Saturday, August 22, 2009

The value of ridership

Back in February, it was all over the news: even with "record numbers of riders," transit agencies were cutting service and raising fares. On Tuesday, Transportation for America released a report giving more details about the service cuts and fare hikes.

Lots of people have been baffled by this, including your Cap'n. We had images of packed buses making their way through cities and money pouring into fareboxes, while guys with green eyeshades still sadly shook their heads: not enough to overcome the declining tax revenues. My first guess was that the agencies simply weren't charging enough. That's a factor, but it's not the main factor.

I finally had a chance to sit down and crunch some numbers from the National Transit Database, and I've figured out that the main factor is that those buses aren't packed. "Record numbers of riders" means that the buses were going by with fourteen people instead of the usual ten.

I took the NTD figures for the 374 bus services "directly offered" (on an individual basis, not under contract to some other entity) for 2007, the latest year we have data for. Some of the systems in the Transportation for America report weren't in there, but I marked all the ones that I could find, 46 of them. On most of the systems identified by T4A as facing service cuts, the average number of passengers was 9.9. The nationwide average number of passengers per bus per revenue mile was 10.8; not too different. The average farebox recovery for the systems facing service cuts was 21%, which is below the nationwide average of 27%.

The Transportation for America authors use some 2008 data; I can't find it on the NTD website. T4A gives some numbers in their report, but some of those are for agencies that run both rail and bus services (maybe ferries too), and that's hard to tease apart. So I figure I'll just do a hypothetical: imagine that the buses were all packed, with forty people on each one. I know that when buses get full, fuel costs and dwell time go up, but for now let's assume that operating expenses stay the same.

With forty people on each bus, farebox recovery goes up to 86% for the services facing cuts, and 94% nationwide. With only twenty people on each bus it still goes up to 43% and 47% respectively, but the systems in Gary and Baton Rouge both make a profit.

So clearly my initial guess was wrong, and the ridership gains were much more modest. It is possible to run a bus without subsidies in any city in the country, if you can get enough people to ride. Of course, that's the big "if"! I'll talk about that in future posts.

If you'd like to play with the data, the spreadsheet is here. Also I'd like to note that some states have no systems facing cuts. For example, no one is stranded in Iowa.


Jarrett at said...

I'm not sure the words match the picture here.

Your y-axis value "passengers per vehicle per revenue mile" is confusing to me. I assume this is the same thing as the standard NTD metric "passengers (or boardings) per revenue mile" since a revenue mile is defined as one vehicle going one mile while available to passengers.

But your narrative talks about this number as though it was an indication of load, i.e. how many people are on the bus. Average load is "passenger MILES per revenue mile" ... You can find that in the NTD too, but it's not what you've charted.

Christopher Parker said...

Actually, average passengers per mile is pretty much meaningless.

Public transit has always suffered from three problems that hurt it's economics:

1. Heavy ridership in rush hour requires capacity that isn't needed the rest of the day. More than 50% of travel happens during just a few hours in the day.

2. Many routes have a big traffic generator (downtown, a subway stop, etc) at one end and then gradually drop their passengers off block by block . . . You can leave full and still have a low average passenger per mile because eventually you have 1 person on board.

3. Paratransit is frightfully expensive and could be distorting the numbers.

If you've got a bus big enough to hold the peak load, you might as well run it for the rest of the day too. It's cheaper to have only 1 big bus than to have 2 buses so that one is smaller for the off-peak service. And labor is the biggest expense anyway.

So what are some solutions?

1. Transit systems need two levels of fares. Commuters can often pay more, so like the airlines, transit can get revenue where it can while still providing cheap off-peak service. Let the market encourage those who can to travel in the off-peak. Off-peak ridership is much more price sensitive in general.

2 Express services for longer trips are nice for those passengers, but they are also nice for the economics of the service. If you have full buses that gradually dwindle, it's better to run some express to some point and then start dropping people and have the others make short trips. You can run less miles with more average passengers and provide them with a faster service.

3. Speeding up service in general makes a huge impact on passengers per mile. In general buses can be *very* slow. Reduce detours, reduce stops, add short jump lanes past traffic, have traffic light pre-preemption and a fare collection system (ie, proof of payment) that doesn't require paying on entry. The same bus can make two trips an hour instead of one, offering a more desirable service!

BruceMcF said...

@Christopher Parker: yes, speeding up bus service does increase patronage ... and also hits the operating ratio on the other side, since both labor costs and capital servicing costs are per hour rather than per revenue mile.

And, yes, off-peak fares are important for the imbalance between capacity to cope with peak demand and using that capacity to service off-peak demand. Cityrail in Sydney has an off-peak discount whereby a return ticket originating off-peak costs 20% more than the one-way on-peak fare.

Cap'n Transit said...

Thanks for the feedback, Jarrett! The Y-axis on the chart is in fact passenger miles per revenue mile; I just mislabeled it.

And yes, Christopher, labor is the biggest expense, at least here in the US. You could even label it as "passenger miles per driver mile." Very informative solutions.

BruceMcF said...

The mislabeling is entirely understandable ... referring to boardings per revenue mile as "passengers" per passenger mile, after all, is a misnomer, even if it is an officially sanctioned misnomer.

Alon Levy said...

Monthly passes are a form of two-level fare, since they're priced to be competitive for people who only use the system to commute. Once you have an incentive to get a monthly pass for your commute, your off-peak fare is zero.