5. How transport network pricing can work in Melbourne
We tested our approach to understand and illustrate the potential outcomes for Melbourne under a hypothetical transport pricing system.
- We applied the five principles of good pricing and included the conditions identified by the community in our work.
- Our work shows significant benefits from the introduction of a network-wide pricing model.
- Up to 85% of Victorians could pay less under transport network pricing with discounts including a safety net of concessions and discounts for the vulnerable and disadvantaged.
- A cordon charge would significantly reduce congestion and improve travel speeds by discouraging people from driving through inner Melbourne.
- Different prices for public transport based on mode and time of day would motivate Melburnians to change their behaviour to cheaper times or modes.
Summary of benefits of long-term reform of transport network pricing
Travelling around Melbourne would be considerably different to what we experience now and particularly what it would be like with projected population growth.
A much greater share of travel would take place on public transport. It is likely that the frequency and quality of service, as well as public transport infrastructure, will have been expanded to enable this. Driving will also be faster and times more certain.
Overall productivity will have increased by reducing the time lost in congestion and, we believe, liveability also improved. Better peak management also means some major transport infrastructure projects may have been delayed, freeing up funds for use on ther projects that make life better for Victorians.
Our modelling of the direct effects of the price changes proposed in the illustrative TNP and TNP with discounts scenarios of transport network pricing shows that:
- Putting a price on roads during peak periods leads to substantial increases in road speeds in the cordon area during the morning peak.
- Differentiating transport modes by price leads to substantial shifts to the cheapest modes.
- The largest impacts of an inner Melbourne cordon fall on those from the surrounding areas, not the outer areas.
- There is mixed evidence of large shifts in the use of public transport by time due to pricing. This may reflect large effects from the inner Melbourne cordon price.
- Adding equity measures reduces the costs on those that travel long distances but does somewhat reduce the efficiency of the system.
Modelling transport network pricing
We have applied the pricing principles and community conditions set out in Section 4 to illustrate an approach to transport network pricing for Melbourne that could reduce congestion, provide more choices and be fairer than the current system.
To analyse the effects of illustrative different types of transport pricing we have used the Melbourne Activity Based (transport) Model (MABM) – one that is used to model the impacts of non-build solutions to transport problems. Using this people-focused model, we have looked at what Victorians might pay on a typical day, who would benefit, who would be worse off and how best to incorporate fairness within the system. All the analysis that follows is based on MABM modelling conducted by KPMG.
Because we see transport network pricing as a long-term reform we have reported results run on the 2031 version of Melbourne, taking new transport projects and forecast population growth into account.
Note that the 2031 version includes as completed the following transport infrastructure projects: North-East Link; Eastern Section of the Outer Metropolitan Ring Road; West Gate Tunnel; Mordialloc Bypass; Westall Road extension; Metro Tunnel Project; Fishermans Bend Tram Link; and upgrades to the M80, Tullamarine, Calder and Monash freeways as well as various public transport and road improvements in the growth areas.
The modelling accounts for capacity constraints. Individuals faced with transport system capacity constraints adjust their mode or travel time.
As there are no `real-world’ examples of comprehensive transport network pricing, we have tested three illustrative scenarios to see how Victoria’s transport network would work in 2031, with and without transport network pricing. These three scenarios are presented in Table 4 below.
The first scenario is the current set of road charges and public transport fares, referred to as the Current System.
The second scenario is an illustrative example of transport network pricing which implements the principles stated in Table 2. This example is referred to as the Transport Network Pricing system (TNP).
The third scenario supplements the TNP example with a set of measures to meet the concerns raised by the community panel as reflected in Table 3. This illustrative example is referred to as the TNP with discounts system.
Running the MABM requires assuming each price in the scenarios. We proposed a set of prices to achieve two goals. First, they needed to be as consistent as possible with the economic principles and community conditions presented in Section 4.
Second, the total revenue raised under TNP should, as closely as possible, equal the total revenue raised when applying the current fares and charges in 2031. This means the fares and charges under TNP will return the same revenue as earned from myki, registration, TAC, car stamp duty and fuel excise. In both TNP scenarios these existing charges are not applied.
This enabled us to focus solely on a change in the pricing structure between the Current System and TNP scenarios. The TNP with discounts scenario returns less revenue than the other scenarios – a decrease of approximately $1.4 billion, meaning total annual revenue from transport reduces from $7.5 billion to approximately $6.1 billion. None of the TNP scenarios change the structure of existing tolls on Melbourne’s roads, they will continue to operate and be charged in addition to any TNP charges. These prices are illustrative rather than estimates of the actual optimal prices.
In the TNP and TNP with discounts illustrative scenarios, all road users in Melbourne are charged $0.155 per kilometre according to the distance they travel.
The inner Melbourne cordon charge would apply to all vehicles entering the cordon area during the AM and PM peaks. We have modelled a cordon charge of an additional $1.00 per kilometre travelled within the cordon. The cordon area used for the TNP and TNP with discounts scenarios is defined in Figure 2 below. It is important to note that this cordon is also illustrative. Either a narrower (as used in Terrill et al. 2019a) or a broader cordon could be applied in practice depending on implementation costs, congestion reduction benefits and impacts on surrounding areas.
The values we assumed for public transport prices are reported in Table 5. Concession holders pay 50% of these prices.
Five pricing principles
- All modes, routes and parking are priced
- All costs are priced
- Provide choices but not too complex
- Different prices for different products in different markets
Table 4: The Current System and two examples of transport network pricing
Table 5: Transport prices in the TNP and TNP with discounts scenarios (before any discounts or concessions)
These scenarios are only illustrative as they are not set to optimise transport network performance. An optimal transport network pricing system would require the body responsible for setting prices to perform detailed analysis, possibly based on trials, of the potential consequences of different prices by route and mode, e.g. to minimise rat-running and make best use of each mode of public transport and the roads.
The TNP example partially implements the first and second pricing principles in that all modes, including roads, are priced with distance, time-of-day and mode-specific prices. Congestion pricing is applied to road use and peak period charges are also applied to public transport fares. Parking associated with public transport is also priced. That said, we have not tried to set fares that will fully recover operating and capital costs or systematically account for environmental externalities.
Differences in the flagfalls across modes reflect the fourth principle. Trains are the most expensive, followed by trams and then buses. The per kilometre charge is also higher on trains compared with trams and buses. On all public transport modes and in all locations, the flagfall and per kilometre charges are higher during peak than off-peak periods. There is no free tram zone or early bird free travel.
This system provides much more choice than the Current System.
Providing concession prices for public transport reflects the fifth principle of equity. In practice, there is no reason why it could also not be applied to road use charges.
The TNP with discounts example incorporates changes that primarily respond to the community panel’s concerns about location not being a disadvantage, particularly for people with low incomes. We make three `safety net’ refinements to the TNP model: quantity discounts, a daily spending cap for concession holders and a certain number of free trips each year.
Each of these refinements, by dampening the incentives provided in TNP, risk losing the efficiency gains achieved by TNP. We have chosen to illustrate how some of the concerns might be addressed and to estimate the efficiency losses from doing so. Similar to the price system itself they are not necessarily optimal ways to achieve fair outcomes.
We are also conscious that we currently have very little information on how low income and vulnerable Victorians respond to transport price changes and so include in Section 8 suggestions on how to address this.
The first refinement re-introduces quantity discounting with a 50% lower per-kilometre distance charge after daily total spending across all modes passes a certain threshold. We have modelled a threshold of $10 for non-concession holders and $5 for concession holders. Quantity discounts can improve efficiency if prices are above marginal operating costs. If prices are below marginal costs efficiency is reduced as the discount encourages trips for which the benefits are less than the costs of providing the transport.
The other refinements introduce a small number of free trips per year for all Victorians and a cap on daily spending for concession holders, effectively making additional trips after the cap free. We have modelled 20 free travel days a year per person and a daily spending cap of $5 across all modes for concession holders. A system with free trips is less efficient because the prices for these trips are below marginal cost. In effect, these changes reduce the efficiency of the transport system.
Table 6: Summary of safety net refinement measures of TNP with discounts scenario
|Scenario name||Relevant discounts|
|TNP with discounts||
While these refinements will reduce the difference between the charges paid across different locations, whether they improve equity depends on whether distance travelled is correlated with disadvantage.
If distance is correlated with disadvantage, the TNP with discounts scenario will improve equity more than TNP. Daily caps are targeted directly at concession holders; the other measures would apply to all Victorians. As the number of free trips is limited, this is unlikely to have a substantial impact on demand.
If distance travelled is not correlated with disadvantage, as suggested by analysis undertaken by the Bureau of Infrastructure, Transport and Regional Economics (BITRE, 2016), then quantity discounts to all may worsen equity outcomes. BITRE’s examination of long commutes found that distance commuting tends to be positively correlated with income. In other words, this suggests that while there may be a concentration of disadvantaged Victorians in a particular area, those travelling long distances to and from the area may not be disadvantaged.
The equity measure of 20 free travel days a year would be particularly beneficial to regional users because their most expensive travel days are likely to be more expensive than Melburnians’.
These free travel days could be especially important in making sure regional Victorians don’t have to pay a premium to visit Victoria’s most important cultural, sporting and health assets, many of which are located within central Melbourne.
Regional Victorians would also benefit greatly from the $5 daily cap on travel spending for concession holders, particularly because the proportion of Victorians older than 60 is higher in regional and rural areas than in Melbourne (41.1% versus 30.9%). Concession holders – such as seniors accessing central Melbourne’s hospitals or students accessing Melbourne’s universities – could travel hundreds of kilometres in a day and only ever pay $5 in transport charges.
Figure 1: Regions of Melbourne
We modelled 11 regions within Greater Melbourne as illustrated in Figure 1. The regions defined in our modelling originate from boundaries as provided in the 2017-2050 metropolitan planning strategy, Plan Melbourne. We have further refined these regions to get greater resolution and clarity into how Greater Melbourne moves using the MABM.
To simplify the presentation for some analyses we group them into three: Inner, Middle and Outer, as illustrated by the different colours across the 11 regions.
The Outer area includes some of the most rapidly growing suburbs of Melbourne and is also where much future population growth will occur. It would be easier and more efficient to get transport network pricing in place before this growth happens so the population can take this into account when deciding where to live. As we will see, this would also minimise the number of people who have higher transport costs resulting from transport network pricing (or minimise the budgetary cost of addressing location-based inequities).
In the MABM, it is assumed that everyone takes the same trips under both systems. Travellers respond to the new prices by changing mode or the time when they travel.
The MABM also does not model land use changes in response to pricing changes. Hence the responses reported by the MABM can be interpreted as minimum estimates as it does not allow for changes in trips or broader economic changes from transport network pricing reform.
Figure 2: Inner Melbourne cordon as modelled
Our analysis compares transport network pricing with existing transport costs set by the state government, plus fuel excise. Under the current system, this includes fuel excise, registration, compulsory TAC charge, stamp duty and public transport fares.
Because registration, the TAC charge and stamp duty vary with the type of car, a standard type of car is assumed for all individuals simulated in the model. Under TNP, this includes the road distance charge, the cordon charge, public transport fares and parking charges at train stations and at the Doncaster Park-and-Ride.
Using a standard car means that the analysis probably overestimates the stamp duty for low income earners and underestimates it for high income earners – but stamp duty is a relatively small component of average daily transport costs.
 For more detail on the MABM and how it works, search for `MABM’ on www.infrastructurevictoria.com and refer to the KPMG-Arup Model Calibration and Validation Report
 Though it would be possible, of course, to set prices at a level that maintains the current level or even increases the revenue earned from travellers
 To implement 20 free trips a year, we assumed 6% of trips were free. This is due to the limitations of the MABM model. Note also that because the simulation is for a particular day, it is as though the travellers with the highest costs all took their free trips on the same day – so these results will probably be an upper bound of the impact
Pricing delivers highly positive outcomes for Melbourne
Our modelling shows there could be substantial benefits from introducing transport network pricing in Melbourne.
Introducing a cordon price in inner Melbourne (Figure 2) substantially reduces congestion in the inner cordon area and improves travel experiences as show in Figures 3 and 4.
Figure 3 shows the average speed in the inner cordon area under the Current System and for TNP.
Figure 4 shows the effect of the cordon price by reporting the percentage change in vehicle kilometres travelled (VKT) in the inner cordon area by time period under TNP compared with the Current System (for 2031).
As Figure 3 shows, during the AM peak, average speed increases by 36% (7kph) more than halving the difference between the AM peak and off-peak speeds. There is an about 10% reduction in time spent in peak congestion. Speeds also improve in the PM peak, although this is less striking as there is less of a gap to begin with.
Figure 4 reveals where the improvement in speed is coming from. During the peak periods there are reductions in the vehicle kilometres travelled in the cordon. Interestingly, during the inter-peak, but not the off-peak, there is also a non-trivial reduction in vehicle kilometres travelled. This could reflect the fact that other trips associated with work travel (like doing shopping during the day) are also being switched to public transport. This is less the case for trips after work.
Figure 3: Cordon average speeds TNP vs Current System
Figure 4: change within cordon of VKT* (per day)
Effects of road pricing on the use of roads and public transport
Figures 5 and 6 show the effects of transport network pricing on road and public transport use in general i.e. not just within the cordon. These results indicate that introducing road pricing across
Melbourne has a substantial effect, decreasing private vehicle trips and increasing public transport use across the network.
There is about a 1.7% reduction in the number of car trips and a 2.5% reduction in VKT (representing a reduction of over 196,000 car trips and a reduction of over 3.6 million VKT per day). This means a 7% increase in the number of public transport trips (representing over 180,000 new public transport trips). Active transport does not change to any substantial degree.
Figure 5: Private vehicle use (per day)
Figure 6: Public and active transport use
More travel choices
The results show that Melburnians could respond to transport network pricing by changing mode and time of travel if offered more choice when `shopping’ for travel within Melbourne.
While people make these choices to benefit themselves, they also benefit others by using the transport system more efficiently. The Current System, while very simple, does not give people the option to seek out cheaper fares by taking the bus or tram instead of the train, or to get to the inner city more quickly by private vehicle (or commercial and freight vehicle) by paying a premium via a cordon charge.
Effects of differentiated pricing on mode choice
Although road pricing increases travel on all modes of public transport (Figure 7), making buses cheaper leads travellers to make greater use of buses as reported in Figure 8, showing a substantial increase in bus patronage of almost 130,000 new boardings.
Figure 7: Change in person trips by mode
Figure 8: % increase in travel by public transport mode
Effects of peak pricing on time of travel
Figures 9 and 10 show the changes in boardings and vehicle kilometres travelled for the four periods of the day under the two new scenarios compared with the Current System.
For buses (noting that most bus trips in Melbourne will not cross the CBD cordon), Figure 9 shows boardings increase at a faster rate for the off-peak (day) compared with the AM and PM peaks. There is a considerably smaller increase in off-peak bus trips.
Figure 9: % change in pulic transport boardings by mode and time
Figure 10: % change in passenger kimometres by mode and time
The increase in passenger kilometres is greater than boardings. This is consistent with the increase in boardings during peak being for longer trips. It is also consistent with the peak cordon price affecting long commuter trips to and from the city.
For trains, there is a larger increase in boardings outside peak times. In contrast, passenger kilometres show a greater proportionate increase during peak times.
This is also consistent with the long trips to the CBD shifting from cars to trains, with people making more efficient use of the network. The proportional increase in kilometres travelled during the inter-peak is smaller than the proportional increase in boardings, suggesting that travellers are switching to trains for short trips. The effect is more muted for the TNP with discounts case, suggesting people are possibly shifting to cars when given a discount compared with TNP.
The direction of effects on boardings for trams are the reverse of those for trains and buses. There is a greater than proportionate increase in boardings during the peak periods than the off-peak. This is consistent with the cordon having a large effect on the demand for trams – most of which run much closer to the CBD.
This suggests that although it costs more to travel on trams during peak than off-peak, any disincentive to travel during peak times is overwhelmed by commuters switching to trams to avoid the cordon price. The greater increase in kilometres travelled during the peak period suggests it is mainly longer trips that are being switched.
The increase in tram use may also be partially due to adding parking charges to train station car parks. While reduced congestion on the roads is likely to improve tram reliability, there may be a need to expand the tram fleet.
Figure 11: Cordon price mainly affects inner Melbourne residents
The effects on buses are positive as these changes involve better use of an underutilised asset — one that is relatively cheap to expand.
Figure 11 confirms that TNP reduces the incentive to drive within inner Melbourne. Our modelling shows the cordon price and other transport price changes lead to a 40% to 50% reduction in the number of people driving within the cordon.
As the number and destination of trips does not change, this is the origin of the shift to public transport.
Before we explore this result in more detail, note that Figure 11 shows that the number of people affected is much greater within inner and middle Melbourne.
Even under the Current System, relatively small numbers of people travel to the inner city, even with all the population growth that will occur there.
Figure 12 shows this in more detail. Under the Current System nearly 18% of travellers who start their journey in inner Melbourne drive within the cordon.
In the surrounding middle regions, this falls to between 3% and 8%. Around 3% of travellers starting from the outer west, north-west and northern regions drive within cordon. In all other regions the share is between 1 to 2%.
Figure 12: Cordon driving entries by home location, Current System 2031
Figure 13 shows that under TNP, in almost all regions except for those next to the inner Melbourne region, the share of those driving to the cordon (and paying the cordon charge) falls to between 0 and 2%. The numbers involved are always below 17,000.
Figure 13: Cordon driving entries by home location, TNP 2031
Impacts of fairness can be managed
Who has to pay more after the shift to more efficient transport network pricing is a central concern when thinking about public acceptance.
Related to this is the community panel’s concern that no one should be disadvantaged based on location when transport network pricing is implemented.
In the first section we have analysed, for the TNP scenario, who will end up paying more. We then analysed the extent to which the set of discounts proposed in Table 4 deal with any problems as captured by the TNP with discounts example. We also checked to make sure that addressing inequity by location does not undo all the gains in travel speeds or change the shift to public transport.
Who bears the cost of transport network pricing and where do they live?
We have compared transport costs under the Current System and under the two examples of transport network pricing.
The specific charges included are all Victorian Government charges, Commonwealth Government fuel excise, public transport fares and parking charges (at stations and park-and-rides), along with the distance and cordon charges under the new systems. Because we work with averages, charges linked with road use make up the largest component and this largely varies with distance. Not surprisingly, transport costs increase with distance from the CBD because the further out people live the further they drive.
One of the community panel’s main concerns was that people should not be disadvantaged because of their location. We have focused on this concern and interpreted this very conservatively in that average transport costs do not rise. Community panel members indicated they were willing to pay more to travel more as long as they got more for what they paid.
Under the TNP scenario, the goal of not increasing average transport costs is not met. People living in the outer suburbs have increased transport costs.
Figure 14 shows the growth rate in average daily costs between the Current System and TNP for the three regional groups across the four income groups. All households in the model are classified into three income groups: Low, Middle and High. In addition, average costs are calculated for all concession holders. We discuss how the classification was done in an appendix.
Before presenting these results it is useful to note three things. First, because most travel is by car the changes in average daily costs are determined by changes in the cost of driving. Secondly, because fixed charges have been replaced by distance charges, charges now increase with distance travelled. Thirdly, distance travelled increases with income. So, from any neighbourhood, high income earners travel the furthest, on average, and have the highest travel costs.
These results show two broad patterns. First, people living in inner Melbourne pay less under TNP than under the Current System. This is because inner Melburnians, irrespective of income, do not drive as far as other Melburnians.
Low income earners in the middle Melbourne region also pay less under TNP. However, high and middle income earners in inner and middle Melbourne pay more and all residents in outer Melbourne pay more. This is because, on average, they travel further.
In general, under TNP, with a distance-based pricing system, transport costs increase more for higher income earners.
Figure 14: % change in average daily transport costs (TNP)
Figure 15 provides more detail on the different geographical impacts of TNP. Circles show the change in average daily travel cost for each of the 11 regions. The larger the circle, the more travellers are affected. The colour of the circle, on the spectrum from blue to red, shows how much average daily transport costs change from – 50% (blue) to + 51% (red).
On average, all regions in inner Melbourne have lower transport costs in a world with TNP without discounts. Average daily transport costs increase in the south-eastern suburbs, but by less than 10%. The greatest increases in average travel costs are in select areas of Outer Melbourne. However, the number of people affected is relatively small (95 and 140 thousand in the outer north-west and outer east respectively).
Note this is for 2031, after a decade of substantial population growth. If we carried out this change now, a much smaller number of people would be affected.
This suggests that it is important to make these changes sooner rather than later. This will allow people to consider them when they decide where they live and work. They will also be relevant to decisions in these areas about services, their frequency, and investments in transport infrastructure.
More generally, it is useful to reform ahead of demand, just as it is sometimes argued that it is cheaper and easier to build infrastructure ahead of demand. This is even more the case for pricing reforms. The consequences of getting pricing reforms wrong if demand doesn’t eventuate are likely to be less than getting infrastructure wrong. It is a lot easier to correct a price than to deal with a piece of infrastructure that is no longer needed.
The final dimension we consider for its impact on fairness is whether TNP is more unequal than the current pricing system. To analyse this, we calculated the ratio of the average daily cost of the people receiving concessions to those in the high income group for each of the 11 regions.
Under the Current System, the ratio ranges from 0.36 to 0.46; under TNP, the ratio ranges from 0.22 to 0.37. This suggests that low income earners pay much less relative to high income earners under TNP.
Figure 15: Change in average daily travel costs (TNP compared with Current System) in 2031
Supporting fairness with a spatial safety net
The TNP with discounts scenario supplements the TNP prices with measures meant to reduce the effect of distance on transport costs.
The results from modelling the TNP with discounts scenario are reviewed in this section. In Figure 16, we compare the change in average daily transport costs for income groups across the three broad regions, as we did in Figure 14.
Figure 16 shows that with TNP with discounts the average daily transport cost for low income earners is lower under transport network pricing than under the Current System. Indeed, only middle and upper income earners in the outer areas will have higher average daily transport costs. This is because some are choosing to pay more to continue to drive, but in less congested conditions. Concession holders and low income earners across Melbourne will have lower average daily transport costs.
Figure 17 further highlights that this set of discounts, in our illustrative example, largely manage any disadvantage linked with location.
While under TNP Melbourne residents outside the inner suburbs had greater average daily transport costs, under TNP with discounts, it is only the relatively small number of people affected in the outer north-east and north-west that still have higher average daily transport costs.
Note that on average, in other areas within the ‘outer’ group, like the outer south and outer west (including Cranbourne, Mornington, Werribee, Sunbury and Mernda), average daily transport costs fall.
Figure 16: % change in average daily transport cost (TNP with disounts)
Figure 17: Change in average daily travel costs (TNP with discounts compared with Current System) in 2031
Supporting fairness doesn’t eliminate the efficiency gains
In this section we analyse whether adding the discounts on distance travelled eliminates the efficiency gains found with TNP.
In particular, we focus on the effects on travel within the inner Melbourne cordon. This is because one of the objectives of TNP is to reduce congestion there.
Figure 17 shows the change in average daily travel costs (TNP with discounts compared with Current System) in 2031.
Figure 18 replicates Figure 3 while adding average speeds in the cordon by time of day if TNP with discounts is applied.
We can see that outside of the AM peak there is almost no difference between average speeds in the cordon under TNP and TNP with discounts.
Figure 18 shows that in the AM peak, the gain in speed from cordon pricing is reduced by a quarter and there is still an 8% reduction in time spent in peak congestion. Nevertheless, the bulk of the gain remains.
Figure 19 explores the differences between the TNP and TNP with discounts examples in more detail by looking at impacts by time of day on vehicle kilometres travelled.
The pattern is similar to what we saw for average speeds. There is a smaller decrease in vehicle kilometres compared to TNP.
The number of car trips still falls by around 168,000.
Under TNP with discounts there is still a more than 40% reduction in the number of people driving within the cordon – particularly from inner and middle Melbourne (Figure 11).
This could mean that there are a lot more drivers that start their journey from these regions. It is also likely that even with quantity discounts, it is still better value to use public transport for long trips from outer areas into inner Melbourne.
Figures 6 to 10 show we get similar, though a bit smaller, effects on public transport under the TNP with discounts scenario. For example, we still get about 110,000 new bus boardings.
Figure 18: Cordon average speeds – all scenarios
Figure 19: % change in VKT* within cordon