Reconsidering Induced Demand
Induced demand is no doubt an important concept in transport planning, with wide-ranging implications for the efficacy of certain investments – especially in large road projects. However, it seems that induced demand is brought up all too often as an excuse for the failure of public transport to compete against the car, fitting into a wider story of new road construction eating away at public transport use. Conversely, it is used by some as an argument to justify the reintroduction of otherwise unsustainable long haul country passenger rail services; because they believe that – quite apart from whether there is any existing transport demand – that the induced demand from any given rail project will make it successful. Whilst I don’t doubt the existence of the induced demand phenomenon, its implications for transport are limited by certain factors and there are other reasons for increased transport use, some of which are discussed below. Importantly, we need to acknowledge that people are travelling more km/year in cars than 25 years ago (as shown below – taken from the Greater Sydney Metropolitan Region Discussion Paper), but that induced demand is not the only explanation.
What is induced demand?
As is often the case, Wikipedia proves to be a fantastic resource – for a good overview of the topic I recommend reading their article on the topic. Ultimately, my understanding of induced demand is that it describes a situation where an increase in supply of a given good causes an increase in demand for that good. In a transport sense, induced demand implies that the provision of new infrastructure causes people to make trips on the new infrastructure that would not otherwise have been taken, leading to increased use of the infrastructure.
Basically, induced demand is caused by a decrease in the effective price of transport caused by the new infrastructure making travel more convenient. A big part of this is opportunity cost – if trip time is reduced by the construction of a new road, the opportunity cost of making that trip is reduced, lowering the effective price of taking the trip. This is represented graphically below:
As shown, an increase in supply leads to a lower equilibrium price on the given demand curve, and quantity consumed increases. Importantly, the movement is along the demand curve, which indicates that induced demand won’t cause big travel increases in transport markets with low demand curves (like Melbourne – Mildura for example).
This relative effective price decrease can cause substitution effects. A decrease in the cost of car travel combined with no change in the cost of public transport will change the budget constraint of the consumer, leading to not only more car use, but less public transport use as well. This is shown graphically below:
Empirically, I can understand that induced demand can and does occur. For example, a friend of mine who lives in Mitcham has had his travel time to Dandenong substantially reduced with the construction of Eastlink. As a result he’s giving consideration to buying/renting a warehouse there, something which he was unlikely to have considered before Eastlink slashed his opportunity cost.
Alternative explanations
However, there’s a plethora of alternative explanations for the increase in car use, such as longer average trips to work, consumer preferences independent of price etc., but I’m going to focus on an alternative explanation for increased car use which is easily confused with induced demand.
Increased use of transport is potentially caused by an increase in aggregate demand quite independent of infrastructure changes. When real GDP per capita grows and the population increases (basically when we have more people who are more wealthy) – as it has done in Australia and a great many western countries, aggregate demand increases, shifting the demand curve out. This is shown graphically below:
Quantity consumed is higher, as is effective price (more congestion). Empirically, we are seeing this in Melbourne now, with increased population and wealth causing substantial congestion on transport networks.
Key consequences
The key consequences of what is presented above are twofold.
Firstly, induced demand, doesn’t really imply ‘build it and they will come’ as some suggest. The movement is only along the existing demand curve – induced demand won’t automatically justify the construction of major transport projects in areas with few people.
Secondly, induced demand is easily confused with an outward shift in the demand curve stemming from factors quite independent of the opportunity cost of making a trip. Increased income (which is highly correlated with increased employment – itself a driver of transport demand) and increased population can and do increase demand for transport, and it’s important to note that income and population have increased steadily in Australia over the past half a century. We should be mindful of the effects these factors have on demand.
[...] This post is published in full at Transport Textbook. [...]
Good stuff Phin
Some of the limitations I see:
-not a shift OF the supply curve, but a shift along the demand curve as the effective price changes (you’ve covered this well). Most demand factors have a price/cost proxy and even taking the motor car as the ‘gold standard’ competition, you can derive service standard/quality/reliability and so on a percentages of that. That would mean that if my journey to work in a comfortable, aircon car along a freeway is $10 of real costs and is defined as 100% quality, then the rail alternative is some % of that. Converts it to a numeric factor which you could then model as you vary the independent variable.
-Australia has a long history of infrastructure deficit, both road and rail. A lot of what we see is not ‘induced’ but ‘released’ demand in othe words, a lot of transport that would be quite reasonable elsewhere has been suppressed here. For example, would you have a quick dash into the CBD from Camberwell for quick feed and look at the shops AFTER WORK? You might be deterred on both fronts, road and rail, bby what you see here.
I should add, talking about setting the car journey to work at 100% rail will always fall short of this in Australia…but many non-urban journeys in Europe and Japan will reach greater than 100% of the service quality of road, usually due to high speed but with other value such as onboard dining, avoidance of bad weather and so on.
I’ve also mentioned, re suppressed and released demand, consider with aviation we are now seeing shifts along the demand curve, but also new supply and supply shifts. But we are long way from the USA, where for example if I was President of the Society of Basketweavers of the USA, and had a meeting of delegates from all over the country, they would fly in at a moments notice. You would struggle to do this in Australia, even though distances are typically similar. The demand that is surpressed therefore is the ability to form national volunteer or professional bodies.
The problem with Mildura (and I’m beyond saying it in other places) was even if you didn’t rob aviation of the passengers to put in a rail service, your cordon counts (say at edge of Red Cliffs and the back road through NSW) would not show enough people actually driving from Melbourne or Ballarat to Mildura to actually fill a train.
And the proposed train offering (via Maryborough) compares poorly with the Swan Hill plus bus option in terms of time.
Most expenditure vectors (varying the expenditure upwards from a few million to $1b) that you could spend on this are unstable, as the first significant changes I would see as I turned up the money dial is a shorter transit time to Swan Hill, which would increase the gap between transit times between the Swan Hill+bus and the all train via Maryborough option.
Once I had spent the amount to the Maryborough Route to get it time competitive with the Swan Hill plus bus option, I then find that the same money spent on the Swan Hill route would have produced an even better outcome. And once I get to a certain point I expect a new railway could be built from Piangil through Robinvale to Mildura for cheaper than what an even larger upgrade to the Maryborough Route would cost.
You would then close the Maryborough line at Birchip or whatever the last silo is, reduce the line speed to sufficient for wheat, and route all freight via the new line, even the Panitya line product should it survive.
Now this is not fantasising idly, it is my best take on what would seem to be the options as you turn up the money dial in a pointless quest to give more land transport options where none have been sought.
The improvement vectors are just so vastly different that trying to save the Maryborough route, for anything other than freight, is just not worth it. The decision to serve Mildura by passenger rail is essentially a decision whether to improve the Swan Hill route, as you might if you wanted to improve Robinvale or Balranald services.
Sorry people, diverging off topic.
Assuming you upgraded Swan Hill to RFR standard (which is not conceivably very difficult, only expensive) and built a new railway, which whereis.com suggests might be 200km long by taking a more direct route to Redcliffs than the current road) and having a 160km/h line speed along it, and maybe a short 1 minute stop at Robinvale, I would guess a total time from Melbourne (assuming stops Castlemaine, Bendigo, Eaglehawk, Kerang, Swan Hill, Robinvale, Redcliffs and Mildura) of 5 hours. More than competitive with road.
If you pushed the envelope even harder, and went for a 200km/h option (no better than a typical European mainline railway) you would get it down to four hours and be competitive with air transport times from CBD-town centre.
But for what? The town has few redeeming features for tourism, and not much that Melburnians can’t see at Echuca, less than half the distance from Melbourne. As a business centre it is important, but Shepparton, which is even more important, makes do with less.
I’d imagine any re-introduced passenger service to Mildura to be a welfare run like CountryLink to Dubbo, Wagga or Tamworth.
Where does the current Swan Hill passenger service fit in to all of this? It seems to be considered by many to be much more feasible than a Mildura service would be, most likely due to Swan Hill’s closer location to Melbourne/Bendigo (despite lower pop) and lack of competing aviation.
Good call Somebody
Having the Swan Hill rail service, in place and terminating 200 km from Mildura tells you that if you were on the other route, you would have an existing service as far as Birchip, running at normal line speed and ready to go. Which means that you only have 200kms to extend and upgrade.
But we don’t, and of the 500 or so kms to Mildura only the first 100kms to Ballarat was fit for purpose until recently.
I suspect some of the gunzel-think on this subject is an ill-founded attachment to the previous service rather than a genuine wish to serve Mildura by rail the best way possible.
Please don’t take this as a tacit endorsement of the Swan Hill service, merely that if they are going to keep running it, they might as well get some extra patronage by running the Mildura buses there rather than say all the way to Ballarat or Melbourne.
I can also conceive of a separate Melbourne Maryborough service, based on the fact that over that short distance rail is probably competitive with car (if they can hold it to about 2 hours). It is probably no worse a proposition that the current Ararat service.
So many of the various lines in NW VIC seem useless in 2008 – I wonder if you were building railways to service this area now if you’d just go for one main line serving the major centres without branches servicing thriving metropolises like Sea Lake.
In terms of servicing the Mallee-Sunraysia area, Mildura and Swan Hill are the two major centres so you’d want to service both of them in one hit (for an intercity service) so the issue should really be more if the last segment to Mildura is more appropriate as a bus or train than running a separate train for Mildura people.
Swan Hill pax are more captive to V/Line than Mildura pax who also have Virgin Blue, REX and QantasLink, so I’d say the current arrangement is better, although the Swan Hill service could be improved (the track beyond Bendigo is relatively slow)
Hope Phin doesn’t mind us discussing this on his blog.
Many thanks Riccardo and Somebody.
Interesting idea to model the price/cost proxy and using the car like a gold standard. I’d really like to work out the substitution price point with respect to cars in terms of some combination of service frequency, comfort etc., but I don’t really have the data/skill to do it properly. It would be a great phd research topic though!
Also agree that Australia has suffered from a substantial transport deficit, or at the very least a chronic misallocation in favour of rural areas. All those rubbish Octupus Act lines which were never justified (and similar ones that followed in Victoria right through into the 20s) saw resources not spent on much more deserving infrastructure projects. Not only was the aggregate level not much good – but the allocation was terrible. Same applies with the city loop etc. Moreover, the subsidies to the out of pocket cost of rail deferred sensible investment which would have lowered the opportunity cost.
Re. Mildura – I can’t see how it’s going to compete with air unless it’s a high standard (200kph) line – and I don’t see how that sort of investment is justified for a town like that. But keep discussing it though – it’s as much your blog as it is mine!
Thanks Phin
We start discussing induced demand if Mildura actually GREW as a result of the 200km/h rail service, which lowered the effective travel time to the same as air (and was operated to the same frequency).
The mechanism for this growth might be as follows:
-businesses that were inclined to set up either in Mildura or Melbourne, based on their activities (eg serving the horticultural irrigation industry, vineyards and so on) now find that Mildura has that much LESS disadvantage than it used to
-a government department is able to transfer up there, and the perception of the pain of the transfer would be reduced by the rail service being there
Now the unreality of this scenario runs as follows:
-Mildura is about as far from Melbourne as Canberra is. Canberra has been through this before, with government departments moved from Melbourne. This happened over many years and I doubt too many public servants were forced to make the move.
-There is no medium to high speed train from Melbourne to Canberra despite the possible demand. The existing line from Melbourne to Wagga is close to the actual chord linking the two locations, and at 200km/h with minimal stopping the train should be able to reach Wagga in 2.5 hours. Another 1.5 hours of direct line is required from the vicinity of Wagga into Canberra. The terrain would be challenging. Notwithstanding this, it would be clearly a better proposition that Mildura.
If Wagga Wagga is on an inter-capital route then it should be on the Melbourne-Brisbane route. The Melbourne-Sydney route should (eventually) be rerouted via Holbrook and Gundagai (the approximate route of the Hume Highway) which would slash 100 km from the route. Then the line via Wagga Wagga could be used for either an Albury-Wagga Wagga-Sydney service or a Melbourne-Wagga Wagga V-Line type service and a Wagga Wagga-Sydney XPT.
Good point Tom
Right now I am looking at the topographic map of the area between Wagga and Canberra looking for the gaps. I believe the VFT investigation had a similar issue – a giant tunnel under the range near Burrunjuck though nowhere near as much engineering as the Cooma route would have required.
Going from Albury would require a direct and very long confrontation with the Great Dividing Range, and the valleys in this area all run NW-SW (as in Victoria) which is the worst possible alignment they could have for a direct rail line.
The further north you go (Tumbarumba, Tumut or Gundagai) the fewer of these ranges you have, until you hit Yass when you are basically hitting the undulations of the Cullerin Range which are far more mild, but still have a significant impact on rail operations today.
If you look at the road map below
http://maps.google.com.au/maps?f=d&saddr=-35.285985,149.133911&daddr=Wagga+Wagga+NSW&hl=en&geocode=%3BFRtE6P0dE7LICA&mra=dme&mrcr=0&mrsp=0&sz=8&sll=-35.759886,148.639526&sspn=2.264367,3.537598&ie=UTF8&t=p&z=8
And see that the road takes over 3 hours and travels 242 kms indirectly from Wagga to Canberra, yet this alignment, shortened by a tunnel south of Burrunjuck and made suitable for 200km/h running, is probably the best you can do.
It is, as you can see, no surprise the VFT people found the alignment difficult, and a lot of what was gained by running the flat inland route to Wagga from Melbourne, was lost in the next 180km as the crow flies to Canberra.
A graph from Perth and one comparing Adelaide and Perth.
Err, that second link is wrong, Adelaide vs Perth.
Hi James. I’m sure the Perth expenditure and service improvements made considerable difference, but there are several logical leaps to be made between a line being electrified and it automatically getting more passengers. Victoria’s RFR, the Hunter class cars in NSW and autonomous changes between Kiama and Nowra would demonstrate that.
Some of the claims in that first link is wrong – the Perth network was never completely “shut down in favour of articulated buses”, only the Fremantle Line was. I think Armadale and Midland have had a continuous service.
I think Phin covered “the spark effect” in his Economics of Electrification series.
Riccardo re “the Hunter class cars in NSW” – has replacing ancient 620/720s with those done anything for patronage? The services haven’t become any less infrequent.
Hi Somebody
The reason that I said autonomous changes for Kiama south would be the same for the Maitland service – the same reason all rail patronage eventually rises, the cost of living and the growth in population.
I would guess that new carriages has had an effect as well, reinvigorating the service.
As we’ve discussed on Phin’s series, electrification can make a difference if frequency is increased and if transit time is reduced.
In fact, it is a sine qua non (without which not) in otherwords, you wouldn’t want to electrify if you aren’t also going to reduce transit time and increase frequency.
But you don’t have to electrify to do these things.
A little explored scenario is what might have happened in Newcastle and Wollongong if the lines had not been electrified, but the money spent on realignments. As we know, these lines did not reduce transit time, but did increase frequency.
On opening day for both the service quality did improve by virtue of the V sets arriving….but U sets and plenty of 46 hauled car sets also arrived, which were no improvement over what had previously run eg the Budds and Tullochs, the 900 cars, the 620s, the diesel hauled loco sets. Many of these non-electric sets provided service under the wires after the wires went live. It was many years before a full airconditioned service was provided.
I don’t know if the electrification budget would have bought the massive base tunnels the routes require, but might have bought a Hawkmount straightening, or a Fassifern to Adamstown straightening. This might have chopped 15 minutes of the Newcastle transit time.
The money would definitely have bought a fleet of airconditioned diesel rail cars (say the Endeavours a few years earlier) that if run to a suitable frequency would have improved service quality.
The project was of course intended for freight, but was never a goer for this – with a change of locos at Broadmeadow and in many cases at Lithgow for the coal.
We know that Qld was able to electrify at a fraction of the kilometre price at high voltage and in NSW this opportunity was lost because of a misplaced desire to get best value from rollingstock that was already becoming obsolete. Cheaper electrification might have enabled the branches eg Kandos, the Hunter Valley lines, the Metropolitan Colliery and so on to be done, which in turn would have ensured no loco changes and cheaper operation into the future.
There is no reason why a competent scheduler couldn’t have kept the 46 fleet for the Glenlee and Lithgow general freight runs until they expired, and kept the existing Vs and Us on the mountains and Gosford runs.
There’s more discussion of those graphs on SkyscraperCity which is where I found them. I think it’s obvious that not all lines were shut down, as rail patronage doesn’t go to zero. My point in posting this wasn’t a direct comment on the article, just as something related I came across.
I’d agree with you Riccardo that the electrification projects on the South Coast in particular have (still) not done anything to fix the other issues that corridor has, and it is still a dodgy mess to this day.
The issues with the Stanwell Park viaduct and Clifton Tunnel have not still not gone away, and the long section of single track south of Unanderra hasn’t changed except for a wire above it.
I wonder how much the Kiama electrification has really that area – there’s still plenty of trips requiring a change of train (between electric sets) or with an Endeavour in the electrified section (imagine the last scenario in Victoria)
Interesting idea in relation to 25kV AC electrification. Pity that QR isn’t using their overheads to Rockhampton for much at the moment – was Emerald also wired?
Back onto the subject of Mildura – I flew on one of Virgin Blue’s new Embraer jets recently (not to Mildura unfortunately). Very nice – would rather sit in one for an hour than 8 hours in an N set at 80km/h.
If the weather is clear the view of the Mallee area from the plane would be better than what you’d see passing through Hattah on 80.
Phin,
You might want to add to your article (if you can) and offer the travel budget concept as a reason for induced demand. You might also want to mention/discuss latent or suppressed demand as a factor of induced demand.