« portland: another challenging chart | Main | comment of the week »


Feed You can follow this conversation by subscribing to the comment feed for this post.


Don't the trains in Japan issue certificates of tardiness to commuters when running late, which said commuters can present to the boss as an excuse?

At any rate, I can think of one sneaky reason for doing that: Once you tie a transit agencies revenues to performance, you now have a better reason to upgrade your infrastructure, so that tardiness is reduced.

At any rate, how would this work for persons with passes?


Here in Melbourne we saw an isolated case of the train operator paying a passenger the taxi fare she incurred due to an extended service disruption.


Jeffrey Bridgman

I can confirm what EngineerScotty said. They (not sure to what extent, but the company I used did this) issue certificates of tardiness to commuters when running late. If fact, that was the only was to get an excused tardy in the morning from my school.


I've never heard of such a thing as "on time or we pay". And frankly, I don't think it's a very good idea, either. The first time a bus driver gets in a wreck because he was pressured to be "on time or else" will be the last time you'll hear an agency make such a promise.


Transport for London has such a policy (15 minutes on Underground, 30 minutes on Overground):


Christopher Parker

Metra's studies show what their riders value above all else was reliability. I imagine that is true most everywhere.

It is very possible that if not collecting fare can get the buses back on time there may be more revenue value in that than in the missed fares(over the long term, from higher ridership). Of course an on-board proof of payment system would eliminate the time spent collecting in the first place . . .

In the old days of privately operated transit systems, bus drivers had a variety of tricks for staying on time, or slowing down if they were running early. Going though all the yellows. Or not. Stopping for passengers running after the bus. Or not. Stopping the bus next to the younger energetic riders, who (in rush hour) will leap on board or next to the elderly ones (off peak) because then the youngsters will politely let them on first anyway. Etc.


I agree with rhywun. There was a case where a pizza delivery driver killed someone, and the grieving family sued the pizza company, claiming that the "X minutes or it's free" guarantee made the company culpable. They won.

Also if an agency finds they are losing too much revenue they will just make the timetables less stringent, which makes the system less efficient.

Wesley Zhao

Great post. One of the better articles I've seen. I especially love the idea of on-time or it's free and how it helps the bus get back on schedule.


One big problem with the "if your bus was late to pick you up, you don't pay the fare" issue is it addresses only one kind of late: if the bus takes longer than it should to get me to my destination, I am just as late but I already paid my fare.

And amen to Pantheon: the typical reaction to any penalty for late transit will be the agency's putting unreasonable slack into the system to prevent late arrivals.
Like if you punish the airlines for flights that arrive late, they could just add an hour to the predicted length of all their flights and they'd almost always be early.

Stupid may not be limitless, but it runs pretty deep.


Actually, the airlines already did this a while back.

Alon Levy

Actually, putting extra slack in timetables is precisely how agencies stick to schedule. It improves reliability at the expense of average speed.


My express limited-stops bus has only one intermediate stop between where I catch it and the city terminus. It takes 5-6 minutes in peak hour to do the 3.5km trip. Yet according to the timetable, it should take 12 minutes......... while the all stops bus following the same route takes around 12 minutes, and should take 20. Did someone mention something about extra slack in the timetables?


As people have noted, the airlines responded to pressure to be on time by adding slack to their schedules, so they end up "on time" no matter how sloppy they are getting out of the gate. Also, they refuse to compensate you if their delays are due to "weather", and "weather" tends to cover a variety of sins. Since many factors making transit vehicles late are outside of the transit agencies' control, it would be a bad idea to implement such a policy without an exception for bad weather, unforseeable traffic situations, etc... which would in turn leave sufficient leeway for the agency to cover it's ass in most cases, making the policy ineffective and the customers cynical about the agency.

Boston's MBTA has a policy where if they are more than 30 minutes late, your fare is free. But 30 minutes is more than generous to the agency. Even when the bus driver doesn't bother to show up for his shift and the buses only run every 20 minutes, I would still be only 20 minutes late.

Louis Haywood

I agree with SpyOne, and would add that riders at the beginning of the route would rarely get on free, as buses generally leave their starting points on time, if there is any slack at all in the schedule for recovery.

I don't think equity is all that important though, and so I agree with Jarrett that not collecting fares is more important, because if the bus becomes free, then the riders who boarded earlier may get back on time, which I think they would appreciate. But do you then begin collecting fares again once you are <10 minutes late? Would this not put you back over the threshold, thus making the bus fare-free again?

Fare collection would simply oscillate on/off/on/off, as the bus became marginally on time and marginally late.

Ethan Tucker

When I visited a friend in Geneva in the late '90s I was told that the city's hyper-efficient bus system was in part due to a policy of fining bus drivers if they were more than 90 seconds late at key stops. But perhaps that was just an urban myth!


Well, it's been often said that Mussolini made the trains run on time...

Alon Levy

He didn't. It's just a myth.


Well, if a fascist dictator like Benito couldn't keep transit on schedule... :)

Richard Masoner

Santa Clara, CA VTA 522 "Rapid" service doesn't have scheduled times along its route, just the times it starts service at either end of its route. If it's ahead of schedule, the bus keeps going. These buses also have traffic signal prioritization for several intersections along its route.

Caltrain has something like 96% on time performance, which is actually down significantly from prior years because of the high number of "track incidents" (suicides and cars on at-grade crossings) last year.

Nathan Williams

I'd suggest x be something like 3 times the published headway or ten minutes, whichever is less.
You're a funny man, you know that? I believe the absolute-most-frequent routes on the MBTA have a published 6-minute headway (key routes like the #39, during rush hour), so the idea that 3*headway is ever less than ten is pretty hilarious.


For very frequent urban routes, a better GPS-driven scheme would be based on actual headway (time between consecutive trips, i.e. maximum wait time) rather than timetable

With GPS, one the agency log actual performance of the bus, make long-term statistics of delays and waitings for departure with respect to time of day and week and use tese stats to make schedules more efficient (adding slack only in times of day when it is needed).

And if the whole system overgrows some critical size, it makes sense to couple GPS with data network and send real-time information to dispatching centre so the agency can quickly respond to any irregularity.

Alon Levy
Caltrain has something like 96% on time performance

Only if you define being 5 minutes late as being on time.


5 minutes seems quite generous. As others have mentioned, the MBTA does 30 minutes. Ive personally asked for and received my money back two times, once on the commuter rail ($7.75 and once on a bus $1.50). The downside is it took 2-3 months for the ticket to arrive in the mail (not cash, but stored value)

Now, you mention letting people on for free instead. I've seen bus drivers do this often, most recently on Wednesday in Vegas on the deuce. The bus pulled up to a stop, there were 20 or so people, and the driver let them in for free. Considering they were all tourists....probably a good idea.

The problem for the transit agency making this standard practice is
1) It screws over people who got on earlier and payed a fare AND annoys those with passes.
2) They lose more money. Letting everyone in for free, you get zero revenue. Making people ask for a rebate means most people won't go through the process....maybe 2 out of 50 riders will go through the online process.

Stuart Donovan

I think the idea of "performance incentives" is worth pursing. In my experience transit agencies and service providers (whether public or private) have a fairly bad attitude to customer service.

This may reflect how transit systems tend to be natural monopolies, in which the consumer has only limited choice especially in the short-term, i.e. once you are their waiting at the stop it is often difficult to change your plans.

They could also use these "offers" as a point of difference, as with the quality and/or price guarantees shops provide with goods/services. And it is similar to airline practices of offering compensation ...

I think you would want a couple:
1. Late service - Ride free. Sure the people onboard have already paid, but they subsequently get the benefit of the faster trip, quid pro quo.
2. No service - guaranteed taxi ride home. This would require 24/7 helpline, GPS tagged services, and excellent comms with every service.

Where drivers continue straight past passengers may be the most challenging "service failing" to address with direct customer guarantees.

An effective complaints resolution system coupled with more generous time-tables (as others have noted) may be a better option in this situation?

Louis Haywood

1. Late service - Ride free. Sure the people onboard have already paid, but they subsequently get the benefit of the faster trip, quid pro quo.

Again, the problem is that the bus would make up time when the riders got on for free. So the next passenger may board on-time (barely within the 10-minute threshold, say 9.5 minutes late). If she pays with cash, the next customers may get picked up 10 minutes late. So these customer would board for free. The bus makes up 30 seconds maybe, so the next customer would pay, and so on and so forth.

Lauri Kangas

At least Stockholm, Oslo and the Skåne region in Sweden offer travel guarantees. The conditions are not identical, but generally they will pay for alternative transportation up to a certain limit of the real if expected delay is over 20 minutes. Stockholm probably has the most favourable conditions and even provides them in English: http://www.sl.se/templates/Page.aspx?id=4664

I'm not sure that diverting funds from actual service is necessarily a problem. If unreliable service has a price, it is an incentive to be realiable. That said you can only offer these guarantees if you are already running an efficient system. It is not realistic to assume an underfunded and/or underperforming system can be improved just by adding sanctions.


Didn't the Bush Administration consider a program called NPLB (No Passenger Left Behind), which was designed to help underfunded transit agencies improve their peformance by subjecting them for penalties for failing to keep to schedules?


Alon Levy
Again, the problem is that the bus would make up time when the riders got on for free. So the next passenger may board on-time (barely within the 10-minute threshold, say 9.5 minutes late).

You're assuming two questionable things:

1. If the bus gets just under 10 minutes late again, it gets to charge customers.

2. Payment is at boarding.

Problem #2 is a general one and needs to fixed as soon as possible everywhere.

Problem #1 would be fixed by saying that once a bus is really late, it doesn't get to charge people again until it's exactly on time, say under 1 minute late. Then if it falls behind schedule again it gets to charge until it hits 10 minutes, and so on.

Louis Haywood

Alon, great points. I agree waiting to charge until 1 minute late is good enough.

Payment should be made by touchcard as a goal, combined with proof-of-payment for all-doors boarding. This is widely done in Europe, e.g. the Parisian RATP bus network, but also in small-town Italy POP is done with tickets. But we digress into system change and capital-driven wishes, whereas we're talking operational/policy changes.

NextBus technology (cellphones, smartphones and/or kiosks) would allow for customer information, and thus a larger margin for error before reparations are demanded.

Finally, how to prove that someone would have taken the bus if it had been on time? If I know a certain bus is running 30 minutes late, and I was planning on taking a taxi anyway, then couldn't I just write to the transit company and have them pay, even though I'm not really a customer at all?



I have great respect for your input, but your solution to problem #1 is insane. How do you explain to your customers why yesterday when their bus was 5 minutes late they didn't have to pay, yet today it is 5 minutes late and they do. "Because", the driver says, "yesterday I was making up time from being really late, and today I'm not". And how does the customer verify the honesty of the driver? It sounds like a recipe for confusion, chaos, and angry customers. The very opposite of what this policy intended.

Alon Levy

The verification doesn't have to be done at the customer level; there has to be some manager input in either case. (What if the driver just charged people anyway?)

The purpose of the policy is to avoid charging for rides when the bus is 10 minutes late or more. The bit about not charging at 5 minutes when recovering from 10 minutes' lateness is an extra bonus for customers as well as a way of discouraging straddling the 10-minute boundary.


The only way I would support these crazy ideas is in a system that is completely separate from traffic. Which pretty much eliminates buses. As for subways, sure, *in theory* any lateness is their fault, but then again... there are so many reasons trains are late and most of them are completely outside the control of the operators. Also, the main reason airlines are so accommodating to reparations is that they are in competition with one another. They want to retain your business. Not so with nearly every public transit operator in the world. If the NYC MTA makes me 5 or 25 minutes late to work, it's not like I'm gonna take the taxi from now on.

The comments to this entry are closed.

the firm

Jarrett is now in ...

Related Posts Plugin for WordPress, Blogger...
Related Posts Plugin for WordPress, Blogger...