You know how sometimes you see a common phenomenon, and you think there should logically already exist word for that phenomenon, but no matter how hard you search, you can find no such word? That’s what this article is about. I want to create new terms for two existing phenomena that, in my limited understanding, are not well-explored1 in spite of how common they are. I’m going to call them the sunk cost discount and the trip cost scaling problem.

The topic I want to explore is the price of public transport for car owners.

Setting the stage

I grew up in a small village in Fryslân, the Netherlands. As is common in rural Netherlands, my family owned (and still owns) a car to get to most places. Since moving to Brussels, I have been radicalised against the automobile. In discussions with my family back home, I sometimes try to encourage them to take the bus or train. But my family reports back with a problem: public transport is expensive.

At first glance, this makes no sense. It is well-documented that cars are expensive. But when I run the numbers, to my surprise, they have the right of it: public transport is more expensive. Kind of.

Napkin mathematics

Let’s run the numbers together. Let’s say my family wants to visit me in Brussels, and they depart from Lemmer, Fryslân. They first take the bus to Heerenveen, then take the train to Amsterdam or Rotterdam, and take another train to Brussels. The bus fare, according to, is €6,70 or €6,11, depending on which bus you take. The quick-but-more-expensive bus takes 25 minutes, and the slower bus takes 40 minutes.

The train fare is more flexible, depending on how far in advance you book. According to, a short-notice fare costs €55,10, and a train ride 2 weeks from now costs €36,50. This is the slower IC train, and the total journey time from Heerenveen to Brussels is approximately 4h30. The faster (and way cooler) Thalys is more expensive, but I’ll leave it out of consideration.

All things considered, you can go from Lemmer to Brussels within 5 or 6 hours for between €42,61 and €61,80.

Now the car. According to Google Maps, it is 276 km by motorway. According to the ANWB, the current price of E5 petrol is €2,129, and my family’s car drives at 5,1 L/100 km according to the manufacturer. Parking at the Brussels P+R is free, and the MIVB-STIB trip to my house costs €2,10.

Run the numbers, and you pay €32,07. To boot, the trip takes between 3h10 and 5h40, so let’s say it takes 4 hours.

And there you have it. The car trip is both cheaper and faster. Taking the average cost and time of the public transport option, it is €20,13 cheaper and 1h30 faster.

Wait a second…

Of course, the above example isn’t fair. It doesn’t account for the cost of the car purchase, the monthly insurance, the taxes, the maintenance2, or even the externalised costs of cars (infrastructure, damage to the environment and public health, et cetera). If we account for all of the previous (minus the externalised costs), the cost of owning and operating that vehicle becomes a lot higher.

But here’s the trick: my family already owns the car, and will pay those monthly expenses regardless. Whether they use the car or not, they pay the monthly upkeep, so why not use the car?

Those with a keen mind (or an ability to read the title of this article) may notice the similarity to the sunk cost fallacy. The thought process goes that you’re already committed to the car now that you’ve spent so much money on it, so you may as well keep on using it, even though it’s not the best or cheapest choice.

And that’s all well and fine and true—it would certainly be cheaper for my family to ditch the car and use alternate means of transportation—but they’re not going to. In my family’s case, it’s because public transport is rather rubbish in the rural village where they live, and they have thus determined that the higher cost of the automobile is worth it. In another family’s case, it may well be because of the sunk cost fallacy. The real rationale doesn’t really matter.

What does matter is that—because my family owns and operates a car—they get a relative discount on individual journeys taken by car when compared to public transport. The take-away is this: when you own a car, it is cheaper to drive the car.

And that’s bad, because we want to replace as many individual car journeys with other modes of transport as possible. Every car trip replaced by a bus or bike ride is one car fewer on the road, reducing congestion and pollution.

I am christening this phenomenon the sunk cost discount. It can succinctly be described as follows: for so long as you are invested in resource X, it is cheaper to use resource X compared to resource Y, even though investing in resource Y is cheaper on the whole. Or applied to cars and public transport: for as long as you own a car, it is cheaper relative to public transport to use that car for individual journeys, even though you would save money if you got rid of the car and exclusively used public transport.

No man is an island

There is another problem, however. My family doesn’t travel alone. The cost of the car trip is the cost of the car trip, regardless of how many occupants are in the car. But the cost of public transport doesn’t work like that. Each passenger pays for their own ticket. And while there are group discounts for the Dutch railways, those discounts are only available to people who have a subscription. My familly doesn’t have a subscription, both because there’s no train station immediately close to them, and because of the aforementioned sunk cost discount.

So for my family of two, the average cost of visiting me in Brussels using public transport doubles to €104,41. If my family were bigger, the cost would be even higher. There are some ways to reduce that cost, but because it’s an international journey to Belgium, that all becomes rather more complicated.

This is a problem. And like the sunk cost discount, it’s not a problem that I often see discussed. The common argument is that public transport is cheaper, which is true, but not when you already own a car, and especially not if you travel together.

Towards that end, I need to christen an additional phenomenon. I could be a nerd and call this the inverse proportionality of cost per passenger, but maybe it’s better to call it the trip cost scaling problem. It’s not nearly as catchy as the sunk cost discount, but it’ll have to do. This phenomenon can succinctly be described as follows: for each additional passenger taking a journey together, the cost per passenger decreases when travelling by car, but stays the same when travelling by public transport. Alternatively: for each additional passenger taking a journey together, the cost of travelling by public transport increases, but the cost of travelling by car stays the same. Of course, this hits a limit as soon as the car is full, and doesn’t entirely account for group discounts in public transport, but it holds true enough.

Make public transport free, damn it

The best solution to the sunk cost discount and trip cost scaling problem is to just make public transport free already. It would improve the world by a lot, and it would make me very happy indeed. But failing that, public transport just needs to be a lot cheaper (or car journeys should be more expensive, or both). The cost of a public transport journey for, say, four people should be less than or equal to the petrol cost for operating a car across the same distance. The cost for a single traveller should be substantially lower.

We want to get people out of their cars, even while they still own cars. And although some car owners voluntarily pay the premium of public transport instead of taking their car’s sunk cost discount (because driving a car is an exercise in masochism, and riding a train is sunshine and unicorns), many people won’t leave the car at home for so long as it is cheaper for them to drive it. In turn, this means that those people mightn’t give public transport an honest try, reducing the odds that they’ll ever ditch their car.

Now go forth into the internets, and spread the word. I want Wikipedia pages for these two phenomena, because I feel a little alone in making these observations.

(Towards that end, this article is licensed CC-BY-4.0. Ignore whatever the website footer says.)

Update 1 (2023-07-30): I’ve received and welcomed a decent bit of feedback on this article. Many thanks for that. This is a tiny addendum about some oversights.

There are many, many, many reasons why people take the car over public transport. The two phenomena outlined in this article are barely a factor in some cases. I am aware of this, but wanted to hyper-focus the article on one thing only: the cost of public transport for car owners. I did this for a simple selfish reason; I want my family to visit me more often. They don’t much like long car rides, but the alternative journey by train makes no economic sense for them.

Furthermore, I’m aware of the externalised costs of taking a car, and did mention them in the article, but it bears repeating: the only reason car journeys are so cheap is because we subsidise the hell out of cars and their infrastructure and off-load all the damages incurred by them (opportunity cost, productivity loss, health, environment, climate) onto society. But everything else remaining the same, I’d still like my family to be able to visit me by train more cheaply than by car.

I’ve also learnt the word I would have needed to type into DuckDuckGo to learn about the economics involved: marginal cost. It’s jargon related to economies of scale, that when translated to transportation means loosely the following: the increase in total cost [of a mode of transportation] by increasing the [journeys taken] by one. I could rephrase the definitions in terms of marginal cost, but I’m afraid to get it wrong, so I’m not going to.

There are also a few ways around the sunk cost discount (some good, some bad).

  1. If these topics are well-explored, or if terms already exist for these phenomena, then that would be rather very embarrassing. Sorry in advance. ↩︎

  2. The wear and tear caused by the individual journey also isn’t accounted for, but I think the point stands regardless. ↩︎