore not polluting, using gasoline or running people over. But that's where the park-and-rides come in. If a commuter is only on your train for half their commute, they're still behind the wheel for the other half, polluting, burning gas and putting lives in danger the whole way. And if they then stay in the car for errands on the way home, that's more driving, plus they're shopping at car-oriented stores.

What's worse is that a park-and-ride can induce more driving, because it can make a car-oriented suburban life affordable and convenient for people who work downtown but wouldn't pay to drive all the way downtown, either in pure dollar terms or in convenience cost. If a park-and-ride encourages someone to move from an apartment in a walkable neighborhood to a house in a car-dependent suburb, that may make Joel Kotkin happy, but it sucks for the environment.

That's what's missing from the Federal metric. What is the incremental cost per incremental rider in terms of transit trips still half driven? What is the cost in terms of induced driving for errands? In terms of induced driving by spouses and children? Nowhere.

So what about these new guidelines? Well, I'm not too impressed. They talk a lot about parking - presumably from the point of view that the more parking there is at transit stations the better, but it's actually ambiguous.

What I'd like to see is some attempt to quantify how much less driving people would do with the project in place than without it. Of course, you'd also have to take into account competing road projects that are being funded at the same time.


Alon said...

Don't know about you, but I think that transit should be built with the goal of getting people to ride it.

Pantograph Trolleypole said...

I agree with Alon, getting the most people possible to ride it should be the goal, if we believe in the benefits of transit.

Another issue that comes up though is that previously the issue was that a new built project was compared against a no build alternative that was never going to get built. And then you measured the benefits to the NEW riders, not ALL the riders. As the FTA joke goes, the lines that best pass the cost-effectiveness test are those that don't have any stations along the line.

Seattle enviros actually got the new light rail corridor to minimize park and ride construction, because they felt that it spurs sprawl. If you go back through the Seattle Transit Blog history I'm sure you'll find lots of commentary on that subject specifically.

This is also why the Livermore extension in the Bay Area is as perverse as it gets. Instead of spending $3B to get into downtown Livermore, which could be a new job center and walkable place, they are going to end it in the center of the freeway and build parking garages for the people in the Central Valley to drive halfway and then take the train to Oakland and San Francisco.

Alon said...

...or not spending $3 billion on Livermore and instead using it to get to the Richmond District. No big deal, it's only 100,000 daily riders stuck on a bunch of slow buses, passing through neighborhoods that are some of the densest in the US outside New York. There's no possible way they could make use of rapid transit.

Susan Pantell said...

Actually, the proposed FTA guidelines would count "limited availability of parking" as a positive under the Land Use criteria. They specify the rating factors for the CBD based on parking costs and spaces, with "high" being >$16 and <0.2 spaces per employee and "low" being <$4 and >0.5 spaces. They do not say how they would evaluate non-CBD areas, and I plan to ask them about that.

Peter McFerrin said...

While he was at Berkeley under Bob Cervero, Michael Duncan (now of UNC Charlotte) did a great paper on the ridership generation effects of TOD vs. park-and-rides. Long story short, the further out from the CBD you go, the less effective TODs become and the more effective park-and-rides are.

Instead of trying to create artificial TOD islands in distant suburbs, it might be make more sense to focus efforts on increasing population density and pedestrian-oriented retail within walking distance of transit stations in inner suburbs and the outer parts of central cities. In the case of BART, this means beefing up densities near stations like Rockridge and North Berkeley.

arcady said...

You could boost the ridership on the Livermore extension if you somehow convinced the whole Central Valley to commute to SF on BART using Livermore as a park and ride. But is that really a better outcome than a line that has a much lower ridership but doesn't cause a whole lot more driving? Or just not building the line at all?

Alon Levy said...

Peter, what you're missing is why extend BART so far out in the first place. If the only kind of travel it can generate is peak commutes to San Francisco and if population density is going to stay low, there's no point spending all that money on rapid transit. There are places in the Bay Area that need rapid transit more than Milpitas, Livermore, Bay Point, Pleasanton, and Millbrae.

Cap'n Transit said...

Alon and Jeff, yes, in general the more riders the better, but there's no point in being absolutist about it.

Thanks for the clarification, Susan.

Peter, that's a very interesting article; thanks for telling us about it. Maybe you don't remember the part where he says, "The model predicts that connecting buses at either
the origin or destination will significantly increase ridership,
which shows that bus feeders can be an effective tool along with parking in extending the reach of a rail system." Also, "ln(auto tt/bart tt)" is one of the major predictors, indicating that highway removal is another potential tool to increase ridership.

I don't think there's anything "artificial" about removing height restrictions or parking requirements near train stations, but I agree that the low-hanging fruit is near the CBD, which brings me back to where I started this: the Northern Branch.

alai said...

There's a parking garage under construction at MacArthur station at a cost of $40m for 400 spaces. It seems that the park-and-ride must be maintained at any cost.

busplanner said...

Re: Park and Rides (A comment on a number of recent posts where this topic was raised.)Part 1

In surveys, interviews, focus groups, and outreach sessions with individuals who do use or might use public transit by choice, not by necessity, the four most common responses to why do you use transit/what would it take for you to use transit (assuming it goes where you need to go for your daily commute) are:





Everything else (cost, time of trip, comfort of trip, schedule reliability, other reasons) pales in comparison. When you bore down into the reason frequency is so important, the answer (said in a variety of ways) is flexibility. (What if I oversleep? What if I need to be at work earlier than normal? What if my kid’s school bus is late? - AM. What if I have to work overtime? What if I need to leave work early for a medical appointment? What if I need to leave work early for my kid’s soccer match? - PM) These individuals, many of whom grew up in a suburban environment, are accustomed to getting in a car or minivan and going. They don’t want to rely on a public transit service that operates every thirty minutes or less frequently in peak times and not at all or in an extremely limited fashion in off-peak times.

To offer frequent service, the public transit operator needs to have a sufficient ridership level to support frequent service. (Operating empty buses and trains draws frequent criticism from those paying for public transit through their taxes but who do not use it and the politicians they support.) However, in much of middle and outer suburbia, the population density does not support the provision of frequent public transit. Put another way, park and rides consolidate the users along one or more corridors in a manner that permits frequent bus and/or rail service attractive to these choice users and relatively efficient for the public transit operator to provide.

Why aren’t shuttles used to collect riders and bring them to the train station or bus stop? Simply put, in most cases, the demand per trip for the shuttle cannot justify the cost of the shuttle (labor, fuel, maintenance). Also, it is difficult to design shuttle routes that meet the schedules of the buses or trains in an efficient manner.

This does not mean that shuttle routes or local bus routes cannot work in specific circumstances. For example, the Metropark Train Station in New Jersey has five shuttle bus routes and a local bus route (NJ Transit routes 801-805 and 62 respectively) feeding commuters into the station. These routes work because of relatively high density housing near the station and, on most of these bus routes, office complexes or other employment sites along the route that transport reverse commuters arriving at the train station as well. Still, the station has two large parking decks and additional surface parking.

busplanner said...

On Park and Rides - Part 2

And that brings us to why planners are proposing park-rides along the proposed Northern Branch extension. First, one needs to ask, “What is the purpose of the Northern Branch extension of the Hudson-Bergen Light Rail Line?” It is to serve an already developed corridor where, while there is lots of one-seat ride bus service to the Port Authority Bus Terminal in midtown Manhattan and some one-seat ride bus service to the George Washington Bridge Bus Station in northern Manhattan, there is no bus or rail service to the thousands of jobs that have developed along the Hudson-River Waterfront in Hoboken and Jersey City, New Jersey. People are driving to these jobs and the new service is designed to capture many of these people much closer to home.
What about local buses feeding these stations? Many of these stations do have connecting bus services. Some of the bus service is frequent. But park and rides are also needed.

Are the park and rides properly sized? Only time will tell. For example, NJ Transit has opened three major park and rides in recent years. The Wayne Transit Center (serving both bus and rail routes to midtown Manhattan and Hoboken) filled on day one. The Ramsey Route 17 Rail Park-Ride and the Montclair State Rail Park-Ride have so much unused space that NJ Transit leased out portions of these park-rides to a car dealership for storage and Montclair State University for student parking respectively.

Finally, this leads us to BRT on the Tappan Zee Bridge. Could the reversible lane on the Tappan Zee be converted to a BRT/HOV lane almost overnight? Probably, yes. But one needs BRT routes to run over the bridge. And that is a problem. Origins and destinations on each side of the bridge are more spread out than the extreme concentration of jobs that exists in Manhattan. On the more residential Rockland side, one probably needs park-rides to collect bus riders in a cost-effective manner. On the Westchester side one needs a major distribution system to get these bus riders to the various office campuses.

Cap'n Transit said...

Ah, Busplanner! I think you missed my post on the idea that transit ridership has any direct relationship to population density.

Matthew said...

The Northern branch is located in a place that used to see many crossing trolley routes (and there are still NJTransit buses). There were and are no park and rides for trolleys and buses. The fundamentals haven't changed. It does not need park and rides. The corridor passes through the centers of Tenafly and Englewood. While it skirts Leonia and Palisades Park, both of these towns are small in area and nothing is really more than a 15 minute walk away. These are streetcar suburbs, not quite the sprawling suburbs like Wayne.

NJTransit operates several heavily used commuter bus services in nearby parallel corridors. We're talking buses that have a 20 minute OFF-peak frequency, 14 hours worth of express service, and overnight span. The buses do not require park and rides, and neither does HBLR.