The Economics of Electrification

The economics of electrification is an issue which has interested me for some time. So when drwaddles contacted me with some interesting information on the issue, I jumped at the the chance to have a look at it on this blog. Unfortunately, I haven’t been able to come up with any hard and fast rule on when a line should be electrified, but hopefully the information here can offer some insights when looking at whether a given line should be electrified.


The main costs of electrification are in the construction and maintenance of infrastructure; as well as the need to purchase different rolling stock (or at least locomotives). Stringing up wires can be expensive – the recent Craigieburn electrification cost an exorbitant $115 million for 10km of works (thanks for the tip DB!). Whilst that price tag also included two new stations and re-signalling (as well as goodness knows what else – DoI often like to hide several years worth of operating costs in these numbers), we’re still talking about a fair bit of money. Even if we’re really optimistic and say that the actual electrification (consisting purely of wires, stanchions and substations) cost half that, the per kilometre cost is still $5.75 million.

Put into context, that would see the cost of electrifying 48km Geelong line as far as Marshall around the $276 million mark. Obviously, there are economies of scale for bigger projects and the Craigieburn project was outrageously expensive for what it was, but I’d already halved the per km Craigieburn figure to get $276 million for the Geelong line. To illustrate opportunity cost, it is worth noting that the original Vlocity order (for 38 2 car sets) cost $535 million.

Maintenance of the infrastructure is a crucial cost, but it is often more than offset by reduced maintenance levels required for electric trains. I’ll discuss this issue below.


The potential benefits of electrification are fourfold – they consist of better acceleration, lower running costs when a large number of services are provided, the so called ’spark effect’ and lower carbon emissions (depending on the energy source).

Improved acceleration

Electric trains generally accelerate faster than their diesel counterparts, so for lines with closely spaced stations (like a metropolitan rail system), electrification is often a must for the sake of maintaining a reasonable average speed. However, as station spacing moves further apart – as it does in the country – the benefits of faster acceleration are reduced. Consequently, I’d argue they aren’t a significant factor for a line like Geelong. I should also point out that metropolitan railways also tend to have a high level of service – for the implications of this, see below.

Lower running costs for high frequency services

Amos and Galbraith suggest that while capital costs for electric traction are higher, operating costs can be lower. This is because of lower train maintenance and fuel costs for electric traction. Indeed, Electric and diesel services have very different supply curves. Electric trains have high initial fixed costs (because of the extra infrastructure required), but a low marginal cost. Conversely, diesels have low fixed costs but high marginal costs. These supply curves are represented graphically (and somewhat badly!) below:


Basically, if fewer than Qx services are provided, diesel trains should run, but if more than Qx services are provided, electric trains should run. Calculating Qx in terms of trains per hour is something I’d love to do, but sadly I don’t have enough data to do it properly. Furthermore, the costs are going to be distorted in favour of electric traction over the longer term, as many of the capital works are a one off (well for 80 years anyway).

On top of this, I haven’t mentioned opportunity cost and the discount rate, so what you see above is a very basic approximation.

The ‘Spark Effect’

The ’spark effect’ describes the apparently oft occurring phenomenon, in which patronage increases after electrification occurs because passengers like electric trains more. I’m fairly sceptical of this, largely because electrification seldom occurs without a concurrent change in rollingstock and/or service level.

Intuitively, I’d argue that a change in the type of train and how often that train runs are more likely to change passenger behaviour more than whether diesel or electric traction is provided. So in absence of a good sample of lines in which the means of traction was the only thing changed, I’ll rule this out as a definite benefit.

Lower carbon emissions

Electric traction can be better for the environment than diesel traction, although this gets a lot murkier when the electricity is generated from dirty sources like brown coal. I’ve seen a number of studies on this issue (some more dubious than others), and they have varying values for the carbon emissions and disparate average numbers of passengers per service. The basic rule of thumb should be that – if a decent number of passengers are carried – they will both be better than cars, and electric trains should be better than diesels (although this depends on average passenger numbers and the source of power). This is a massive topic in itself, and really deserves it’s own post.

This is not to say electric is always better for the environment – there’s no real envrionmental benefit to electrifying a line like Swan Hill which only carries two trains each way per day. The billions required have a much greater impact on the environment if spent elsewhere – that’s opportunity cost in action.

Calculating a Net Present Value

In order to assess whether electrification of a given line should go ahead, the costs need to be weighed against the benefits through a formal cost-benefit analysis. On top of this, we need to account for the cost of capital and opportunity cost (or what else we could spend the money on). A broadly accepted way of doing this is by calculating the Net present Value (NPV) – the Wikipedia article explains this far better than I can, and is worth reading. Also worth reading is the article on the time value of money – don’t get too worried about it though.

Here’s what the NPV equation looks like, taken from Wikipedia:

\mbox{NPV} = C_0 + \sum_{t=1}^{N} \frac{C_t}{(1+r)^{t}}

If you find the maths intimidating, don’t worry about it – I’m generally terrible at maths so I just use excel to work everything out for me. There’s a link later in this post which allows you to plug in your own numbers and see how changes impact on the viability of a project.

The discount rate and opportunity cost

The discount rate puts money in the future into present value terms. Economists tend to believe that a dollar is worth more to you now than it will be next week – the discount rate measures how much less it will be worth next week. The higher the discount rate, the greater the value placed on the present. In practice though, the discount rate used in these calculations is the cost of capital. The discount rate is also really handy for measuring opportunity cost, so if you can earn 8% by putting that money in the bank, or another project (like using the money to buy more trains) brings in 8% – these figures can be used very effectively as the discount rate and deliver some interesting insights.

Setting the right discount rate is really important – set it too high and worthwhile projects look non-viable; set it too low and non-viable projects look worthwhile. I think a discount rate of somewhere around 7-8% is reasonable for these sort of projects. This article explains very well why setting an appropriate discount rate is necessary.

Most intriguingly, it has been suggested that in reality, politicians face very high effective discount rates – around the 20% mark. With a short electoral cycle, politicians often view the present as much more important than the future. A project with benefits in 20 years (well beyond the current political cycle) is less likely to occur than a pork-barrel which yields political benefits almost immediately. This theory is, in my view, a pretty good explanation of why politicians in Australia are so unlikely to properly support rail.

Some results

So now I’m going to set up a very basic NPV calculation in excel, looking at the potential electrification of the Geelong line. The main assumptions I’m making are below. All of these assumptions have problems, but if anything they favour electrification (with perhaps the exception of the 50 year life of the project).

– The capital works for the project would be in the order of $276 million. This is based on the very optimistic assumption of the project costing half of what Craigieburn cost per km. See part one for details.

– No new expenditure for rollingstock would be required. V/Line are going to have to buy lots more anyway and it makes little difference whether they buy more diesel trains or new electrics and cascade the Vlocitys off the Geelong line.

– Service levels would be held constant

– The life of the project would be 50 years

– The discount rate is 8%

Quantifying costs is not so hard – quantifying benefits is difficult. Now to be clear, I don’t believe that there are any economic benefits to electrifying Geelong while service levels are so low. This diagram from part one explains why. But I’m going to make up some absolute best case benefits for the sake of the exercise. Let’s pretend that that electrification will reduce net maintenance levels such that it cuts 5% out of V/Line’s $342.3 million annual expenditure. That’s $17.1 million. Furthermore, let’s pretend that the ’spark effect’ exists, and that each of the Geelong line’s 2.57 million annual passengers gets an extra $2 utility per trip. That’s $5.1 million per year.

These are the results when the numbers are put into excel (thanks to this website for showing me how to do it properly). Electrification still doesn’t stack up for Geelong even when the costs are underestimated and the benefits overestimated – the NPV is negative to the tune of $4.4 million!

So that’s that – I don’t think the numbers support Geelong electrification unless there are going to be a lot more services running. I’d highly recommend playing around with the NPV in excel by changing around the discount rate, initial investment, benefits and life of the project. Just for fun, behave like a politician and crank up the discount rate to 20%. You’ll understand why we don’t have more infrastructure with long term benefits!


10 Responses

  1. The problem with the Geelong route in my opinion is that I dont think enough services are running. 29% growth in patronage last financial year and 25% since is pretty big and stretching staff to the limits.,21985,23742646-2862,00.html

    I also think that electrification would increase the patronage because people prefer to catch electric trains.

  2. Thanks Sean. Why would people prefer electric trains?

    I stand at Pakenham, see people diving for seats on a Vline train while a Connex train sits on the platform opposite unloved. Why is that? The Vline train is diesel and the Connex train is electric.

    Re Geelong – I suspect when we see 6 car velos with middle cars used, and some improvements at Laverton that have been promised – the Geelong service will easily accommodate more people.

  3. i would say it’s more about express running.
    Like you said in your Pakenham series, on the fastest VLine ‘table it is 42 minutes into the city.
    Clearly a better service

  4. Let’s not loose sight of Phin’s diagram. If Adelaide boosts service frequencies past a certain level, boosts train lengths and if the oil price keeps rising, they may well get their money back by electrifying. I have been quite surprised, as I noted on my own blog, that Grange and Outer Harbour are getting heavy rail, not light rail, standard electrification.

    Phin, any ideas on how to cost the difference?

  5. Would the red curve for electric trains be lower or higher according to whether the wiring was 25kv AC or 1500v DC.

    Regarding RIccardo’s Pakenham station comment, people ar obviously prefering the more comfortable trains. Maybe we need an electric Velocity.

  6. Re: the Spark affect – In Melbourne at any rate, I’d say this should describe the phenomenon whereby rail planners and politicians see electrification as providing a standard metropolitan service level (eg. 20minute off-peak and operating hours).

    This probably has more to do with most electrification projects in Victoria in recent memory being extensions (hence the existing 20-minute off-peak service is extended a couple of stations).

    Things like Melton and Sunbury electrification which are often touted illustrate the some of the other benefits – Sunbury would most likely be an extensions of the Sydenham line and would improve matters, since lower capacity V/Line service (express) would likely be replaced by higher capacity all-stations service in the peaks. Most the benefit comes from getting better efficiency of the line utilisation nearer town due to all services having similar run-times and having more capacity on the services.

    Melton electrification on the other hand would mean running a weird mix of electric services from Melton and Diesel services from beyond and bumping up service levels to a standard 20-minute frequency to Melton would make a mess of the existing flat junction at Sunshine. It would also make even more interesting the DoT’s exciting plan for a new “outer circle” line – the Tarneit link – to separate V/Line services completely from metros on approach to Melbourne.

    Interesting aside – I can’t see too many impediments to operating 9-car or longer V’locity sets express into Southern Cross (with the cost of lengthening any outer stations reasonably cheap). Longer trains don’t fit at the loop stations and a lot of inner city stations would be hard/costly to lengthen. Except by electrifying a couple of platforms in the Southern Cross terminus, I can’t imagine an easier way to provide some peak metro-express capacity than running 9-car or 12-car V/Lines and picking up at 1-2 key suburban stations.

  7. I can’t see any benefits to anyone living in Sunbury or Diggers, particularly Diggers where the small local station is set to become a focal point for transients ie travellers who don’t actually live there. This will add to noise, graffiti and violence. A Vline train currently takes from 29 to 32 minutes from Diggers to Spencer St. A metro train, stopping at all stations via the loop, is likely to take an hour, in squalor and discomfort. Where is the benefit???

    • Replying to Scarlet.

      I dimly remember Geelong commuters resisting electrification proposals as the current country trains are more comfortable and run express compared to electric trains.

  8. how do we calculate Qx?what data do we need?

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: