Catbus

1) Passenger-Kilometers of the STM

The STM publishes very little information about the passenger-kilometers they achieve. Luckily we have access to the Canadian Public Transit Association (CUTA) factbook, which provides this number. According to the CUTA factbook of 2015, the STM moved passengers a total of 3.45 billion km.

Unfortunately, we are still left with the problem of splitting the kilometres between the bus and the metro. There is almost no information available about this. After much searching, the only reference I found is in a 2013 presentation given by the STM at a conference in the Czech Republic. It indicates that the metro accounts for 64% of the passenger kilometers of the system. This number is a few years old, but it still provides the best estimate for our calculations — even if the total ridership has changed, the proportions are likely similar.

2) Cost of Operations of Bus and Metro

The budget of the STM provides the the following revenue and cost summary in their 2015 budget (page 10):

On the revenue side, we see plenty of income from fares, but also a lot of subsidies.
On the expenses side, most of the money is spent operating the bus and metro, but a large chunk also goes towards paratransit. The budget also includes some investments (capital costs).

To split the costs by metro and bus, we can look further in the budget, where the STM provides a breakdown of operating costs by administrative unit (page 61):

Luckily for us, the operating costs of the metro is shown separately from bus and paratransit. However, to get the complete picture, we have to figure out how to allocate the cost not directly associated with the metro or bus system, but represents services shared between them. In the absence of specific information, the best we can do is find a reasonable split. On the one hand, the metro transports more passengers and is generally more technically complicated to operate than buses. On the other hand, the bus network has twice the numbers of employees (page 62 in 2005 budget) and costs twice as much to operate compared to the metro. Considering this, I decided that splitting the shared services 40% for the metro and 60% for the bus was reasonable.

Strangely, the costs shown in the STM chart do not add up to the budget’s total operating cost but we will ignore this for our calculations.

3) Revenues of Bus and Metro

The next step is to allocate the revenues between the metro and bus, as they are reported together in the STM budget. Again, this is not a trivial task.

To illustrate the complexity, imagine a passenger purchases a single ticket and takes a bus for 2 km, then the metro for 6km, then another bus for 2km. How should we split the fare between the bus and metro? Should we split the fare 60/40 based on the number of travelled kilometers? What if the passenger transfers from one metro line to another? Or 50/50 between bus and metro? Or 33/66 because the metro was taken once, but buses were taken twice (or, in industry terms, “by boarding”)? What if the passenger transferred from one metro line to another?

To answer this question, I tried to find out how the STM and agencies share revenues for passes that span multiple transit agencies. I found that a hybrid is used: for TRAM monthly passes covering multiple agencies, the first 23.75$ are shared in proportion to the number of trips, the remainder according to passenger-km travelled (budget 2015, page 141). I applied a similar sharing model to the calculations, and allocated 20% of the revenues by boarding and the rest by passenger-km.

While analyzing revenue-splitting, I considered factoring-in the type of tickets to get more accurate numbers. Analyzing STM OPUS card tap-in data, I found that there are more single and short term tickets used on the metro compared to the bus, which in turn sees more weekly and monthly passes. This may be relevant, since single tickets bring in more revenue per trip compared to pases. On the other hand, the Metro sees more monthly passes that are shared with other agencies (TRAM passes), which complicate revenue allocations.

Without being able to model these accurately, we’ll simply assume that the usage of different tickets across the different networks has a negligible impact on revenue.

4) capital cost

Capital costs are the big one time costs used to build lines, upgrade them or make large renovations.

It is possible to trade operating costs for capital costs — for example, a system could decide to spend a lot of money to replace an expensive-to-operate heavily used bus line with an automated light metro line, which may reduce the number of staff required and thus the operating cost – but requiring a lot of capital expenditure.

On the other hand, a system could decide to never build capital-intensive rail lines at all and transport everybody by bus all the time. Then the capital costs will be much lower (since you basically only need the buses and garages), but the operating costs will be very high.

So in order to get a complete picture of the total cost of running transit, we need to consider the capital costs.

Generally, we expect the capital costs for a metro system to be large due to the expensive infrastructure, and the capital costs for a bus system to be small. For the operating cost it’s the other way round — and depending on how many users there are running on the line, the total cost of one or the other may be cheaper per passenger.

Figuring out how to count the capital costs can be quite tricky. Some investment spendings come out of the operating budget. Capital spending may vary from year to year, depending on the construction projects. Capital costs are also often paid via debt, which then spreads the costs over many years and makes it hard to identify when the costs happen.

From an accounting perspective, there are two approaches to counting the capital costs: when the money is made available and when it is spent. Both approaches have shortcomings, as debt makes it hard to track when the money for capital projects is paid out. The STM budget also doesn’t provide a detailed view to track this information. Further, it’s impossible to split the information to bus and metro spending.

Tracking when the money is spent is much easier: each year there is a capital budget, and there is even a listing of the projects where most of the money is spent. This allows to differentiate bus and metro spending. The problem is that this pretends that the capital cost of the system is only due to projects that are being done in one year, when in reality most of the capital spending may have been done in the past, with the debt being paid off slowly.

Looking at the capital budget over multiple years, we see that the spending has been relatively constant, around 600M$ per year (although there were some spikes in recent years due to the purchase of new metro cars). Also consider that the only metro extension in nearly thirty years, the 2007 Laval extension of the Orange line, cost 745M$ in total — not much more than a single year of capital budgets now.

The capital budget can be relatively easily split into bus and metro, since most of the budget is for major projects:

It seems overall, using the year-of-expenditure view of capital cost provides a reasonable approximation of the total capital cost of the system, and allows us to break it down into bus and metro spending.

One thing I found during this work is that the story could change if the numbers were picked slightly differently, if I made different choices and assumptions. It would be possible to make the metro look unprofitable or make it look much more profitable, by changing the way costs and revenues are split between buses and metro, or if one were to sneak the cost of paratransit into the equations.

So if somebody provides numbers like this, it’s important that they publish their assumptions and calculations alongside it.

This brings us back to the graph made by CDPQInfra, the private promoters of the REM. Firstly you can see that the Montreal metro is as cheap or cheaper to operate than the fully-automated REM would be in the future (19c vs 19-25c).

But also, note that both the Metro system itself and the STM overall are cheaper to operate than whatever ‘existing networks’ CDPQInfra presents, and that the total cost including capital costs is much lower as well. They are showing numbers without providing the details of their calculations, trying to tell a story that may not withstand scrutiny.

This is a bit concerning, because their whole story of profitable transit, which is presented as having the same overall cost as the existing transit lines, is used to justify important policy decisions and spending of large amount of public money.