Are passenger trains to Mildura worth it?

The T&G Mutual building in Mildura (via Mattinbgn)

A few months ago, the Parliamentary Budget Office released the results of modelling instigated by Liberal Democratic MLC Tim Quilty, who represents the Victorian upper house region of Northern Victoria. Quilty had asked the PBO to cost a proposal to run three shuttle services per week in each direction from Maryborough to Mildura, and to pay for this by scrapping the Victorian Government's subsidy for the Overland. To my mind this is completely the wrong way to go about things - running a train every second day makes it unlikely to be useful for people, and cannibalising the Overland to do it is just petty and unnecessary - but it does at least give us some very useful numbers from which to extrapolate. 

Firstly - why is running a train every second day a bad idea? 

For a train trip to be viable, it needs to be running at a suitable time for both legs of a trip - both outbound and inbound. This is because taking the train one way and driving the other usually isn't an option - if you take the train there, your car won't be available for the return trip, so it's quite all-or-nothing; if the train doesn't work for both legs, you'll drive both legs. 

If a train runs one day a week, the odds of it running on the day of your outbound trip are 1/7 = 14%, but the odds of it running for both your inbound AND outbound trips are (1/7) x (1/7) = 2%. Compound probability is a hell of a thing. 

A service's usefulness rises exponentially with every additional service

Now, intuitively you might expect a train that runs every second day to be viable for 50% of trips, but it isn't - a train that runs three days a week in each direction has only an 18% chance of running on both the days you want it to. With every extra day you run it, its usefulness grows exponentially, so even at six days a week it's only 73% - that one missing day costs you 27% of potential riders. If we want a train to Mildura to be successful in attracting passengers, it must run in both directions every day. 

So what would this cost? 

The PBO estimates that it would cost $22.2 million to upgrade Maryborough, Ouyen, Woomelang, Birchip, Donald, St Arnaud and Mildura stations, and to re-gauge one N Class locomotive and set of carriages. To run the service seven days a week instead of 3, the upfront capital costs probably wouldn't change all that much; the infrastructure works should be the same (1), the main difference is you'd need to re-gauge two trains instead of one (2). Unfortunately the PBO doesn't break down their estimate, so I don't know how much of the $22.2m is for stations and how much is for re-gauging, but it's unlikely adding an extra train would bring the total much over $25m. But I want to really err on the side of caution, making sure I overestimate costs rather than risking underestimating them, so let's be safe and round it up to $30m. 

N457 Mildura at Southern Cross Station (via Kelisi)

Reading between the lines, it seems the PBO has estimated the actual operating costs of these Mildura shuttles to be $4.6 million per year - but because the proposal is to offset this by cancelling the Overland's subsidy of $3.8 million per year, the net effect on the budget would be $0.8m/year. (It's on this dubious basis that Quilty has claimed the service could run a profit with $1m/year fare revenue). We don't want to cannibalise the Overland, so let's take that $4.6m figure and extrapolate it out; let's triple it, to account for the other four days of the week, for some weekend penalty rates, and some padding to be safe. That's $13.8 million per year. 

The PBO also calculated farebox revenue of $1m per year, which if you assume is $333k per operating day per direction would mean $2.3m revenue for a seven-day service. However, as we saw above, a seven-day service is around 5.4 times as useful as a three-day service, which suggests revenue would actually be more like $5.4m/year. But the PBO's original revenue figure seems like quite a back-of-the-envelope guesstimate anyway, so who knows. Again to err on the side of pessimism, let's ignore fare revenue altogether. 

N458 Rural City of Maryborough speeding through Sunbury (via Aulj7)

Even with my likely-overestimated costings, and even completely ignoring any farebox revenue coming in, $100 million would easily cover the costs of building this and running it for at least five years. Which in the context of the State Government's transport budget is genuinely just pocket change - in the 2021-22 budget, they allocated $100m to "commence detailed design and planning" to upgrade the Calder Freeway. Not to build any actual upgrades - just to do the planning. And of course it pales in comparison with the $16.5 billion being wasted on the North East Link. 

One minor caveat is that you'd need to run more trains between Melbourne and Maryborough to connect with this train, which obviously has its own costs and revenues. This absolutely needs to happen anyway, so it's its own kettle of fish, but suffice to say this wouldn't exactly break the bank either. 

Returning passenger trains to Mildura is definitely affordable. We don't need to beg for the scraps of a three-day service, we have a strong case for a full seven-day service. And we certainly don't need to pit regional voters against each other, threatening to take away the west's Overland in order to pay for the north's Mildura train. This is something the State government can and should fund, and we should be demanding that they do. 

1. There are several crossing loops between Maryborough and Mildura for the two passenger trains to pass each other. Adding extra crossing loops would be a very good idea, to provide as much flexibility as possible for the freight trains that would be running amongst the passenger trains, but probably isn't strictly necessary for the project to proceed, so I haven't accounted for it here. 

2. The PBO seems to assume no redundancy is required for trains being out for maintenance, presumably since the rolling stock allocated to the North East line already provide this; I'm assuming the same.