It’s been embarrassing to be traveling in Europe during critical weeks when several states I care about — including Pennsylvania, Oregon, and Illinois — are going through major transit funding crises. But the same crises are coming for much of the US in the next year. So I wanted to lay out everything I could offer in the way of arguments for why US transit deserves funding, why it’s going to need more, and what arguments we can make to win this difficult battle. Bloomberg Citylab published it today.
I’m especially proud of the line, “If you drive to the mall, bus riders subsidize you.” But there’s a lot there and I hope you find it useful.
Again, it’s here.
I agree that both urban and rural areas should decide what they want to prioritize in their respective areas without one side forcing the other to do as they demand. That way, both sides get what they want and they’ll have more incentive to leave each other alone.
However, it should be mentioned that rural areas disproportionately lose money when compared to their urban brethren. Just taking Illinois as an example, the area outside of Chicagoland costs over $2 to maintain for every $1 they send to Springfield (which makes Downstate’s “efforts” to secede and join Indiana all the more laughable).
I bring this up because it must be asked: What would be the best way to fund transit for those specific communities? Rural areas don’t make that much income to begin with (hence the subsidies), but there are plenty of rural folks who still need transportation despite not owning cars…
Do you have a source on the Downstate vs. Chicago claim?
I’m not doubting you, I’d just love to have that study in my back pocket for the future!
And just to verify the specifics, of course.
@Carter, there’s one glaring problem with the idea that “urban and rural areas should decide what they want to prioritize in their respective areas without one side forcing the other to do as they demand.”
Urban areas cannot impose domestic passports on rural residents who might want or need to work or visit urban areas.
Urban areas are wealthier because each acre/hectare of land is far more productive and can support far more activities. Suburbanization happens because of proximity. Farmland becomes subdivisions because the proximity to urban activities makes the land more valuable than the activities on top of it.
In rural areas, land can often only be one thing. Agricultural land can only be agriculture, mines can only be mines and so on. This is why, throughout human history, the arrow of economic mobility is rural -> metropolitan and not the other way around or bidirectional.
Urban and rural areas that go their own way, pay their own freight, or such similar metaphors will consequently lead to each area treating residency as a club good, rather than a public good. Elites in either areas will game these dynamics to leverage wealth and power through the stratification that results.
Good word Jarrett! I wish there were answers easier to arrive at in our fractious nation.
“But as long as we need paid employees on board, to drive either buses or taxis, this hope is a delusion. The cost of passenger transport in the US is mostly the cost of labor,”
“The only car-based technology that could change the equation is the fully driverless vehicle, where no paid employee is needed at all. If robotaxis became so cheap and reliable that they could offer individual rides at the cost of a bus fare,”
“Waymo is also nowhere near offering its service at the price of a transit fare, even if there were space on the streets for all the resulting traffic. For now, its fares are more comparable to Uber and Lyft.”
The solution is beyond obvious – transit agencies need to switch to driverless vehicles. The vast cost reduction would not only remove the fiscal cliff, but allow for more service than today. Waymo might cost as much as Uber when it is one rider, but split that cost 7 or 8 ways and it is not the cost of a bus fare it is CHEAPER than a bus fare – you are paying the same cost for the vehicle but not the cost for the operator. And that is with Waymo’s current prices, which are inflated due to supply and demand.
The next obvious conclusion is to move from a single quarter full bus to 3 or 4 smaller vehicles. This would allow huge increase in frequency (frequency is freedom!) without similar increase in cost. Time and again the largest factor in size of transit ridership is size of transit service – vehicles that come more often or for a longer period of the day get ridden more.
This leads to the final obvious conclusion, which is that driverless vehicles allow transit to be viable at lower densities and wider areas than now. This would greatly reduce the ‘suburb-vs-city’ issue that Jarret raises. Some people in the suburbs who won’t ride (or by extension support) a bus that doesn’t visit their neighborhood or only comes every 30 minutes will ride a bus than comes every 8-10 minutes because it is convenient. Driverless vehicles make this possible.
The non-obvious conclusion is to move from a make-every-stop paradigm to a demand response system. Elevators have had these for a few decades now. It used to be you got on the elevator and then pressed a button to pick your floor – someone going to floor 4 and floor 74 could be together. Now you press the button for your floor first and you are told what elevator to get on. Passengers to floors 4, 6, and 7 are grouped so the elevator returns to pickup other passengers quicker, passengers to floor 74 and 78 are grouped for a faster ride without stops. With frequent driverless vehicles you can put a stip of buttons for every stop down the line at a bus stop. Press the button where you want to go and the system will have the best bus pick you up. A few might pass by, but once you are on you have a much faster ride with fewer intermediate stops. Zero headway and express trips are why cars dominate travel (even in Europe) but a demand response line can take much of that advantage away [the other major advantage of cars is infinite span – but driverless vehicles can allow 24/7 service too].
The largest customers for Waymo and Zoox should be transit agencies. If Jarret does not say this every time he consults he is doing transit agencies a disservice and condemning them to irrelevance when someone takes the Uber/Lyft/Waymo/Boring Co. model and applies it to medium capacity vehicles not passenger cars.
The problem there is that self driving street vehicles just aren’t ready for prime time yet, and waiting around for that will only delay something we’re long overdue on.
I know they are driving around now, but those are much smaller vehicles with a lot more training data associated with them. And even those tend to require human intervention constantly behind the scenes.
There’s also the old “Doorman Fallacy” to take into account. That being that the driver is doing more than just driving. They’re also being a person in the role, caretaking the bus in other ways, such as providing a sense of security and orderliness. Something that automation can’t replace. That’s less of a problem for a personal vehicle, but for a shared bus it becomes more acute.
I do agree that automated busses would be a game changer here, but it’ll probably be a good ten years or more before they become even remotely viable. The fact is, there’s no reason we should wait around for that when we could be doing good stuff now. Let the clanker busses roll out gradually with time as they become more viable, but do that alongside better transit now.
As a last aside, I do like your point on using smaller shuttle style busses in lower demand areas instead of the big busses. While that obviously doesn’t remove the largest cost factor (labor), it does have other advantages, such as better gas mileage and better acceleration. Plus, personally, I find the shuttles less intimidating than a big bus. I just think it could be worth a shot. In addition, if the shuttle has less than 16 seats, the driver no longer would need a CDL, at least in theory. I’m not sure how that works for transit agencies, though.
“The problem there is that self driving street vehicles just aren’t ready for prime time yet,”
“I do agree that automated busses would be a game changer here, but it’ll probably be a good ten years or more before they become even remotely viable.”
The simple fact is that this is not true. In 2017 Drive.ai had a vehicle operating autonomously at night in the rain; by 2020, half a decade ago, a total of four companies had permits to test self driving vehicles on California roads. Waymo has been offering full commercial service almost as long, and is now operating in five different cities; Zoox started commercial service a few weeks ago; Auroa has Class 8 self driving trucks on the road, Gatik the same with medium duty vehicles like Ford Transit vans; Avrride is testing self driving vehicles in the Gangnam district of Seoul, one of the most crowded and complex urban environments on earth. That’s just a selection of the American based companies that are farthest along. Self driving vehicles are not just ready for prime time, they are performing in it, right now, the very instant that you read this.
“I know they are driving around now, but those are much smaller vehicles with a lot more training data associated with them.”
The fully autonomous taxi service provided by Wamo and the like is actually HARDER than what a transit agency would need. Busses operate on a fixed route, with known conditions (‘after two stops signs there is a stop light’ etc.) and only have to pull over at well marked and dedicated spaces. If self driving cars can provide anywhere to anywhere taxi service (and they can) they can easily deal with a fixed route and its simplified route.
“That being that the driver is doing more than just driving. They’re also being a person in the role, caretaking the bus in other ways, such as providing a sense of security and orderliness.”
I know of many transit agencies which have given their bus drivers enclosed cabs or other barriers separating them from riders because of concern over crime and bus drivers being targeted. Those drivers and not doing anything more than driving. What’s more, millions of riders everyday take rides on fully driverless metros or airport people movers with no drivers or staff on board, despite being much more crowded. Having a “doorman” on board is no a requirement for successful transit.
“While that obviously doesn’t remove the largest cost factor (labor), it does have other advantages, such as better gas mileage and better acceleration.”
Jarrett has explained before how using two different sized fleets (regular busses and shuttle busses) is not actually a cost saver. Because so much of the fixed cost is labor the small gas milage savings are overwhelmed by the cost of procuring and stocking two different spare parts streams, training two different sets of mechanics, having enough spare vehicles of both types to handle breakdowns, etc. Larger agencies obviously do run more than one type of vehicle, but for most places a single bus size is most economical. Switching to smaller busses to serve less dense areas is only viable with self-driving vehicles where removing the cost of the driver means you can run the vehicles more often and give decent frequency to areas that won’t have as many riders.
Even if self-driving vehicles have reached the point where they are technically capable of operating a bus route under most conditions, they haven’t established the ability to do so under all possible conditions (e.g. severe weather), or at lower cost than simply paying a human driver to drive a traditional bus. In addition, transit agencies, by nature, have to be very risk averse, as the consequences of a transit system having to shut down due to some technical issue are worse than with a private taxi company, as transit systems are expected to serve people who, for financial or disability-related reasons, don’t have other alternatives.
And, even if autonomous buses could lower an agency’s operating costs, and were proven safe, it’s not a given that agencies would actually do it. After all, the U.S. doesn’t really have a strong political constituency for making transit more efficient, with the left mostly interested in other progressive priorities, and the right just wanting to paint every government program (including transit) as wasteful, in order to justify cutting it completely.
So, I can actually see transit vehicles being along the last in the U.S. to automate, outside of specialized services, such as airport rental car shuttles.
The one thing I can agree is that autonomous vehicles have not been fully tested in severe weather, particularly snow on the ground. But you better believe AV companies are working on that. Two decades ago the idea of a vehicle driving itself was almost science fiction, being worked on for various DARPA challenges where often no vehicle could complete the course. Now AVs are driving all over the world everyday. The weather challenge will be solved faster than you think (and of course, in very severe weather like blizzards even human driven cars often cannot run so its not like AVs have to be perfect.
“or at lower cost than simply paying a human driver to drive”
That AVs will operate at lower cost is so obvious it should almost be a tautology. You can have the vehicle run for 16 hours a day without paying it overtime, or paying it straight time for that matter. Running the vehicle late at night does not incur a night shift premium. In absolutely every industry everywhere automation reduces cost, transit will be no difference.
“it’s not a given that agencies would actually do it.”
This is the key problem. Government unions are a major constituency in many cities, and it seems likely that too many of them will be like the UAW, which resisted assembly line robots and saw its membership shrink 80% over the past 50 years, and not like the Teamsters, which adapted from driving teams of horses to driving trucks, and has retained 75% of its peak membership. Cities that do not challenge their unions on this will wreck their viability, just as much as the cities that do not fund transit. Other cities will not resist driverless vehicles, either for their transit agencies or for a private company that figures out by putting AV technology in a van they can move 10 people for the cost of Waymo moving one. Those cities will see an explosion of transit availability and use, and people and prosperity will move there from cities still using drivers. What person wanting an urban lifestyle will choose to live in the city where they wait 15 min for a bus and pay $3 to travel 12mph, when another city offers a 3 min wait to pay $2 and travel at 25mph?
Onux, the problem with driverless buses being so “obvious” is a paradox: They are only financially justifiable in places where ridership is extraordinarily high to begin with.
Vancouver has driverless, fully automated trains running at very high frequencies. But again, Vancouver looks like Hong Kong and has both very high densities and very high labor costs.
This does free up some money to operate buses at high frequencies, with highly paid drivers. However, Vancouver’s bus drivers are supremely productive. There are routes with articulated buses at very high frequencies with sardine-can boarding conditions that should be candidates for another SkyTrain line.
To put everything in context: the reason why a driverless train can exist is because it has the ecosystem in place to reinforce its utility. It has pre-existing high ridership, the population density to support and need high frequencies, as well as the high operating costs to make driverless trains attractive on a marginal basis.
In other words, if a transit system serves a low-density area and has low passenger utilization, it does not warrant the capital-intensive investment of a driverless system, even with the promise of point-to-point service with low wait times. It’s actually much more cost effective to buy existing technology, buses that need to be driven, and staff drivers to bump up service levels to make it desirable and useful enough.
Driverless systems vs. status-quo labor intensive systems are the age-old accounting and financing question of capital vs. operating costs. With capital costs, you are front-loading years of what could be paid out in recurring operating costs for the future benefit of long-term operating cost savings. A Vancouver Skytrain trip is probably doing the work of 5-6 bus drivers, and each bus driver is already moving sometimes 100 passengers per hour.
A driverless bus is not just a computer bolted-on to a standard transit bus. It has to be engineered, trained, and require a specialized training and maintenance staff to keep running. As a capital expense, this will be necessarily more expensive than a regular transit bus.
Manufacturers set their prices with this expectation. A bus in the U.S. has a service life of 12 years. A manufacturer has to be able to build a bus with an as-good-or-better-than-a-bus-driver “brain”, but the price of this bus has to be weighed against the cost of annual compensation per bus driver. If this premium is a 12x or more over a conventional bus, the manufacturer is out of the money because it is still less expensive to pay a human to drive the bus.
The buses also have to be built at such enormous scale that the premium must come down from 12x, but it is still out of the money even at 10x, 8x, or 6x premium. The buses will still physically deteriorate from everyday use and require more maintenance as they get older. Maybe if the vehicles get down to a 5x premium (a savings of 7 times a driver’s annual compensation) will you start to see adoption. But … BUT! … the vehicles are so costly a transit agency is only ordering 20% of their usual fleet need.
“But again, Vancouver looks like Hong Kong and has both very high densities and very high labor costs.”
Ha! Made me laugh. Hong Kong has a density of 6,800 pop/sqkm, while metro Vancouver is . . . 918 pop/sqkm. The densest parts of Hong Kong like Kowloon reach 47,500+pop/sqkm, while the city of Vancouver is 5,750 pop/sqkm. .
“They are only financially justifiable in places where ridership is extraordinarily high to begin with.”
You have this completely backwards, because you are wildly overestimating the cost of the hardware for self driving vehicles. There are no official figures, but estimates suggest that the sensors for the vehicles cost around $10,000, and that total modification cost is $100,000. The biggest cost should be the four NVIDIA H100 GPUs used for processing the camera data. Those list for $25,000 each, which would be $100,000 alone, but probably are purchased cheaper in bulk. An average cost for city transit busses is $500k; at $100k of modification isn’t 5x a regular bus, but 1/5th. Even if you take the high end estimate for Waymo’s ($300k) and deduct out the cost of the vehicle (~$80k) that still leaves a modification cost of less than half a bus’ cost, not multiples of it.
If you believe that autonomous vehicles begin to make sense at 6x the cost of regular bus (base cost + 5x premium) then at 1.2x cost they are so beneficial it shouldn’t be a question.
Thus where you have it backwards. Autonomous vehicles are not best suited to high density because their capital cost is so high, they are most advantageous in lower density because their operating cost is so low. A longstanding rule of thumb is that half the cost of operating a bus is driver salary. Cut operating cost in t half and you can provide the same service in areas where you will get half the ridership, or double the service with the same ridership. At really high densities the many riders per vehicle mean that driver cost is not as big an issue; the NY Subway has a lower cost per rider than NY busses, even though subway trains have a driver and a conductor. By NYC would not lower its cost by replacing all busses with subways. The trains cost more to operate individually, only in the areas where they carry enough people would cost per rider be lower than a bus, and the capital investment for a subway is very high. With autonomous vehicles you get the lower cost per rider without the large capital cost, and without needing a large number of people per vehicle to achieve it.
Onux, I am not wildly overestimating the cost of a self-driving bus because, for starters, one such vehicle does not exist.
And it might not, considering the state of the North American transit market.
It’s a monopsony. In the U.S., the Federal Transit Administration effectively sets the standards for bus manufacturers. Transit agencies in order to receive federal funding for said buses, order these to federal specs. This effectively limits the bus market to an oligopoly.
Right now, the U.S. bus market is dominated by two manufacturers: New Flyer and Gillig. Two smaller players, Novabus and ENC/Eldorado National, have exited the market. There were two electric bus startups that went bust: Proterra and DesignLine.
The bus also must pass Altoona testing in order to qualify for federal funding.
Then you get to the bidding process. The purchasing people are fleet managers and operations managers, who are organizationally rather conservative. Needless to say, they are rather skeptical of introducing even more simple technologies than buses that don’t need drivers. There is also the matter of permanently eliminating the workforce in an industry that is thoroughly unionized. These collective bargaining agreements make severances so costly that it’s often more cost-effective to carry the operating costs and attrition the drivers out over time.
The fleet and ops managers are going to think of so many scenarios that happen in the course of the operating environment, and then examine the cost and time the driverless bus takes to solve the problem vs. how long a human driver can solve the problem?
As someone mentioned, the “doorman fallacy”. The bus driver does more than turn the steering wheel. Take the human out of the loop, and how will the driverless bus 1) board a wheelchair passenger who asks for securement (this is the ADA and technology can’t handwave away), 2) board an ambulatory-challenged passenger (a person who might not qualify under the legal standard of disability, but use a mobility aid like a walker or rollator) who might need the ramp deployed, the bus knelt, and/or the bus to minimize the gap with the curb; 3) riders who need directions; 4) handle crowd control when a bus carries too many passengers; 5) handle bad conduct by passengers; 6) reorient routes around detours, and anything else I couldn’t think of here?
We know in real life how this is solved. Heavy amounts of capital are deployed to design these scenarios to not require manual intervention, or in the case of conduct, require spending on law enforcement? For an automated system like Skytrain, the disabled accessibility is designed into the elevators, platforms and vehicles themselves.
This is a substantially higher capital cost, because a lot of it is the primitive analog issues that have persisted before the advent of digital technology. If you are talking about buses running in mixed traffic with stops not designed but picked based upon space availability, you must bake into the budget the infrastructure that will need to be installed and maintained to reduce operational friction — things like raising or lowering sidewalk heights to be flush with the bus floor, doing something about the tight corners and obstacles posed by wheel wells, off-board fare collection, robust UX for riders to not need humans to help them navigate the transit system, etc.
It’s not bus sticker price + a couple thousands of dollars in hardware.
The capital investment is based on units of driver FTE time. The evaluation unit is going to be cost per annual driver FTE, measure in years.
This is why the vehicle is going to cost, not as the cost of an add-on, but a premium based in years of bus driver compensation. Even if the premium is 2x, it means the transit agency can only afford to purchase half as many buses as it could if they just bought a bus that needs a human to drive it.
When you look at the hard infrastructure and its maintenance, that premium is going to be 6x or more. Around, at or over 12x, the driverless bus is not worth building. Remember, 12 year service life is the FTA standard for buses, with the transit agency responsible for all operations and maintenance costs.
Upon reflection, Vancouver is actually a great example of how driverless vehicles are best suited for lower density areas. As mentioned metro Vancouver is 918 pop/sqkm while city of Vancouver is 5,750. New York City is 11,300 while Manhattan is over 28,000. But Vancouver Skytrain runs service almost as frequently or even more frequently than the New York Subway!!! The service to Braid on a BRANCH LINE runs every 6 min peak, midday and evenings; the 1 and 2 lines in NY are running 5-6 min service midday and 10-12 min in the evening. The combined section of the Expo line has 2-3 min service at peak, the same as the 7 line, NYC’s most frequent peak service. The disparity grows late night, when service is least used. NYC runs a flat 20 min headway late at night, in Vancouver the airport BRANCH runs 15 min headway late at night! On the combined section of the Canada line late night headway is 8 min, while on the Expo line late night service comes every 6 min or every 4 min on Fri/Sat nights! Vancouver runs better service late at night than NYC runs during the middle of the day!!!
Vancouver schedules here: https://www.translink.ca/schedules-and-maps/skytrain
NYC schedules here: https://en.m.wikipedia.org/wiki/List_of_New_York_City_Subway_services
The conclusion is inescapable: driverless operation allows a low-mid size metro to run subway service as good or better as the second largest metro area and biggest subway system on the continent. And the lower the usage (I.e. evenings and late night) the greater the disparity in favor of driverless. Driverless operation is not a luxury of dense areas as you claim, it is of enormous benefit to less dense area.
Upon a bit of reflection, Vancouver is not an example of driverless vehicles being practical because it is high density, it is an example of how much driverless vehicles help LOW density areas. As mentioned, Metro Vancouver has a population density of 918/sqkm, and City of Vancouver is 5,750. New York City has a population density of 11,314/sqkm, while Manhattan is more than 28,000. Yet Vancouver Skytrain provides frequency as high or higher than the New York Subway!!!
At peak, the Expo line has a train every 2-3 min, as frequent as the most frequent NYC Subway service at Peak, the 7 Train. The lowest frequency *branch* lines in Vancouver run every 10 minutes and most are every 6 minutes, equal to or better more than half of NYC subway services. This discrepancy only grows at off times, when ridership is lower. Only two Skytrain branches run at more than a train every 8 minutes in the evening while almost every NY Subway service has less frequency; the main section of Skytrain has a train every 3 to 5 minutes in the evening, matched by no NY Subway service and better than most subway services at peak.
But the greatest discrepancy is late at night, when rider density is lowest. The NY Subway runs a flat 20 min frequency, while every Skytrain branch is 15 min or better, and service on combined sections runs 6 to 8 minutes equivalent to NY Subway service in the middle of the day!
To summarize, with driverless operation a mid-size metro area can offer service as good or better as the third largest metro and busiest subway system in the hemisphere, and the times when ridership is lowest is when the less dense area outperforms the most. Far from being a luxury of very dense areas autonomous operation is an enormous benefit to less dense ones.
Onux, frame-flipping is bad faith and I did not misstate my argument.
Transit agencies, pace Donald Rumsfeld, must operate with the densities their service areas have, not the densities they wish to have.
The decision of a driverless train network to be built in Vancouver started with a political choice to go driverless from the get-go. Even though the transit agency has a unionized bus driver workforce that lost out on future work, Skytrain did not result in the loss of jobs. The train did make bus drivers more productive by making routes more crowded.
Whatever the density of the Vancouver service area is, it’s generally true that ridership and density are related but the relationship is not perfect. For instance, the UBC campus on the city’s westernmost tip is a major ridership generator. However, between the campus and the center city area is a wilderness area as well as some suburban-character residential neighborhoods that are on the way. Not optimal, but this doesn’t diminish the ridership of those east-west routes terminating at the campus.
When school is in session, any of these routes can be made driverless. But in a world of budget and capital constraints, only one of those routes can and should be converted. It should be the busiest one, with the capital expense being used to justify more frequent service and/or schedule reliability approaching 100%.
Whatever Vancouver’s density is, Vancouver can expect most of its buses throughout the service day to have a full-seated load to a crush load where riders may be passed up. A 40-foot bus with sardine can conditions can hold about 80 passengers, but seat about 36.
If we are talking about most of North America, where the productivity of a route is often less than the length of a bus (most routes with 40-foot buses don’t approach to anything even close of 40 passengers per hour), the services are so underproductive that capital investment do not change the underlying operating conditions of those transit services.
If you go to the San Francisco Bay Area, BART centers around San Francisco, where Muni has very high frequencies and very high ridership. Then you have outer Alameda, Contra Costa and San Mateo counties, where services are infrequent and ridership is very low.
Which is the transit system most likely going to see success deploying driverless buses? San Francisco, because the ridership is there and confers the most benefit.
You’re not going to turn over the small fleet of, say, Wheels — which coincidentally serves Livermore, where America’s No. 2 bus manufacturer Gillig is based — because the driverless buses are so much more expensive that they aren’t abundant enough to deliver high-frequency services in a suburban area that inhibits productive transit services by their urban form and street design (very few connected grids, and bus routes cannot “be on the way” because they’ll have to deviate or dogleg to serve destinations the riders are going to (like onto the premises of a college, hospital or shopping mall that is set back far from the road).
“First, fare revenue continues to be low, because ridership has not returned to the levels it was at before Covid.”
For all of the culture war references at the end of the article the group most responsible for the collapse of transit in the US is not rural conservatives but urban liberals. It was the urban liberals during COVID that told people coming together in cities was dangerous and that they should stay home instead. Many people who used to ride transit every day discovered they did not have to ride anymore to be productive, and now they still don’t. That loss of fare revenue is why budgets are collapsing, however patchwork and inconsistent funding was before. Talking about expanding transit to match population growth (in the traditional way, see above for different path to greater service) is foolish when ridership is lower than it was six years ago. Telling a person in the suburbs who works from home that transit to downtown is necessary for their cities prosperity is not a convincing argument, it is an exercise in futility.
Blaming the other side always feels good and gets a great response from the true believers. But Jarrett needs to grapple with the reality of how transit got to the crisis it is in, and it has nothing to do with the side that was opposing stay-at-home mandates.
I think the problem can’t be explained by loss of fare revenue, alone. Not when agencies are being faced with service cuts far exceeding the percentage of operating revenue that came from fares, even before COVID.
I think the bigger reason is that the cost of providing transit service keeps getting more and more expensive each year, that it’s outpacing inflation, and that tax revenues are failing to keep up. And, COVID has made the measure worse, due to increasing opportunities for transit drivers to deliver packages instead, and also, all those home offices leading everyone to demand more space, driving up the cost of real estate, causing bus drivers to demand more money so that they can afford rent. More recently, the effect of tariffs on the bus to produce and repair buses is probably not good either.
Asdf2, transit services outpace inflation because bus driving remains one of the most union-dense occupations in the U.S. Most bus driving jobs are school or transit, and those jobs are most likely unionized. The unions set the wage for the occupation as a whole.
Generally, a collective bargaining agreement will include a cost of living increase plus a negotiated wage increase over the life of the contract. When inflation goes up, so do wages. Bus drivers stay somewhat ahead of inflation because their wages are Inflation+Salary increase. if inflation is 5% and they get a 2% increase each year, that’s a 7% increase in operating costs.
Unfortunately for transit, it requires a 7% increase in revenue to match the costs, but ridership rarely ever grows at that level to cover costs.
Note that I said revenue, and not ridership. Even adult fares don’t cover the cost of service; seniors, disabled and children get reduced fares but don’t cost the transit system any less to move around.