benefits of Sydney Ferry report to IPART - … · External benefits of Sydney Ferry ... Mike Smart is a director of Sapere Research Group in Sydney. He advises industry leaders in

External benefits of Sydney Ferry

services—FINAL report to IPART

Mike Smart 23 August 2012

FINAL report—Sydney Ferry externalities ii

About the Author

Mike Smart is a director of Sapere Research Group in Sydney. He advises industry leaders in telecommunications, rail, gas, logistics, mining, electricity and aviation. Mike has given expert evidence in the Federal Court of Australia and the Australian Competition Tribunal. He is a member of the Competition and Consumer Committee of the Law Council of Australia and the Economics Society of Australia. Mike is the author of previous reports for IPART on externalities for CityRail and Sydney Buses.

About Sapere Research Group Limited

Sapere Research Group is one of the largest expert consulting firms in Australasia and a

leader in provision of independent economic, forensic accounting and public policy

services. Sapere provides independent expert testimony, strategic advisory services,

data analytics and other advice to Australasia’s private sector corporate clients, major law

firms, government agencies, and regulatory bodies.

For information on this report please contact:

Name: Mike Smart

Telephone: +61 292340210

Mobile: +61 407246646

Email: [email protected]

FINAL report—Sydney Ferry externalities iii

Executive Summary

Like other modes of public transport in Sydney, ferries are subsidised by the Government. The rationale for this subsidy is that public transport generates external benefits that are not properly captured in the price system. Forcing public transport prices to equal costs, so that there would be no need for subsidy funding, would involve a loss of social welfare. In that event, more travellers would use private cars, contributing to road congestion, air pollution, and other disbenefits that are felt widely across the community as a result of the decisions of others.

This Final Report employs an approach that has been used in previous studies for IPART to quantify the external benefits for CityRail and Sydney Buses. The adaptation of this approach to ferries has involved some new challenges, including the need to incorporate substantial elements of tourist demand and the need to work with a less satisfactory data set on ferry costs.

The principal external effects for public transport in quantitative terms are road congestion and emissions—both of conventional air pollution with its associated local health costs, and greenhouse gases with their global impacts. Other externalities were considered, but judged to be of second‐order importance only in the present circumstances.

The oft‐mentioned icon value of Sydney ferries is derived from the existence of ferry service and various private sector alternative harbour cruising services, rather than the quantum of service offered by Sydney Ferries. For this reason, icon value was judged not to be relevant to Sydney Ferries’ pricing.

In broad terms, it was found that tourism and commuter‐based demand for ferry services are approximately equal in magnitude, although this varies by route. Tourism demand takes place principally in the inter‐peak period of the day when road congestion effects are not quantitatively significant.

Emissions per person‐kilometre travelled are nearly the same for private car and for ferry. This surprising conclusion arises because ferries are not particularly fuel‐efficient means of travelling, and because average seat utilisation on ferries (on a 24 hour basis) is quite low. Thus, there is no emission externality advantage to ferries compared to private cars. In contrast, rail and bus modes of transport are quite fuel efficient and experience higher seat utilisation factors than ferries. Therefore, the emission performance of these two public transport modes is superior to both ferries and cars.

Traffic modelling performed on behalf of IPART by the New South Wales Bureau of Transport Statistics indicates that approximately one quarter of ferry passengers would travel by car if ferry service were suddenly unavailable, while three quarters would travel by either bus or train.

FINAL report—Sydney Ferry externalities iv

Putting these findings together, the overall external benefit from ferries is small. One quarter of ferry commuters avoid travelling by car, and thus these passengers contribute to a reduction in road congestion. There is no appreciable congestion relief benefit from ferry tourists. The emission performance of ferries overall is negative, as three quarters of ferry users avoid using more emission‐efficient public transport modes. For the one quarter that avoids using cars, this decision is emission‐neutral.

Optimal ferry fares are calculated in this report. While these differ somewhat from long run marginal costs, the difference is small owing to the small magnitude of net externality benefits. Total external benefits are also calculated for Sydney Ferries. The rough magnitude is $1.9m per annum, which is orders of magnitude smaller than the current level of Government funding support for Sydney Ferries. This conclusion is robust to alternative assumptions concerning uncertain input values. These findings suggest that if current subsidy levels are indeed justified, that justification must be on some grounds other than external benefits.

The specific conclusions of this analysis are as follows:

1. Overall, tourism‐based demand for ferry services represents roughly half of all

patronage—considerably more on some routes.

2. The Short Run Marginal Cost of ferry service is lower than the fare implied by

Travel Ten ticket prices for the Manly, Mosman, Neutral Bay and Darling

Harbour services. The Long Run Marginal Cost of ferry service is higher than

the implied fare on every route.

3. For the Parramatta River and Woolwich services, Travel Ten prices are less

than half the Long Run Marginal Cost, and substantially less than even the

Short Run Marginal Cost.

4. On average across all Sydney Ferries routes, pricing falls well short of Long

Run Marginal Cost.

5. External emissions costs for ferries are approximately the same as for

automobiles on a $/person‐km basis because ferries are a relatively fuel‐

inefficient means of transport and average seat occupancy is low. This means

that no advantage to society with respect to emissions costs would be

obtained by convincing car occupants to switch to ferries.

6. While there is a significant road congestion relief benefit from ferries during

commuter hours, approximately half of all ferry demand takes place outside

these hours.

7. Non‐peak ferry usage creates only very slight road congestion relief

benefits—approximately only 10% of the benefit in $/person‐km achieved in

the peak.

FINAL report—Sydney Ferry externalities v

8. The great majority of travellers who elected to switch from Sydney Ferries

services in the event of a price increase would go to other public transport

modes, such as bus or rail, rather than car. These modes would have superior

external benefits in terms of emissions and similar congestion relief benefits

compared to ferries. As a result, it is possible if not likely that a switch away

from ferries could actually be welfare‐enhancing.

9. Optimal ferry fares would be close to Long Run Marginal Cost because the

external benefits associated with ferry use are relatively minor. This is

particularly so for tourist demand for ferry services.

10. The total external benefit from Sydney Ferries is orders of magnitude smaller

than the current level of Government financial support. This conclusion is

robust to alternative assumptions about uncertain input values concerning

the value of travel time savings and the carbon price.

These conclusions are broadly the same as those reached in the 10 January 2012 Draft Report. Some route‐specific differences and minor changes to the quantitative results between this Final Report and the Draft arise because the cost model and externality calculations have been updated to reflect the following changes:

More realistic crew costs are employed here, based on financial data provided

by Sydney Ferries;

Fuel consumption and seat utilisation data for some routes has been updated

to reflect figures published in the Sydney Ferries 2011 Annual Report;

Cockatoo Island is no longer treated as a separate route, as it is part of the

Woolwich/Balmain service;

The Commonwealth’s $23/tonne carbon price is now used for calculating the

Greenhouse Gas emissions externality.

FINAL report—Sydney Ferry externalities vi

Table of Contents

Executive Summary .......................................................................................................... iii

1 Background ............................................................................................................ 1

1.1 What are externalities? .............................................................................. 1

1.2 Previous work on public transport externalities ......................................... 2

1.3 Structure of this report ............................................................................... 2

2 Analytical approach ............................................................................................... 3

3 New challenges posed by ferries .........................................................................5

3.1 Accounting for tourism demand ................................................................. 6

3.2 Simulation modelling of the ferry mode ................................................... 12

4 Marginal costs ...................................................................................................... 13

4.1 Efficiency of marginal cost estimates ...................................................... 19

4.2 Comparison of marginal costs to ticket prices ......................................... 20

5 Marginal external costs ..................................................................................... 20

5.1 Traffic congestion externalities ................................................................ 20

5.2 Emission effect externalities .................................................................... 26

5.3 Accident externalities .............................................................................. 30

5.4 Iconic value of Sydney Ferries and other externalities ............................ 32

5.5 Summary of externalities ......................................................................... 34

6 Current extent of road pricing ........................................................................... 35

7 Optimisation of fares ......................................................................................... 39

7.1 Specification of optimisation problem ...................................................... 39

7.2 Results and analysis................................................................................ 43

8 Total external benefit ........................................................................................ 47

8.1 Base case ................................................................................................ 47

8.2 Sensitivity to assumptions ....................................................................... 50

FINAL report—Sydney Ferry externalities vii

9 Conclusions .......................................................................................................... 51

References ........................................................................................................................ 53

Appendix 1: Technical formulation of fare optimisation problem .............................. 56

Appendix 2: Input data derived from the literature .................................................... 60

9.1 Value of travel time .................................................................................. 60

9.2 Fuel consumption .................................................................................... 61

9.3 Cost of greenhouse gas emissions ......................................................... 63

Appendix 3: STM ferry modelling issues ...................................................................... 63

FINAL report—Sydney Ferry externalities 1

1 Background By way of background to the study, this section contains an explanation of the nature of externalities, a summary of previous IPART work on externalities for public transport, and an outline of the remainder of the report.

1.1 What are externalities?

In an exchange between a seller and a buyer there is often an effect on other people, which is called an externality. This effect could make the others better or worse off. The important point is that the price paid for the exchange ignores the externalities because they do not affect the seller or buyer. This phenomenon is important to economists because the net value to society of the exchange includes the externalities, yet the price system fails to deal with them adequately. Prices may reflect private but not social costs and benefits.

Presently in Sydney and other large cities around the world “second‐best” conditions apply, meaning that car users do not pay for the full external costs they impose.1 Under second‐best conditions externalities provide a rationale for public transport ticket prices that may differ significantly from the cost of providing a passenger journey. Where public transport helps to mitigate the external costs of car use, socially optimal pricing would be below cost, implying a subsidy.

It is a fact that all government‐provided public transport in Sydney is subsidised.2 In this respect, Sydney’s experience is quite typical of urban public transport worldwide. As this subsidy is large in budgetary terms, it is important to understand whether its level represents the social optimum.

The prime motivation for the present study of ferry externalities is to understand the optimal level of subsidy to Sydney Ferries. External benefits3 from Sydney Ferry services determine the optimal gap between ticket prices and journey costs. If they

1 Road user charges, including fuel levies and tolls, can be used as a means of requiring motorists to pay something toward the external costs they impose.

2 As are some privately provided commuter bus services.

3 These benefits are mainly in the form of reductions to external costs imposed by other transport modes.


can be estimated accurately, then the optimal level of subsidy to Sydney Ferries can also be determined.

Putting this in a slightly different way, public transport services that generate positive externalities or help to avoid negative externalities are paid for partly by the traveller and partly by the taxpayer. The part paid by the taxpayer represents the external benefit. It is important that the taxpayer does not pay too much or too little. If too much, then resources are wasted and taxation is higher than it needs to be. If too little, then an opportunity to reduce the social burden of less efficient forms of transport is missed.

1.2 Previous work on public transport externalities

IPART’s December 2008 Final Determination of CityRail fares and regulatory framework for 2009 – 2012 was the first such decision in Australia to take explicit account of quantitative estimates of externalities. The basic externality methodology used in that inquiry was articulated in Smart (June 2008) and Smart (November 2008).

IPART’s December 2009 Final Determination of metropolitan and outer metropolitan bus fares also took account of quantified external benefits. The externality methodology first developed for CityRail in the previous year was adapted to buses and extended in Smart (October 2009).

The principal external benefits identified in these studies were a reduction in traffic congestion and air pollution that would result from traveller decisions to use public transport instead of private motor vehicles. The approach used in this study follows the previous IPART studies. It has been refined in the light of experience, and further adapted to handle some specific characteristics of ferry service.

1.3 Structure of this report

The analytical method is set out in the next section. Further technical details are provided in the Appendix. Section 3 outlines the new challenges posed by applying the externality method to ferries. Section 4 explains how ferry marginal costs are calculated. Section 5 contains the main empirical work for this study. It explains how each of the external costs is calculated, summarises the data and tools employed, then sets out the estimates for marginal external cost for each mode. These marginal external cost figures can be used to develop estimates for total external costs and benefits for each mode. Section 6 sets out the estimation of road pricing based on current levels of fuel excise tax, tolls, and other taxes on automobile use. Section 7 applies the results of the previous sections to estimate optimal ferry fares and section 8 estimates the total external benefit, which is a key indicator of optimal levels of government subsidy to Sydney Ferry services.


Sensitivity testing is applied to these estimates to consider the importance of uncertainty in some key inputs.

2 Analytical approach This section discusses the analytical approach used in this report in procedural terms, including an explanation of data sources and tools. In Appendix 1, I set out the equations that are used, in order to put things into a more precise framework.

The main externalities considered here are reductions in traffic congestion, air pollution and motor accidents. The extent of benefit depends on the extent of modal shift between private car and public transport usage. To a lesser extent, the benefit may also depend on shifts between public transport modes.

A traveller’s choice of transport mode depends on that person’s preferences, the journey characteristics and relative price for each mode. One of the most important journey characteristics is the availability of a given mode at the traveller’s residential location. Within the timeframe of the ferry fare determination, journey characteristics are assumed to be constant for each mode.

There are two policy levers potentially available to influence the relative price of private car and public transport journeys: road user charging and ticket prices. Road user charging has been discussed at length in Australia and overseas, and there is much to recommend it on grounds of economic efficiency. Nevertheless, the political impediments to extending it appear, for now at least, to be insurmountable. Some progress has been made on parking space levies, time of day tolling for the Sydney Harbour crossings and fuel excise reform, but the introduction of more systematic road user charging does not appear likely within the next few years. Road user charges in force at present are assumed to remain constant.

The remaining policy lever for optimising transport externalities is to adjust fares. A change in fares would change modal market shares by altering the relative prices of transport modes. Changing modal shares would alter traffic congestion, fuel consumption, related emissions of greenhouse gases and conventional pollutants, the frequency and severity of traffic accidents.

Because traffic congestion is such a spatially specific phenomenon, it is necessary to simulate it using a detailed traffic model that has been tailored to the street and rail geometry of Sydney and to the origin‐destination movement patterns of Sydney residents. The Sydney Strategic Travel Model (STM) designed by the NSW Bureau of Transport Statistics is used for these traffic simulations.

The modal choice logic is incorporated in the STM, which is used to derive public transport usage by mode and the following properties of automobile traffic:


vehicle kilometres travelled;

vehicle occupancy rates;

person‐hours of travelling time;

the distribution of traffic among different velocity bands.

These usage indicators are calculated for several different ferry availability scenarios. They are capable of being broken down further by purpose of travel, time of day, and geographic regions. For each scenario, this information can be used to determine congestion costs by calculating the value to motorists of the additional travel time they are forced to endure because of congestion. It can also be used to determine fuel consumption, which correlates directly to emissions of greenhouse gases and conventional air pollutants.

Comparison of STM results for different scenarios makes it possible to calculate the marginal external benefit associated with a small increase in public transport use. The marginal external benefit is the sum of avoided automobile congestion, pollution and accident costs less the sum of congestion, pollution and accident costs associated with public transport. The marginal external benefit rate will depend, among other things, on the current level of car and public transport use. The lower the starting automobile modal share, the smaller the effect on road congestion of a given shift to public transport.

A separate calculation of the marginal cost of ferry transport is required in order to estimate optimal fares given information on marginal external benefits, and fare elasticities. In the case of the 2008 CityRail study, it was possible to perform econometric analysis on long‐term data from annual reports to estimate marginal costs. In the case of the 2009 Sydney Buses study, it was possible to do econometric analysis on monthly financial data derived from bus service contracts to estimate marginal costs. For the present study of Sydney Ferries a simple bottom‐up cost model is constructed to estimate marginal costs. Like the marginal external benefit rate, the marginal cost rate for ferries will depend to some extent on the current level of ferry patronage.


Once the ferry demand schedule has been established,4 along with ferry marginal costs and marginal external benefits as a function of ferry patronage, it is possible to perform a mathematical optimisation of the ferry fares. These optimal fares, in conjunction with fixed costs, patronage and marginal cost levels at the optimal fare, implicitly determine the optimal quantum of Government subsidy to ferry services. Recognising that these calculations are subject to uncertainty in some key input values, the optimal fares and subsidies are subjected to sensitivity testing.

3 New challenges posed by ferries Compared to rail and bus networks in Sydney, the Sydney Ferry network is very simple. There are essentially only 8 routes, two of which are point‐to‐point services (Manly and Taronga Zoo). This fact makes it feasible to analyse and optimise fares for each route separately.

Nevertheless, ferries introduce four complications which did not affect the prior IPART rail and bus externality studies to the same extent. First, a high proportion of ferry patronage is tourism‐driven. Many of the existing analytical tools are geared to commuter‐driven patronage of public transport. As a result, comparatively little is known about modal choice behaviour of tourists. This issue is considered further in section 3.1 below.

Second, there are presently two separate private fast ferry service providers on the Manly route. For various reasons it has not proven practical to model this competitive supply situation within the standard traffic modelling tools employed here. In practice, it would be expected that many of Sydney Ferries’ customers would switch to a private Manly ferry service rather than resort to car, bus or train travel.

As a result, the external benefits associated with Sydney Ferries will, if anything, be overstated by this analysis. Much of the external benefit relies on Sydney Ferries’ ability to substitute for congestion and emission‐causing automobile travel. By ignoring the existence of alternative ferry services, this analysis overstates the ability of Sydney Ferries to substitute for automobile travel. Consequently, this

4 It will probably be necessary to assume a particular functional form for this demand schedule due to the relative absence of prior econometric work on this topic. While some characteristics of ferry demand are implicit in the STM’s modal choice logic, the relationship between fares and patronage is clouded somewhat by the fact that most ferry journeys are taken on combined-mode tickets. These tickets make it nearly impossible to isolate the traveller’s payment for the ferry portion of a multi-mode trip.


analysis is conservative regarding the extent of external benefits. Some further commentary on the modelling issues for ferries is given in section 3.2 below and Appendix 3.

Third, most ferry passengers travel on multi‐mode tickets, making it difficult to isolate the effect of changes to the ferry fare when other prices are held constant. Recognising this fact, this report considers the question, ‘how much would an ideal ferry fare differ from the present actual fare if it were possible to isolate the ferry price.’

Fourth, little is known about the marginal costs of Sydney Ferry services. The short history of service contracting with the Government and the infeasibility of comparing annual report data over many decades rules out approaches that were taken for the CityRail and Sydney Buses externality studies. Instead, a simplified bottom‐up cost model for Sydney Ferries was developed as part of this work. That model development is discussed in chapter 4 below.

3.1 Accounting for tourism demand

Unlike trains and buses, which are predominantly commuter services in the peak hours, ferries are significantly used by tourists. This fact creates some new challenges, since the behaviour and price responsiveness of tourists is likely to be quite different from that of commuters. The main traffic modelling tool employed in this study, the STM, is based on data sets that refer only to Sydney residents. The preferences of tourists who are not resident in Sydney are not well documented, and are not captured by the STM.

It is possible to gain some understanding of the prevalence of tourist demand on particular ferry routes by examining the time of day patterns of ferry patronage. One would expect that on weekdays the commuter component of ferry patronage would exhibit a strong morning peak in the direction inbound to Circular Quay and a strong afternoon peak in the opposite direction. Indeed, such a pattern is observed on certain routes. Figure 3.1 shows the time of day pattern for the Mosman service. The source of this data is passenger surveys that are conducted twice yearly (May and November) by Sydney Ferries. It is publicly available on the Sydney Ferries web site.


Figure 3.1 Weekday patronage for the Mosman ferry

Similar patterns are evident for the Neutral Bay, Balmain‐Woolwich, Rose Bay (but not Watson’s Bay) and Parramatta river services other than to Parramatta itself.

An entirely different type of pattern is evident for the Taronga Zoo service, as shown in Figure 3.2.

0

50

100

150

200

250

0:00 4:48 9:36 14:24 19:12 0:00 4:48

Mosman Nov 2009

M‐F inbound

M‐F outbound


Figure 3.2 Weekday patronage for the Taronga Zoo ferry

This figure, for the colder weather period of May, exhibits some evidence of a small underlying commuter pattern (morning inbound peak and afternoon outbound peak), but the most striking feature is the large outbound peak in mid‐morning, followed by a similarly large inbound peak in mid‐afternoon. The long time interval between the outbound and inbound peaks is probably explicable by the fact that it takes several hours to see the zoo.

Another striking feature of tourist demand is that it is much stronger in the summer months than in the winter owing, most likely, to the superior passenger amenity during warm weather. In contrast, commuters must go to work in winter and summer alike, so the commuter component is not seasonally variable. It has proven possible to employ this fact to separate the tourist and commuter components of ferry demand.

A mathematical algorithm was developed to formalise the intuition about seasonal variability in tourist demand. The algorithm consists of the following steps.

0

20

40

60

80

100

120

0:00 2:24 4:48 7:12 9:36 12:00 14:24 16:48 19:12 21:36

Taronga May 2010

M‐F inbound

M‐F outbound


For each ferry service or time period in each direction November (summer)

and May (winter) weekday passenger counts are compared.

For those time periods where they are different by more than 5% of the

winter demand, the ratio of summer to winter demand is calculated.

The median ratio is calculated for those time periods. Separate medians are

calculated for inbound and outbound services.

The implied winter tourist demand is then equal to the difference between

summer and winter demand on that service divided by the quantity (median

ratio – 1). The calculated value for winter tourist demand is prevented from

being negative or greater than total winter demand.

The implied summer tourist demand is just the implied winter tourist

demand multiplied by the median ratio.

Two separate estimates of commuter demand are made: winter total

demand minus implied winter tourist demand, and summer total demand

minus implied summer tourist demand.

It is confirmed that the two estimates of commuter demand are nearly the

same.

The outcome of this algorithm for the Manly service is illustrated in the following two figures.


Figure 3.3 Implied Manly commuter demand by time of day

‐

100

200

300

400

500

600

700

5.00

5.50

6.00

6.50

7.00

7.50

8.00

8.50

9.00

9.50

10.00

10.50

11.00

11.50

12.00

12.50

13.00

13.50

14.00

14.50

15.00

15.50

16.00

16.50

17.00

17.50

18.00

18.50

19.00

19.50

20.00

20.50

21.00

21.50

22.00

22.50

23.00

23.50

24.00

Average

daily passenger count for ferry service

Manly commuter M‐F

comm1 in

comm1 out

comm2 in

comm2 out


Figure 3.4 Implied Manly tourist demand by time of day

It is noteworthy that the implied commuter demand corresponds to the recognised AM and PM peak periods, and that the implied tourist demand corresponds to the Inter Peak period, but the algorithm does not employ any fixed assumptions about journey purpose by time of day. It just happens that the highest seasonal variability corresponds to the Inter Peak period. If it did not, then this algorithm would predict significant tourist demand during peak periods.

The results of applying this algorithm to each ferry route are shown in table 3.1 below.

‐

50

100

150

200

250

300

350

400

5.00

5.50

6.00

6.50

7.00

7.50

8.00

8.50

9.00

9.50

10.00

10.50

11.00

11.50

12.00

12.50

13.00

13.50

14.00

14.50

15.00

15.50

16.00

16.50

17.00

17.50

18.00

18.50

19.00

19.50

20.00

20.50

21.00

21.50

22.00

22.50

23.00

23.50

24.00

Average

daily passenger count for ferry service

Manly tourism M‐F

win in

win out

sum in

sum out


Table 3.1 Implied split of commuter and tourist demand by route

For the Neutral Bay service it was not possible to employ the algorithm since the May passenger survey data was not available. Based on the time of day pattern of patronage on that route it was assumed that commuters dominated patronage.

3.2 Simulation modelling of the ferry mode

It was necessary for the BTS to adapt the STM to enable it to simulate the ferry‐specific scenarios contemplated in this report. A technical note prepared by the BTS contained in Attachment 3 sets out the issues that were considered and how they were dealt with.

Manly

(excl Jet

Cat)

Parramatta

R

Taronga

Zoo Mosman

Neutral

Bay

Watsons

Bay

Darling

H/Balmain

Woolwich

/Balmain Total SF

daily passenger counts

winter commuter

in 2,902 1,323 309 1,083 894 389 804 7,705

out 1,893 1,062 323 976 820 581 577 6,230

summer commuter

in 3,088 1,553 384 1,113 606 1,037 597 767 9,144

out 2,229 1,096 425 1,105 476 932 687 611 7,562

winter tourist

in 2,820 591 735 494 586 759 260 6,244

out 4,043 601 754 602 529 969 439 7,937

summer tourist

in 4,191 1,165 1,306 701 532 944 1,283 440 10,562

out 5,401 979 1,267 930 834 1,004 1,860 686 12,960

winter total 11,657 3,577 2,121 3,154 ‐ 2,829 2,698 2,080 28,116

summer total 14,908 4,793 3,382 3,849 2,448 3,917 4,427 2,504 40,228

passenger‐km/day

winter pkm/day 130,563 76,906 6,936 14,035 ‐ 34,684 11,062 15,350 289,535

summer pkm/day 166,974 103,050 11,059 17,128 6,536 48,022 18,151 18,480 389,400

commuter pkm/day

winter 53,704 51,272 2,065 9,161 ‐ 21,015 3,975 10,194 151,387

summer 59,548 56,939 2,647 9,873 2,889 24,136 5,268 10,166 171,465

tourist pkm/day

winter 76,859 25,633 4,871 4,874 ‐ 13,668 7,087 5,156 138,148

summer 107,427 46,111 8,412 7,255 3,647 23,887 12,883 8,313 217,935

Average daily M‐F

patronage


4 Marginal costs In order to optimise fares it is necessary, but not sufficient, to understand the marginal costs of ferry service. The marginal cost is the additional cost to Sydney Ferries of providing one more passenger journey. It may depend on the particular route or time of day. It will depend on the pre‐existing level of ferry patronage, as adding one passenger to a relatively empty ferry service will have virtually no impact at all on cost whereas, in the worst case, the need to cater for one more passenger when the ferry is full could require an additional ferry journey.

It is usually more practical to examine marginal costs in terms of one more ferry service per day, rather than one more passenger journey. The costs for that marginal ferry service are then averaged over the number of passengers that are likely to use that service. That approach is taken here.

It is the efficient long‐run marginal costs that are of interest for setting average fares and for determining the optimal level of government subsidy. The calculations in this chapter are directed to measuring actual marginal costs. Efficient marginal costs would be lower, but it is difficult to determine how much lower without identifying specific, feasible cost reductions.

For a transport service like ferries or buses in which the major costs are vehicle, rather than infrastructure related, long‐run marginal costs are very close to average costs of owning and operating transport vehicles. That simplification does tend to overlook long‐run marginal costs of owning and operating wharves and the fixed costs of ferry maintenance facilities. The decision to ignore wharf‐related costs appears justified on the grounds that these costs do not depend on the level of passenger demand.5

Ferry maintenance costs are somewhat more directly related to patronage. In this analysis they are included, along with fuel and on‐board labour costs.

Demand for public transport is quite dependent on the time of day. Demand is greatest during the morning and afternoon peaks, and it is highly directional (inbound CBD in the morning and outbound in the afternoon). It is this peak demand that determines the fleet requirements. Off‐peak and in the contra‐peak direction, demand is significantly less than capacity. I assume that the relationship between peak and off‐peak (including contra‐peak direction) demand is constant, so

5 Of course, if the level of demand at a given wharf were to double, it may be necessary to re-engineer the wharf (say to include two-level boarding). For demand fluctuations within quite a broad range, however, the wharf costs should be independent of patronage.


that when demand increases (or decreases) by a given percentage, the peak and off‐peak demand increases by the same percentage. This simplification makes it possible to estimate marginal costs straightforwardly.

Long‐run marginal costs per passenger journey were estimated for Sydney Ferries based on a bottom‐up cost model developed from public data, supplemented with data on capital, labour, fuel and maintenance costs provided by Sydney Ferries. Tables 4.1 and 4.2 below set out the basic input data to the model.

Table 4.1 Sydney Ferries timetable information

route

round

trips per

year

one‐

way

round‐

trip

non

express M‐F Sat

Sun+

Pub Hol

Manly (excl Jet Cat) 11.2 22.4 30 36 35 32 12,800

Parramatta R 21.5 43 78 25 21 17.5 8,420

Taronga Zoo 3.27 6.54 12 25 29 29 9,560

Mosman 4.45 8.9 20 28 18 13 8,727

Neutral Bay 5.34 12.5 25 18 13 7,980

Watsons Bay 12.26 24.52 25 28 19 19 9,157

Darling H/Balmain 8.2 25 34 37 37 12,721

Woolwich/Balmain 7.38 14.76 30 29 17 12 8,861

TOTAL inner harbour 11.48

distances

(km)

daily service

frequencies (round

trip)

timetable

1‐way

duty time

Sydney Ferries downloaded 15

Feb 2011

Source:

Google Earth


Table 4.2 Sydney Ferries fleet information

Tables 4.3 to 4.5 present basic calculations of key operating parameters based on the data from tables 4.1 and 4.2.

Source: Sydney Ferries web site

Ferry type

passenger

capacity

speed

(knots)

fuel

consu

mptio

n

(l/hr)

#

crew

# in

fleet

Freshwater 1100 18 236 6 4

Super Cat 275 24 97 3.5 4

Lady Northcott 811 12 76 4 1

Lady Herron 552 11 42 3 1

First Fleet 396 12 51 3 9

River Cat 230 22 87 3 7

Harbour Cat 150 22 45 2 2

TOTAL 28


Table 4.3 Ferry utilisation

Source: Sydney Ferries, Sapere Research Group estimates

route

2010‐11 Full

yr

patronage

person km

/ ferry km

Assumed ferry

type

pass‐

enger

capacity

annual

capacity

km (m)

avg

capacity

utilis

Manly (excl Jet Cat) 6,047,000 237.80 Freshwater 1100 313.28 22%

Parramatta R 1,610,000 59.80 River Cat 230 133.12 26%

Taronga Zoo 1,188,000 61.37 Lady Northcott 811 51.34 8%

Mosman 1,066,000 55.94 Lady Herron 552 46.81 10%

Neutral Bay 652,000 35.45 First Fleet 396 19.44 9%

Watsons Bay 1,298,000 89.45 Super Cat 275 48.92 33%

Darling H/Balmain 1,912,000 77.08 First Fleet 396 40.27 19%

Woolwich/Balmain 729,000 43.92 First Fleet 396 48.51 11%

TOTAL inner harbour 6,845,000 65.54 425.99 255.30 15%

TOTAL 14,502,000


Table 4.4 Ferry capital costs

Source: 40 7% SRG Estimates

216.16 @ 30 June 2010

Ferry type

SF

insured

RC per

ferry

($m)

linear

dep @

life

($K$1)

ROA @

WACC

($L$1)

Ann. Cap

cost ($m)

/ ferry

Ann. Cap

cost for

type ($m)

round

trips per

year

Ann Cap

cost per

RT ($)

Ann Cap

cost per

seat‐RT

($)

Freshwater 21.88 0.55 0.77 1.31 5.25 12,800 410.16 0.37

Super Cat 4.81 0.12 0.17 0.29 1.16 9,157 126.13 0.46

Lady Northcott 8.13 0.20 0.28 0.49 0.49 9,560 50.99 0.06

Lady Herron 8.13 0.20 0.28 0.49 0.49 8,727 55.86 0.10

First Fleet 5.84 0.15 0.20 0.35 3.16 29,562 106.75 0.27

River Cat 4.81 0.12 0.17 0.29 2.02 8,420 240.07 1.04

Harbour Cat 3.44 0.09 0.12 0.21 0.41

TOTAL

NB: 50%

life

expiry

assumed


Table 4.5 Summary of service and ferry inputs

With this background, the following estimates of long‐run marginal cost can be made. Table 4.6 presents these estimates and compares them to approximate fares on each route.

#days = 364 1.852

route

annual

ferry‐km in

total

non EX

duty

hrs/yr

EX duty

hrs/yr

Assumed ferry

type litre/hr

speed

(knots)

TOTAL non EX EX non EX non EX

Manly (excl Jet Cat) 284,800 13,062 1,483 12.10 24.19 Freshwater 236 18 67% 33%

Parramatta R 578,800 24,612 6,639 8.93 13.25 River Cat 87 22 41% 59%

Taronga Zoo 63,300 3,700 ‐ 8.83 Lady Northcott 76 12 74% 26%

Mosman 84,800 6,439 ‐ 7.21 Lady Herron 42 11 66% 34%

Neutral Bay 49,100 3,865 ‐ 6.92 First Fleet 51 12 58% 42%

Watsons Bay 177,900 5,178 ‐ 15.89 Super Cat 97 24 66% 34%

Darling H/Balmain 101,700 9,867 ‐ 5.31 First Fleet 51 12 44% 56%

Woolwich/Balmain 122,500 7,540 ‐ 7.97 First Fleet 51 12 66% 34%

TOTAL inner harbour 599,300 36,589 ‐ First Fleet 51 12

TOTAL all services 1,462,900 74,263 8,122

Implied avg

speed (knots)

avg

speed /

service

speed

% duty

time

spent

idle


Table 4.6 Long‐run marginal cost estimates

Given the relatively low utilisation of ferry seats on most routes these long run marginal cost estimates may be somewhat excessive, as they imply the need for increased numbers of ferries in proportion to any increases in patronage. To the extent that there is slack in the system at present, this proportionality assumption may not be warranted. In making that point, however, it should be borne in mind that whenever a transport service experiences peaky‐directional demand patterns full peak loading could be consistent with a 24‐hour average utilisation of approximately 25% (as was observed for Sydney Buses.)

4.1 Efficiency of marginal cost estimates

Noting that efficient prices should be based on efficient costs, and that the marginal cost model developed above has referred principally to Sydney Ferries’ actual costs, it is necessary to consider the efficiency of these marginal costs.

The efficiency of capital and fuel costs depends on the appropriateness of Sydney Ferries’ fleet acquisition decisions. As I am not qualified to evaluate those decisions, I make the rebuttable presumption that the current fleet approximates an efficient choice of vessels from the standpoint of capital and fuel costs. In the event that a detailed analysis of those fleet decisions arrives at a different conclusion, then the marginal cost model should be modified accordingly.

The unit maintenance cost rate per ferry revenue hour of service was derived from Sydney Ferries’ accounting records. A discrete rate was calculated for each vessel, and these vessel rates were averaged across vessels of each given class. Only the maintenance costs that were directly attributable to a specific vessel were included.

route

duty

time

(hrs)

fuel

(litre)

Rostere

d crew

cost/hr

mtce

cost/

rev hr

capa

city

%

capaci

ty

utilis

PJ/

ferry

round

trip

fuel

cost/RT

labour

cost/RT

mtce

cost/RT

fuel lab

mtce

cost/RT

fuel

lab

mtce

cost

/PJ

Avg

cap

cost

/PJ SRMC LRMC

Syd Ferry

ticket

prices/PJ

Manly (excl Jet Cat) 1.00 228.26 789.86 94.56 1100 22% 237.80 158.19 789.86 94.56 1,042.61 4.38 1.72 4.38 6.11 5.09

Parramatta R 2.60 212.76 334.21 77.54 230 26% 59.80 147.45 868.94 201.62 1,218.01 20.37 4.01 20.37 24.38 5.09

Taronga Zoo 0.40 29.60 650.59 95.62 811 8% 61.37 20.51 260.24 38.25 319.00 5.20 0.83 5.20 6.03 4.07

Mosman 0.67 27.03 272.36 20.00 552 10% 55.94 18.74 181.58 13.33 213.64 3.82 1.00 3.82 4.82 4.07

Neutral Bay 0.42 20.35 272.36 20.00 396 9% 35.45 14.10 113.48 8.33 135.92 3.83 3.01 3.83 6.84 4.07

Watsons Bay 0.83 78.10 272.36 147.51 275 33% 89.45 54.13 226.97 122.92 404.02 4.52 1.41 4.52 5.93 4.07

Darling H/Balmain 0.83 40.13 272.36 20.00 396 19% 77.08 27.81 226.97 16.66 271.45 3.52 1.38 3.52 4.91 4.07

Woolwich/Balmain 1.00 49.29 272.36 20.00 396 11% 43.92 34.16 272.36 20.00 326.52 7.43 2.43 7.43 9.87 4.07

TOTAL inner harbour

Parramatta express 1.83 153.16 334.21 77.54 230 26% 59.80 106.15 612.71 142.17 861.02 14.40 14.40 5.09

TOTAL all services

Unit prices 0.69304 TravelTen

$/litre price/10

physicals per round trip financials per round trip $/PJ


The fixed costs associated with Sydney Ferries’ workshops and maintenance facilities were excluded from this calculation. The practical effect of this approach is that the maintenance costs included in the marginal cost tend to approximate those that might arise from a very large‐scale ferry maintenance organisation in which fixed facility costs can be amortised over a large fleet. In this respect, these maintenance costs would approximate efficient costs, since they explicitly exclude the scale diseconomies that Sydney Ferries likely incurs.

Estimates of average hourly labour costs for rostered crew were provided by Sydney Ferries.

4.2 Comparison of marginal costs to ticket prices

Table 4.6 suggests that fares are above long‐run marginal costs only for the Manly, Taronga Zoo, Mosman and Darling Harbour ferry services. Cost recovery is lowest for the Parramatta River, Woolwich, Balmain and Watsons Bay services.

5 Marginal external costs The primary focus of this study is the estimation of marginal external costs associated with, or avoided by the use of Sydney Ferries’ services. This chapter presents estimates of these marginal external costs. There is one subsection devoted to each of the most commonly cited externalities of transport: congestion and pollution. Within each of these subsections, I present the method, the data, and then the results.

The third subsection argues that marginal external costs for motor accidents are too small to measure reliably. In a fourth subsection, I consider other types of externalities that may be relevant. The final subsection summarises the overall marginal external cost results.

5.1 Traffic congestion externalities

Broadly speaking, transport gives rise to two types of external costs: those experienced by other travellers, and those experienced by non‐travellers. Congestion is in the former category, while air pollution is in the latter. When one more car journey is added to an already busy stream of traffic, it slows the other cars down by a small but measurable amount. This slowing effect causes the occupants of those other cars to consume more time and petrol than they otherwise would


have to complete their journeys. The cost of this additional time and petrol is an externality.

Consider the possibility that a person who had been a passenger in a car in a busy traffic stream yesterday decides to drive her own car instead today. As a result of that decision, that person will experience a slower journey today, because there is more traffic on the road compared to yesterday. The additional travel time experienced by that person is not an externality. In deciding whether to drive her own car or to car pool, this person will consider the additional cost of petrol, tolls, and of her own travel time. The switching motorist’s travel time is just part of the “price” that is taken into account in determining her choice of transport mode.

Putting this logic in a more precise way, the sluggishness of traffic can be measured by the number of hours taken to travel one kilometre on average on a typical commuting journey. The greater the congestion, the larger this figure will be. In general, in a given traffic network, the average hours per kilometre will increase as the total number of car passenger kilometres in a given time period increases. The total travel time for all car drivers and their passengers is the average hours per kilometre multiplied by the number of passenger kilometres (counting each driver as a passenger also). The increase in total travel time with a small increase in car passenger kilometres has three components:

A. The initial passenger kilometres multiplied by the increase in hours per

kilometre;

B. The increase in passenger kilometres multiplied by the initial hours per

kilometre; and

C. The increase in passenger kilometres multiplied by the increase in hours per

kilometre.

Of these components, only A is a true external cost of car travel. B and C are experienced only by the occupants of cars newly joining the traffic, so they would be taken into account by a rational person deciding whether to travel by her own car. It is very important to recognise the distinction between these types of congestion costs. Otherwise congestion externalities will be overestimated.

Figure 5.1 illustrates these points conceptually.


Figure 5.1 Travel time burden of congestion on existing motorists

5.1.1 Method for congestion

The marginal external congestion cost for motorists has two components: excess travel time and excess petrol consumption. The value of travel time lost due to congestion is the value of a traveller’s time, expressed in dollars per hour, multiplied by the number of hours lost. To evaluate the latter for existing motorists, let Q be the total number of automobile passenger kilometres travelled during the morning

automobile passenger kilometres travelled

Area A: externaltravel time Area C

Area B

hours / km

∆ q


peak hour on a weekday. Let Y be the average number of hours required to travel one kilometre in morning peak hour conditions. Y = Y(Q) is an increasing function of Q. The total travel time is simply Q * Y.

We are interested in the increase in travel time experienced by existing motorists when the total number of automobile passenger kilometres increases by a small amount = ∆q (Area A in figure 5.1). This increase in travel time is equal to

Q * (Y(Q+∆q) – Y(Q))

The marginal external cost of congestion‐related travel time is

mec = VOT * Q * (Y(Q+∆q) – Y(Q)) / ∆q

where VOT is the average value of time to automobile occupants. If Y were a linear function of Q, say Y = A * Q + B, then the marginal external cost would be

mec = VOT * Q * A

Nearly the same formula can be used to calculate the marginal external cost of congestion‐related fuel consumption. Let V be the average number of litres of petrol required to propel an automobile occupant one kilometre in morning peak hour conditions. This would be the average litres per vehicle kilometre divided by average vehicle occupancy. V, like Y, is an increasing function of Q. Total fuel consumption is Q * V. The marginal external cost of congestion‐related fuel consumption is

mec = fuel price * Q * (V(Q+∆q) – V(Q)) / ∆q

If V were a linear function of Q, say V = a * Q + b, then the marginal external cost would be

mec = fuel price * Q * a

5.1.2 Data for congestion

The relationship between automobile passenger‐kilometres and passenger hours is subtle. It depends not only on the specific geometry of Sydney’s roads, which is substantially fixed at all times of day (apart from lane changes on the Harbour Bridge, etc.), but also on the origin‐destination patterns of journeys, which change substantially during the day.

STM runs that were previously done for IPART by the Transport Data Centre (as it then was) for the rail externality study were able to be used to gain meaningful


insight into the congestion effect in a way that distinguishes between four time periods (i.e., the AM peak, the PM peak, the Inter‐Peak period, and the Evening), four distance bands (i.e., Inner, Middle, Outer and Regional), and two directions. For each of the 32 permutations of these time/location/direction subsets of the traffic, the STM calculated the number of vehicle‐kilometres and vehicle‐hours for the Sydney transport system as it was in 2006 (“Business as usual”), and as it would be if no rail services were available. The “No Rail” scenario would involve a significant shift of non‐road users onto the road network, either as car or bus occupants. This shift would greatly exacerbate peak hour congestion. While hypothetical, this thought experiment provides quantitative insight into how congestion would worsen as traffic levels increase.

For each permutation of time, location and direction, it was possible to define three points: business as usual, no rail, and no traffic. The no traffic scenario involves zero passenger kilometres travelled with a time cost of zero. These three points define a quadratic approximation to the travel time‐travel distance relationship. The coefficient for the quadratic term is the parameter A for that permutation.

5.1.3 Results for congestion

Following the procedure set out in the previous subsection, I compiled table 5.1. For each of the 32 permutations of time period, location and direction, Q, Q*Y, A and Q*A are shown. The figure in the final column, Q*A is the one of interest for calculating the congestion impact on travel time.

Commuter ferry services involve an AM peak service in which the Inner ring is the destination, and a PM peak service in which the Inner ring is the origin. The calculations for Q*A in those two cases are highlighted.

Tourist ferry services mainly occur in the Inter Peak period (“IP”). For this time period, the value of Q*A is not substantially different to zero for the Inner ring, indicating that congestion externalities from automobile traffic at that time of day are likely to be negligible.


Table 5.1 Calculation of congestion parameters

The results of employing these values are summarised in section 5.5 below.

quadratic coeff A Q * A

Time period Ring O or D BAU NoRail1 BAU NoRail1AM Regional Dest 3,392,765 3,418,538 77,557 79,568 1.61E-08 0.0548AM Middle Dest 5,214,235 5,815,685 200,344 258,187 9.93E-09 0.0518AM Inner Dest 3,510,556 4,393,678 146,733 223,937 1.04E-08 0.0365AM Outer Dest 8,273,316 8,929,608 242,883 283,499 3.64E-09 0.0301EV Middle Dest 21,577,486 23,647,237 576,163 644,160 2.60E-10 0.0056EV Outer Dest 47,843,729 54,806,926 1,122,486 1,310,983 6.58E-11 0.0032EV Inner Dest 12,541,632 13,780,605 341,299 375,157 8.27E-12 0.0001EV Regional Dest 17,225,040 18,275,501 335,985 356,123 -1.84E-11 -0.0003IP Middle Dest 21,099,580 23,333,474 603,388 678,472 2.15E-10 0.0045IP Outer Dest 44,013,870 50,678,493 1,090,655 1,275,551 5.85E-11 0.0026IP Regional Dest 15,884,988 16,933,350 316,183 336,463 -3.31E-11 -0.0005IP Inner Dest 12,353,382 13,238,146 359,988 385,135 -5.42E-11 -0.0007PM Middle Dest 8,321,000 8,975,350 278,553 325,550 4.27E-09 0.0356PM Outer Dest 21,874,718 25,593,640 625,749 794,850 6.59E-10 0.0144PM Regional Dest 7,475,339 8,080,897 163,716 180,579 7.36E-10 0.0055PM Inner Dest 4,309,652 4,259,345 143,526 145,926 -1.90E-08 -0.0820

AM Middle Origin 3,846,089 4,047,167 146,552 177,948 2.92E-08 0.1122AM Outer Origin 10,906,356 12,665,077 354,202 471,177 2.69E-09 0.0293AM Regional Origin 3,705,593 3,998,732 85,697 96,438 3.38E-09 0.0125AM Inner Origin 1,932,832 1,846,536 70,570 72,461 -3.16E-08 -0.0612EV Middle Origin 22,348,534 24,663,530 594,901 670,320 2.42E-10 0.0054EV Outer Origin 46,377,168 52,653,443 1,082,445 1,248,605 5.95E-11 0.0028EV Inner Origin 13,423,250 15,269,115 366,430 418,616 6.38E-11 0.0009EV Regional Origin 17,038,934 17,924,179 332,576 349,736 -7.50E-12 -0.0001IP Middle Origin 20,341,879 22,353,850 585,728 656,001 2.74E-10 0.0056IP Outer Origin 45,596,615 52,758,665 1,132,667 1,332,112 5.70E-11 0.0026IP Regional Origin 15,996,907 17,136,775 317,891 339,785 -3.88E-11 -0.0006IP Inner Origin 11,416,413 11,934,171 335,043 349,361 -1.42E-10 -0.0016PM Middle Origin 10,286,810 11,572,642 343,679 420,989 2.31E-09 0.0237PM Inner Origin 6,665,447 7,996,936 237,479 313,405 2.68E-09 0.0178PM Outer Origin 17,965,324 20,031,470 484,404 568,668 6.90E-10 0.0124PM Regional Origin 7,063,128 7,308,183 154,586 161,994 1.14E-09 0.0081

VKT (i.e., Q)Implied vehicle hrs

(i.e., Q * Y)


5.2 Emission effect externalities

External costs, such as pollution, that are experienced primarily by non‐travellers can be estimated straightforwardly. While the car driver and her passengers may experience some air pollution if they drive with the windows down, the greater effect is felt by people living near a busy road. In the case of greenhouse gas emissions, the effects are felt by everyone on the planet. The dispersion of air pollution is so great that most of the polluting effects created by a given car are felt by people other than that car’s occupants.

A car’s emissions are directly proportional to the amount of fuel consumed, since pollution (both conventional and greenhouse) is the result of a chemical reaction. 6 Therefore, total air pollution depends on car passenger kilometres in a direct way. More passenger kilometres mean more litres of fuel consumed, which imply more pollution. The incremental impact on pollution of an additional car passenger kilometre is proportional to the additional fuel that must be consumed to produce that passenger kilometre. There are some subtleties arising from the influence of congestion on the relationship between car passenger kilometres and fuel consumption, and from the fact that health impacts of air pollution may not be linearly related to the concentration of pollutants in the air.

The implications of these facts are illustrated in figure 5.2.

6 This fact arises from the chemical equations for fuel combustion. The proportionality between quantity of pollution and litres of fuel consumed, while strong, is not quite exact. It depends also on the thoroughness of combustion of the fuel. In turn, this depends to some extent on the condition of each vehicle, how fast it is travelling, and whether the engine is warmed up. I ignore these second-order complications.


Figure 5.2 Estimation of emission externalities

Figure 5.2 is similar to figure 5.1, apart from the fact that the units on the vertical axis are litres of fuel consumed per automobile passenger kilometre. In figure 5.2, areas represent litres of fuel consumed. As air pollution is proportional to the fuel consumed, these areas are proxies for air pollution. The external cost of an increase of ∆q in total automobile passenger kilometres travelled is the sum of areas A, B and C because the population at large is affected adversely not only by the emissions from existing car users, but also by emissions from new car users. All fuel consumption causes external costs.

automobile passenger kilometres travelled

Area A: litres of fuel consumed Area C

Area B

litres fuel / km

∆ q


5.2.1 Method for quantifying emission effects

Cars and buses emit conventional air pollutants (including fine particulate matter, volatile organic compounds, and nitrous oxides) that have adverse health consequences for the urban population generally. They also emit greenhouse gases that make some contribution to global climate change. The amount of air pollution and greenhouse gas emission from a given type of vehicle burning a given type of fuel is proportional to the number of litres of fuel burnt, for the simple reason that all of these pollutants are the by‐products of the chemical reactions that accompany internal combustion. Quantifying the emission externalities is a simple matter of calculating the following ratios, and combining them:

Health cost per kg of each conventional pollutant;

Kg of each conventional pollutant per litre of fuel consumed;

Carbon price ($/kg);

Kg of carbon emitted per litre of fuel consumed;

Litres of fuel consumed per car‐km or bus‐km;

Average vehicle occupancy for cars and buses.

The marginal external emission cost rates are:

mecferry emissions = (vkt/pax‐km) (ldiesel/vkt) { (CostGHG/kgGHG) (kgGHG / ldiesel)

+ ∑pollutant (Costpollutant/kgpollutant) (kgpollutant / ldiesel) }

meccar emissions = (vkt/pax‐km) (lulp/vkt) { (CostGHG/kgGHG) (kgGHG / lulp)

+ ∑pollutant (Costpollutant/kgpollutant) (kgpollutant / lulp) }

A similar formula applies to bus emissions.


The empirical determination I wish to make is whether ferry usage reduces the costs of emissions and by how much. I am not attempting to endogenize this calculation.7 The emissions externality calculation will be performed once the change in road vehicle‐kilometres is determined by the STM runs. The core steps in the analytical approach are:

1. Estimate the fuel savings per passenger‐kilometre associated with a mode shift from automobile to ferry;

2. Quantify the associated reduction in emissions of carbon dioxide and conventional pollutants such small particulate matter, sulphur dioxide, nitrogen oxides, carbon monoxide, benzene, and lead;

3. Cost the avoided externality on the basis of an assumed carbon price and published values of the marginal external health costs per litre of fuel consumed.

Regarding greenhouse gas emissions, I assumed that the pre‐2010 cost sharing arrangements apply, there are no ETS in place and therefore I simply value the emissions externality avoided.

However, if I were taking a longer term perspective, then I would need to consider the feedback effects from a carbon price into fuel costs (relative rail and road fuel costs) and ferry fares. Given that increased fuel prices infer some degree of internalisation of the externality associated with carbon emissions, I would probably need to reconsider the question about whether any of that additional cost should be borne by government with respect to ferry fares.

5.2.2 Data for emission effects

Smart (2009) employed health cost estimates derived from Beer (2002) and fuel sales estimates from ABS Survey of Motor Vehicle Use (2003) to derive conventional pollutant health costs of $1.24/lulp and $1.36/ldiesel.

Watkiss (2002) calculated air pollution costs for Australian Vehicles, employing a methodology that was similar to Beer (2002). Watkiss distinguished between emission rates from vehicles manufactured between 1980‐89, 1990‐95 and 1996‐99. The figures cited below are taken from Watkiss’ Table 25, p. 39. Air pollution costs in cents per litre were given for four geographical bands, of which Band 1 is of interest. Band 1 is the inner areas of larger capital cities, including Melbourne, Sydney,

7 In other words, the impact of carbon pricing on fuel prices is not taken into account in this analysis.


Brisbane, Adelaide and Perth. For diesel passenger cars manufactured between 1980‐89, the air pollution costs were $1.38/litre. For heavy buses manufactured in the same years, the air pollution costs were $0.97/litre. These figures are relatively close to the figures derived from Beer (2002).

Notably, however, for vehicles manufactured between 1996‐99, the air pollution costs were significantly lower: diesel passenger cars = $0.46/litre and heavy buses = $0.58/litre. As Watkiss is the authority cited by the Australian Transport Council’s 2006 National Guidelines for Transport System Management in Australia (Appendix C: default externality values) I adopt these Watkiss air pollution cost estimates in this report.

Smart (2009) employed CO2 emission rates of 2.34 kg CO2/lulp and 2.68 kg CO2/ldiesel. Since the Draft Report, the Commonwealth Government has introduced a legislated carbon pricing scheme setting the level at $23/tonne CO2.

Average fuel consumption rates per automobile kilometre and ferry kilometre are used. An increasing proportion of Sydney buses uses compressed natural gas instead of diesel, with some reduction in emissions.

An overall estimate of external costs for train services was obtained from figures contained in Karpouzis (2007).

5.2.3 Results for emission effects

The results of the emission externality calculation are summarised in section 5.5 below.

5.3 Accident externalities

By reducing automobile usage, ferries reduce the likelihood of traffic accidents. Published figures are readily available on the rate of accidents per vehicle kilometre, and the total costs imposed by these accidents. However, it is important to distinguish between internalised accident costs and external costs. The accident externality phenomenon involves two complications that must be considered.

First, some of the costs of accidents are borne by the accident victims. If the accident victim is a marginal motorist (i.e., one who decides to switch from bus to car commuting or vice versa) then the probability‐weighted cost to that victim of the accident is an internal cost, not an externality. This logic applies whether the accident cost is a cash cost (vehicle repairs or property damage), or the loss of quality of life associated with permanent incapacitation or death. The latter may be difficult to quantify, but it is a cost to the marginal motorist associated with the decision to drive—not an externality.


The fact of automobile accident insurance tends, if anything, to internalise more of the accident‐related costs.8 For example, third party injury and property damage insurance brings the costs borne by non‐motorists who are injured or lose property in a car accident into the motorist’s modal choice calculation.

Nevertheless, there remain some types of accident‐related costs that are borne by the community at large, rather than the marginal motorists, even when insurance premiums are taken into account. The standby capacity at public hospitals for accident victims, police and emergency services, traffic congestion caused by accidents, and the uninsured detriment to the quality of life of third parties are examples of these external costs of traffic accidents.

The second complication is that one must establish a quantitative relationship between the incidence of traffic accidents and the number of automobile passenger kilometres travelled. This link is difficult to establish empirically, beyond making the intuitively obvious observations that the likelihood of accidents should generally increase with automobile passenger kilometres travelled, and that higher average speeds should lead to more frequent and more severe accidents. In the absence of detailed information on this relationship, the most plausible simplifying assumption is that the incidence of accidents is proportional to automobile passenger kilometres or bus passenger kilometres travelled.

If one assumed that the accident rate per automobile passenger kilometre is constant, then the decision of the marginal motorist to join the traffic would have no effect on the likelihood that an existing motorist would have an accident. In other words, because of this assumption, all of the increased accident risk caused by the marginal motorist is internal to the marginal motorist’s modal choice decision. There is no external accident cost.

Note that this counterintuitive conclusion is dependent on the assumption that the accident rate per passenger kilometre is constant. There may be grounds to believe that the accident rate falls as passenger kilometres increase. Congestion slows the traffic, making it easier to avoid accidents and lessening the severity of those accidents that do occur. It is not clear from the available material that the traffic accident externality is necessarily a point in favour of increasing public transport patronage.

It is recognised that this finding runs counter to the conventional wisdom on accident externalities. There is no denying that increasing usage of automobiles

8 This statement assumes, of course, that the insurance industry is workably competitive so that insurance premiums change in response to changes in accident costs.


increases the total cost of accidents, some varying proportion of which may be borne externally to the marginal motorists. However, when calculating the marginal external benefit to ferry usage the best that can be said is that it is too close to zero to measure accurately with the information available, and possibly it is negative.

The total external benefit of accident avoidance through current total levels of ferry patronage is likely to be not insignificant, but the marginal external benefit from an incremental increase in ferry patronage is too small to measure reliably. Given these problems with measuring the marginal external benefits of ferries in reducing accident costs, I do not attempt a quantification of this type of externality.

5.4 Iconic value of Sydney Ferries and other externalities

Aside from emissions, congestion, and accidents, a number of potential external costs or benefits of transport are frequently discussed. Among the most commonly mentioned are:

Iconic value of Sydney Ferries;

Avoidance of over‐use of other public transport modes that are more heavily

subsidised than ferries;

Impacts on land use, including agglomeration benefits;

Promotion of social equity by providing equality of access to economic

opportunities and social services;

Promotion of mobility and social inclusion.

Concerning iconic value, it is often remarked, not least by Sydney Ferries itself, that ferry services on the Harbour are one of the quintessential features of Sydney as a city and tourist destination. The case is certainly arguable that part of Sydney’s attractiveness to tourists is the ability to enjoy the harbour’s splendours from a boat. Following on from that observation, as tourism generates spillover benefits to the wider Sydney and NSW economies, the availability of harbour cruising services could be said to generate external benefits.

That is not to say, however, that these wider benefits should be uniquely attributed to the Sydney Ferries organisation. While Sydney Ferries has a monopoly over Government‐owned commuter services on the harbour, it is certainly not the only provider of harbour cruising services, or even of harbour commuter services. If Sydney Ferries did not provide the tourist drawcard, some other organisation would provide an equivalent tourist experience. As a result, I conclude that any wider tourism benefits from the iconic nature of ferries on the harbour should not be


attributed to Sydney Ferries, nor should it influence the calculation of passenger fares or Government subsidy.

Concerning the avoidance of more heavily subsidised public transport services, the point is well taken that rail services, for example, require significant subsidies per passenger journey. The welfare consequences of this subsidy are comparable to external costs and should therefore be taken into account in estimating the optimal fares for ferries. In simple terms, the point is that one factor arguing in favour of somewhat lower ferry fares is the fact that if high ferry ticket prices persuaded travellers to use rail instead, the burden on taxpayers would increase.

When the formulae for optimal pricing are introduced in chapter 7 below, it will be shown that equation (9), which forms the basis of these price calculations, fully takes into account this rail subsidy effect on the optimal ferry price. Equation (12), developed in Appendix 1, can be used to show that the effect is even more pronounced when the marginal excess burden of taxation (required to fund the subsidy) is greater than zero.

Land use is most strongly affected by investments in public transport infrastructure, such as railway lines and stations, dedicated busways, and bus interchanges. Taking these investments as fixed, the setting of public transport fares would have a negligible impact on long‐term land use decisions.

Social equity goals would be met at lower total cost by providing targeted subsidies to disadvantaged families, rather than by discounting public transport fares to affluent and disadvantaged passengers alike. To a large extent, the benefits of mobility and social inclusion are private benefits of public transport. They form part of the passengers’ consumer surplus rather than an external benefit. Therefore, none of these benefits should properly be included in the calculation of optimal public transport fares.


5.5 Summary of externalities

The results of the marginal external cost calculations in this chapter are summarised in table 5.2 below.

Table 5.2 Summary of marginal external cost calculation for all modes

MARGINAL EXTERNAL COSTS car train bus ferry

VOT

($/hr)

car

occupancy

congestion

commuters ($/person‐km) 0.65 0 0 0 20 1.2

tourists ($/person‐km) 0.06 0 0 0 10 1.2

emissions petrol diesel

litre unleaded petrol/veh‐km 0.250 0 0 0 1.24 1.36 Beer $/litre

litre diesel/veh‐km 0 0 0.278 0.46 0.58 Watkiss $/litre (band 1)

litre unleaded petrol/person‐km 0.208 0 0 0 2.34 2.68 kg CO2/litre

litre diesel/person‐km 0 0 0.011 0.137 23 $/tonne CO2 price

health costs (Watkiss) ($/person‐km) 0.096 0.006 0.080

GHG cost ($/person‐km) 0.011 0.001 0.008

Total emissions cost ($/person‐km) 0.107 0.007 0.007 0.088 Rail figure from Karpouzis 2007, pp. 21‐22

accidents

by argument, mec ($/person‐km) 0 0 0 0

other externalities

by assumption, mec ($/person‐km) 0 0 0 0

Total mec ($/person‐km)

commuter 0.76 0.01 0.01 0.09

tourist 0.17 0.01 0.01 0.09


6 Current extent of road pricing

Given the dependency of optimal public transport fares on road pricing, it is important to estimate the difference between marginal social cost and price for private car use in Sydney. As noted earlier, the marginal social cost of a transport service is the sum of the marginal cost and the marginal external cost for that service.

The price of private car use can be itemised under the following headings:

marginal cost of vehicle operation;

fixed costs of car ownership;

accident insurance premiums;

risk‐weighted human cost of death or injury in an automobile accident,

including pain and suffering of family and friends.

tolls;

fuel excise tax; and

Government parking space levy.

The first two dot points can be combined into a single long‐run marginal cost of car ownership and use.9 The competitive nature of most markets for automobile inputs tends to make these prices converge to cost. It is assumed, therefore, that the price of private car use at least covers these long‐run marginal costs.

Accident insurance premiums serve to internalise a significant proportion of the risk‐weighted cost of accidents. It is important, when calculating the external accident costs, not to include these insured costs. The risk‐weighted human costs associated with accidents are assumed to form part of the car driver (and passenger) decision to travel by that mode. In other words, while this cost is not monetised, it is assumed to be included in the private traveller’s reckoning of the full private costs of automobile travel. For this reason, these human costs are also excluded from the calculation of external accident costs—they are assumed to be internalised.

9 Included are costs of vehicle ownership or lease, repairs and component replacements, and the non-tax component of the fuel price. Road infrastructure pricing is not included under this head.


Tolls on the Sydney Harbour Bridge and Harbour Tunnel apply only to Southbound journeys. The toll rate depends on the time of day. For the AM and PM peak periods, the toll is $4 per car. For the Inter Peak period it is $3, and for the Evening period it is $2.50.

The Federal Government applies a fuel excise tax of 38.143 cents per litre to unleaded petrol and diesel, from which $9.124 billion in tax revenue was raised Australia‐wide in 2006‐07. Of this sum, only $2.772 billion was spent on roads nationally by the Commonwealth.10 Assuming that the national proportions are valid for Sydney, these facts imply that 26.555 cents per litre of the excise tax (i.e., the fuel excise less the avoidable cost of road usage) could be considered a user contribution toward the marginal external costs of private car usage. This particular road price does not differ between peak and off‐peak use.

For several years the NSW Government has levied a tax on parking spaces in the CBD and several other concentrated commercial precincts.11 In 2010‐11 this levy is $2,040 per car space in the CBD and North Sydney.12

These price points are summarised in table 6.1 below.

10 BITRE (2009) Information sheet 37, pp. 3, 12.

11 The parking levy applies to Sydney CBD, North Sydney and Milson’s Point. At a somewhat lower rate, it also applies to Chatswood, St Leonards, Parramatta, and Bondi Junction.

12 Source: www.transport.nsw.gov.au/aboutus/psl.html.


Table 6.1 Government taxes on car users

Under the following assumptions, table 6.2 summarises the calculation of each of these elements of the road price in units of average price per person‐kilometre. The assumptions are that:

1. Tolls are a pure tax, as the LRMC is approximately zero for a harbour

crossing; and

2. Parking prices equal the sum of the parking space levy plus the LRMC of

providing a parking space. Therefore the parking space levy is the only

component of price that is in excess of cost.

Excise tax on unleaded petrol

tax rate 38.143 cents/litre

excise revenue 9124 $m in 2006‐07

Cwlth road funding 2772 $m in 2006‐07

excise not spent on road

mtce, etc. 26.555 cents/litre

Toll on harbour crossings $ per crossing Source: RTA web site

AM/PM Southbound 4

IP Southbound 3

EV Southbound 2.5

Parking space levy 2010‐2011 $/space/yr Source: Transport NSW web site

Category 1 areas 2040 CBD, North Sydney

Category 2 areas 720 Bondi Junction, Chatswood, Parramatta, St Leonards

Source: BITRE (2009) Information sheet 37, pp. 3,12


Table 6.2 Calculation of road price components in units of $/p‐km

Note that the commuter toll is higher than the tourist toll because tourists generally travel during the Inter Peak period when tolls are lower.

Finally, table 6.3 summarises the overall road price applicable to various types of car journeys.

Table 6.3 Overall road price for different journey types

p ‐ c for car journeys

avg litres/veh km 0.25

occupancy 1.2

avg litres/person‐km 0.21

pure fuel tax $/p‐km 0.06

avg car distance 1 way 10.73

occupancy 1.2

RT person km 25.75

commuter toll $/p‐km 0.16

tourist toll $/p‐km 0.12

commuting days/yr 249

CBD PSL/commuting day 8.19

RT person km per commute 25.75

PSL $/p‐km 0.32

$/person‐km Total p ‐ c for car journeys

0.06 No park, no harbour crossing

0.17 Tourist, harbour crossing, no leviable park

0.21 Commuter, harbour crossing, no leviable park

0.53 Commuter, harbour crossing, CBD parking

0.37 Commuter, no harbour crossing, CBD parking


Any significant change to road pricing, either in the form of an amended rate of petroleum products excise, cordon pricing, increased parking space levy, or more extensive tolling arrangements, would have implications for optimal public transport prices.

7 Optimisation of fares This chapter explains the method for calculating optimal ferry fares, and presents the results of the calculation.

7.1 Specification of optimisation problem

The optimal ferry price, pf*, is the marginal social cost of ferry service less an adjustment factor that depends on the optimality of prices for other transport modes. The marginal social cost is simply the sum of marginal cost and marginal external cost. Marginal costs were calculated in chapter 4. Marginal external costs for ferries and other transport modes were calculated in chapter 5.

In a first‐best world, all transport services would be priced at their own marginal social costs, implying zero adjustment factor. Otherwise, the adjustment factor is the weighted sum of the quantity (marginal cost + marginal external cost – price) for all transport modes other than ferry. That is, car, train and bus. Marginal external costs were calculated for each of these three modes in prior studies I have completed for IPART (Smart 2008 and Smart 2009). Marginal costs were also calculated for train and bus services in these studies.

Strictly speaking, neither the marginal cost nor the price for car use is known with any certainty. Nonetheless, it is reasonable to suppose that as most inputs to motoring are supplied in competitive marketplaces, their prices would closely approximate long‐run marginal costs plus any government taxes that apply. For this reason, the difference between price and marginal cost for motoring would simply be the sum of government taxes that vary with car use. This difference can be determined with a high degree of certainty, and that calculation was the subject of chapter 6.

The weightings used in the adjustment factor are the marginal rate of technical substitution between ferry and each of the other transport modes. This substitution rate is the answer to the following question. If 100 ferry passengers choose to travel by a mode other than ferry (either because the ferry price has risen or because ferry service is now unavailable), what percentage travel instead by car, train, and bus? The STM runs provide answers to that question for ferry services overall, and for ferry service on each of the routes.


Under the assumption that the marginal excess burden of taxation is near zero, and that there is no change to train or bus fares or to the rate of taxation of motoring, the formula for the optimal ferry price is given by equation (9) below. This equation is developed in Appendix 1.

pf * = cf + ∑h mecfh + ∑i<>f [ci – pi+ ∑h mecih] (∂X

i / ∂Xf) (9)

The straightforward interpretation of equation (9) is that optimal prices should equal the sum of marginal costs, marginal external costs, and an adjustment factor that compensates for any departures from optimal pricing that occur in competing transport modes. Evidently, if the price is equal to marginal cost plus marginal external cost for all of the other modes, then the adjustment factor is equal to zero.

Where one or more of these other modes requires significant subsidy, as in rail for example, it will have a downward influence on the optimal ferry price, all else being equal. This point can be seen by recognising that the need for subsidy implies that ci > pi. The modal substitution factor, (∂Xi / ∂Xf), is always negative. Therefore, the larger the rail or bus subsidy, the lower the optimal ferry price.

Prices, marginal costs and marginal external costs as calculated in chapters 4 to 6 and previous studies are summarised in table 7.1 below.

Table 7.1 Summary of p, c, mec by mode

Note that the marginal external cost for car trips during commuter periods is higher as a result of congestion effects. The extent of road pricing (price – marginal cost for cars) depends on whether the car journey involves CBD parking (in which case it

$/person‐km car train bus ferry

price 0.10 0.14 0.73

marginal cost 0.27 0.22 1.17

price‐marg cost ‐0.17 ‐0.09 ‐0.43

commuter/bridge/CBD pa 0.53

commuter/no bridge/CBD 0.37

tourist/bridge 0.17

tourist/no bridge 0.06

mec

commuter 0.76 0.01 0.01 0.09

tourist 0.17 0.01 0.01 0.09


bears the NSW Government parking space levy tax) and whether it involves the use of the Sydney Harbour Bridge or Harbour Tunnel (in which case it incurs the toll.) For tourist traffic involving no CBD parking and no toll, the road price is simply the fuel excise tax less an amount spent by the Commonwealth on road maintenance.

It is noteworthy that the marginal external cost of ferries is not much lower than that of non‐commuter car trips for the reason that ferries are a relatively fuel‐inefficient means of travel. Therefore, emission externalities per person‐km travelled are nearly as great for ferries as they are for car occupants.

The marginal rate of technical substitution for ferries can be calculated from the STM runs. Table 7.2 shows the STM substitution data for work trips. Table 7.3 shows the substitution data for trips with purposes other than work, work‐related, study‐related, and shopping.


Table 7.2 Modal substitution effects away from ferries for work trips (passengers per average weekday)

Where do ferry passengers go when ferry not available?

Work purpose

No_All_

Sydney

_Ferry

No_Manly

_Ferry

No_Parr

amatta_

Ferry

No_Mosman_

Zoo_Neutral_

Bay_Ferry

No_Wats

ons_Bay

_Ferry

No_Darling

_Harbour_

Blamain_C

ockatoo_F

erry

Car Driver 3,080 990 370 460 510 540

Car Passenger 760 210 110 110 120 160

Bus or rail 10,940 3,890 2,390 1,070 1,230 1,510

Bike 80 20 10 10 10 20

Walk 430 160 20 80 60 60

Taxi 80 20 10 20 20 20

Don't travel 40 ‐30 ‐30 ‐40 ‐30 120

Ferry ‐15,410 ‐5,260 ‐2,880 ‐1,710 ‐1,920 ‐2,430

delta car /delta

ferry ‐25% ‐23% ‐17% ‐33% ‐33% ‐29%

delta (bus or train)

/delta ferry ‐71% ‐74% ‐83% ‐63% ‐64% ‐62%


Table 7.3 Modal substitution effects away from ferries for trips with ‘other’ purpose (passengers per average weekday)

7.2 Results and analysis

The results of the optimal ferry fare calculation for commuter trips are shown in table 7.4 below, under the assumption of no marginal excess burden of taxation. It was possible to calculate separate optimal fares for each route.

Where do ferry passengers go when ferry not available?

Other purpose

No_All_

Sydney_

Ferry

No_Manly

_Ferry

No_Parr

amatta_

Ferry

No_Mosman

_Zoo_Neutr

al_Bay_Ferry

No_Wats

ons_Bay

_Ferry

No_Darling

_Harbour_

Blamain_C

ockatoo_F

erry

Car Driver 750 280 80 110 120 80

Car Passenger 270 80 20 30 50 40

Bus or rail 3,860 1,000 920 420 640 570

Bike 0 0 0 0 0 0

Walk 90 20 0 20 20 20

Taxi 10 0 0 0 0 0

Don't travel 20 0 0 10 20 70

Ferry ‐5,000 ‐1,380 ‐1,020 ‐590 ‐850 ‐780

delta car /delta

ferry ‐20% ‐26% ‐10% ‐24% ‐20% ‐15%


/delta ferry ‐77% ‐72% ‐90% ‐71% ‐75% ‐73%


Table 7.4 Optimal ferry fare calculation for commuter trips (other prices fixed, λ=0)

It was assumed that travellers on the Parramatta, Watsons Bay, Darling Harbour, Woolwich, and Balmain services would not incur the Harbour Bridge/Tunnel toll in the event that they drove instead.

While the optimal fare for ferry services overall is 48% higher than current fares, there are important differences between routes. Optimal fares are lower than current prices on the Manly and Watson’s Bay routes. Large price increases would be optimal for the Parramatta River and Woolwich routes.

The results of the optimal price calculation for tourist trips are shown in table 7.5 below.

ferry

Manly (excl

Jet Cat)

Parramatta

R

Taronga

Zoo Mosman

Neutral

Bay

Watsons

Bay

Darling

H/Balmain

Woolwich

/Balmain

work purpose mec+c‐p

car‐commuter/bridge 0.23 ‐25% ‐23%

train 0.18 ‐50% ‐46% ‐72% ‐46%

bus 0.09 ‐19% ‐25% ‐9% ‐17%

car‐commuter/no bridge 0.39 ‐17% ‐33%

sumproduct dXi/dXf * (mec+c‐p) = Z ‐0.17 ‐0.16 ‐0.20 ‐0.17 ‐0.17 ‐0.17 ‐0.22 ‐0.20 ‐0.20

ideal fare commuter p* = c + m 1.09 0.44 1.04 1.77 0.98 2.53 0.30 1.08 1.24

ideal fare ‐ price 0.35 ‐0.01 0.80 0.52 0.07 1.01 ‐0.03 0.09 0.69

% price increase 48% ‐3% 338% 42% 7% 66% ‐8% 9% 125%

ideal fare ($/pj) 6.44 4.93 22.32 5.78 4.37 6.76 3.73 4.44 9.14

dXi/dXf

‐33%

‐42%

‐19%

‐40%

‐13%

‐29%


Table 7.5 Optimal ferry fare calculation for tourist trips (other prices fixed, λ=0)

This table shows that for tourists, as congestion effects are less important, the adjustment factors are generally closer to zero. As a result, optimal prices are closer to ferry marginal costs for each route. Consequently, optimal fares are somewhat higher in this case than they were for commuters.

When the marginal excess burden of taxation is not zero, the taxation needed to fund subsidies induces distortions to consumption decisions that harm welfare. In this case, equation (12) is needed to derive the optimal ferry fare.

pf * = {(cf (1+λ) + ∑h mecfh + ∑i<>f [(ci – pi) (1+λ) + ∑h mecih] (∂X

i / ∂Xf)}

/ (1+λ+ λ/eff) (12)

Equation (12) is derived in Appendix 1. In order to evaluate equation (12) it is necessary to know the ferry own‐price elasticity. Hensher and Raimond (1996) provide a range of own‐price elasticities for ferries of approximately ‐0.2 to ‐0.44 (see table on p. 32). Price elasticity is lower for single tickets, higher for travel ten tickets, lower for commuters and higher for non‐commuters.

Adopting a mid‐range value of ‐0.35 for the own price elasticity and λ = 0.1, optimal prices are as shown in tables 7.6 and 7.7 below.

ferry

Manly (excl

Jet Cat)

Parramatta

R

Taronga

Zoo Mosman

Neutral

Bay

Watsons

Bay

Darling

H/Balmain

Woolwich

/Balmain

other purpose mec+c‐p

car‐tourist/bridge ‐0.00 ‐20% ‐26%

train 0.18 ‐73% ‐68% ‐86% ‐72%

bus 0.09 0% 0% 0% 0%

car‐tourist/no bridge 0.11 ‐10% ‐20%

sumproduct dXi/dXf * (mec+c‐p) = Z ‐0.13 ‐0.12 ‐0.17 ‐0.12 ‐0.12 ‐0.12 ‐0.15 ‐0.13 ‐0.13

ideal fare tourist p* = c + m 1.12 0.48 1.07 1.82 1.03 2.58 0.38 1.15 1.31

ideal fare ‐ price 0.39 0.02 0.84 0.57 0.12 1.05 0.04 0.16 0.76

% price increase 53% 5% 353% 46% 13% 69% 14% 16% 137%

ideal fare ($/pj) 6.64 5.36 23.08 5.94 4.59 6.89 4.62 4.72 9.66

dXi/dXf

‐15%

‐24%

‐68% ‐60%

0% 0%


Table 7.6 Optimal ferry fare calculation for commuter trips (other prices fixed, λ=0.1)

Table 7.7 Optimal ferry fare calculation for tourist trips (other prices fixed, λ=0.1)

Tables 7.6 and 7.7 show that the inclusion of the marginal excess burden of taxation has a significant upward influence on optimal ferry prices. This change reflects the fact that of all the public transport modes, ferry requires the highest Government subsidy per passenger kilometre (see table 7.1, “price – marg cost” row).

ferry

Manly (excl

Jet Cat)

Parramatta

R

Taronga

Zoo Mosman

Neutral

Bay

Watsons

Bay

Darling

H/Balmain

Woolwich

/Balmain

work purpose

mec+(c‐

p)*(1+λ)

car‐commuter/bridge 0.18 ‐25% ‐23%

train 0.20 ‐50% ‐46% ‐72% ‐46%

bus 0.10 ‐19% ‐25% ‐9% ‐17%

car‐commuter/no bridge 0.35 ‐17% ‐33%

sumproduct dXi/dXf * (mec+(c‐p)(1+λ)) = ‐0.16 ‐0.16 ‐0.21 ‐0.16 ‐0.16 ‐0.16 ‐0.22 ‐0.19 ‐0.19

ideal fare commuter

p* = (c(1+λ) +

mec +

Z)/(1+λ+λ/eff) 1.48 0.61 1.41 2.41 1.35 3.43 0.44 1.48 1.69

ideal fare ‐ price 0.75 0.16 1.17 1.16 0.44 1.91 0.10 0.49 1.14

% price increase 102% 34% 494% 93% 48% 125% 31% 49% 206%

ideal fare ($/pj) 8.77 6.83 30.22 7.88 6.01 9.17 5.35 6.07 12.46

dXi/dXf

‐33%

‐42% ‐40%

‐19% ‐13%

‐29%

ferry

Manly (excl

Jet Cat)

Parramatta

R

Taronga

Zoo Mosman

Neutral

Bay

Watsons

Bay

Darling

H/Balmain

Woolwich

/Balmain

other purpose

mec+(c‐

p)*(1+λ)

car‐tourist/bridge ‐0.02 ‐20% ‐26%

train 0.20 ‐73% ‐68% ‐86% ‐72%

bus 0.10 0% 0% 0% 0%

car‐tourist/no bridge 0.11 ‐10% ‐20%

sumproduct dXi/dXf * (mec+(c‐p)(1+λ)) = ‐0.14 ‐0.13 ‐0.18 ‐0.13 ‐0.13 ‐0.13 ‐0.16 ‐0.14 ‐0.14

ideal fare tourist

p* = (c(1+λ) +

mec +

Z)/(1+λ+λ/eff) 1.51 0.65 1.44 2.45 1.39 3.47 0.51 1.55 1.76

ideal fare ‐ price 0.78 0.19 1.20 1.20 0.47 1.95 0.18 0.56 1.21

% price increase 106% 42% 508% 97% 52% 128% 53% 56% 219%

ideal fare ($/pj) 8.94 7.23 30.96 8.01 6.18 9.27 6.23 6.36 12.98

‐15%

‐24%

‐68% ‐60%

0% 0%

dXi/dXf


8 Total external benefit An alternative perspective on optimal subsidy funding for a public transport service can be obtained by examining the total external benefits created. The taxpayer would ideally pay an amount equal to the external benefits generated by the service, while users of the service would pay for the remainder of the efficient cost of service. Under this conceptual framework, the Government would also bear the cost of inefficient operation, but could exert its power to change management of the service to motivate efficiency increases.

Figures for total cost and total subsidy to Sydney Ferries are readily available from official documents. It remains to translate the marginal external cost estimates derived in chapter 5 into total externalities in order to compare present subsidies with external values.

It is assumed that the marginal external cost rates for all modes are constant within the patronage range from zero to current ferry patronage. This assumption is valid because the total ferry patronage represents such a small percentage of car, train and bus patronage that the complete removal of ferry services would have only a slight impact on total road congestion.

The calculation of total external benefit is performed in the series of tables below.

8.1 Base case

Table 8.1 Differences in marginal external cost for ferry‐car and ferry‐(train or bus)

car train bus ferry ferry‐car ferry‐torb

Total mec ($/person‐km)

commuter 0.758 0.007 0.007 0.088 ‐0.670 0.081

tourist 0.168 0.007 0.007 0.088 ‐0.080 0.081


Table 8.2 Person‐km per workday by route, season and journey purpose

Table 8.3 Mode switching behaviour by route and journey purpose

The data in Tables 8.1 – 8.3 is derived from other parts of this report. For each combination of journey purpose, season and ferry route, a total external benefit to ferries is derived by multiplying daily patronage by (% of ferry journeys switching to car * marginal externality benefit from ferries relative to car + % of ferry journeys

Manly

(excl Jet

Cat)

Parramatt

a R

No_Mos

man_Zoo

_Neutral_

Bay_Ferry

Watsons

Bay

No_Darling

_Harbour_

Blamain_C

ockatoo_F

erry Total SF

commuter pkm/day

winter 53,704 51,272 11,226 21,015 14,169 151,387

summer 59,548 56,939 15,409 24,136 15,434 171,465

tourist pkm/day

winter 76,859 25,633 9,745 13,668 12,243 138,148

summer 107,427 46,111 19,315 23,887 21,196 217,935

Work purpose

No_All_Sy

dney_Ferr

y

No_Manl

y_Ferry

No_Parra

matta_Fe

rry

No_Mos

man_Zoo

_Neutral_

Bay_Ferry

No_Watso

ns_Bay_Fe

rry

No_Darling

_Harbour_

Blamain_C

ockatoo_F

erry

delta car /delta ferry ‐25% ‐23% ‐17% ‐33% ‐33% ‐29%

‐71% ‐74% ‐83% ‐63% ‐64% ‐62%

Other purpose

delta car /delta ferry ‐20% ‐26% ‐10% ‐24% ‐20% ‐15%

‐77% ‐72% ‐90% ‐71% ‐75% ‐73%


/delta ferry


/delta ferry


switching to either rail or bus * marginal externality benefit from ferries relative to rail or bus). The results of this calculation are shown in Table 8.4 below.

Table 8.4 Calculation of total external benefit to ferry service

Annual figures were derived from workday figures by multiplying by the number of work days in a year (249).

It is noteworthy that ferries provide a net external benefit for commuter journeys, owing to the congestion‐relief effect, but there is a net external disbenefit to ferries for tourist journeys. That arises because the congestion relief effect is relatively unimportant during the hours that tourists are most likely to travel, and ferries displace other public transport modes to a greater extent than they do private car use, leading to an adverse performance on emissions.

The total net externality effect is positive for Sydney Ferries, but modest. Overall, the total external benefit per annum is only $1.9m, which is orders of magnitude lower than the $58.7m Government funding of services noted in the 2009‐10 Annual Report for Sydney Ferries. The true extent of Government financial support for Sydney Ferries is larger than this figure. In 2009‐10, farebox revenue was only $36.7m while total expenses (including losses) were $137.9m, implying nearly a $100m cost to Government.

Total ext

benefit

$/day

No_Manl

y_Ferry

No_Parrama

tta_Ferry

No_Mos

man_Zoo

_Neutral_

Bay_Ferry

No_Wats

ons_Bay_

Ferry

No_Darlin

g_Harbou

r_Blamain

_Cockato

o_Ferry

Sum all

routes

comm‐w 4,986 2,271 1,938 3,529 2,021 14,744

comm‐s 5,528 2,521 2,660 4,053 2,201 16,963

tourist‐w ‐2,929 ‐1,679 ‐380 ‐619 ‐577 ‐6,184

tourist‐s ‐4,094 ‐3,020 ‐752 ‐1,081 ‐999 ‐9,948

all‐w 2,056 592 1,558 2,910 1,444 8,560

all‐s 1,434 ‐499 1,907 2,971 1,202 7,016

$/day 1,745 46 1,733 2,941 1,323 7,788

$/annum 434,511 11,535 431,460 732,238 329,388 1,939,133


8.2 Sensitivity to assumptions

In light of the unexpectedly small quantum of total external benefits to Sydney Ferries’ services it is important to consider whether plausible changes to the input assumptions could lead to significant changes in these results. The main uncertainties involved in the externality calculation were the value of time used in the congestion cost estimates and the carbon price used in the emissions cost estimates.

Higher values of time would tend to place more weight on the congestion‐reducing benefits of ferries, tending to increase the estimated total external benefit. A lower carbon price would tend to reduce the gains in social welfare arising from tourists switching to rail or bus travel. The switch to rail or bus reduces the quantum of carbon dioxide released by the relatively high‐emission ferries.

The most extreme plausible values for these uncertain inputs are value of time = $40/hr for commuters and $20/hr for tourists, and carbon price of zero. With these settings, the total external benefit from Sydney Ferries would be as shown in Table 8.5 below.


Table 8.5 Sensitivity analysis: total external benefit to ferry service (VOT x 2, CO2 price = zero)

Table 8.5 shows that, even under the most extreme plausible assumptions regarding the uncertain inputs, the total external benefit created by Sydney Ferries at its current service levels is less than $9.2m per annum, which is a small proportion of the total Government support currently provided.

9 Conclusions Based on the analysis in this report it is possible to reach the following conclusions about various economic aspects of Sydney Ferry services.

1. Overall, tourism‐based demand for ferry services represents roughly half of

all patronage—considerably more on some routes.

2. The Short Run Marginal Cost of ferry service is lower than the fare implied by

Travel Ten ticket prices for the Manly, Mosman, Neutral Bay and Darling

Harbour services. The Long Run Marginal Cost of ferry service is higher than

the implied fare on every route.

Total ext

benefit

$/day

No_Manl

y_Ferry

No_Parrama

tta_Ferry

No_Mos

man_Zoo

_Neutral_

Bay_Ferry

No_Wats

ons_Bay_

Ferry

No_Darlin

g_Harbou

r_Blamain

_Cockato

o_Ferry

Sum all

routes

comm‐w 13,243 8,146 4,420 8,107 4,737 38,654

comm‐s 14,684 9,046 6,067 9,311 5,160 44,268

tourist‐w ‐1,333 ‐1,353 ‐191 ‐380 ‐398 ‐3,655

tourist‐s ‐1,863 ‐2,434 ‐379 ‐664 ‐689 ‐6,029

all‐w 11,911 6,793 4,229 7,727 4,339 34,998

all‐s 12,821 6,613 5,687 8,647 4,471 38,239

$/day 12,366 6,703 4,958 8,187 4,405 36,619

$/annum 3,079,135 1,669,016 1,234,528 2,038,508 1,096,848 9,118,034


3. For the Parramatta River and Woolwich services, Travel Ten prices are less

than half the Long Run Marginal Cost, and substantially less than even the

Short Run Marginal Cost.

4. On average across all Sydney Ferries routes, pricing falls well short of Long

Run Marginal Cost.

5. External emissions costs for ferries are approximately the same as for

automobiles on a $/person‐km basis because ferries are a relatively fuel‐

inefficient means of transport and average seat occupancy is low. This

means that no advantage to society with respect to emissions costs would

be obtained by convincing car occupants to switch to ferries.

6. While there is a significant road congestion relief benefit from ferries during

commuter hours, approximately half of all ferry demand takes place outside

these hours.

7. Non‐peak ferry usage creates only very slight road congestion relief

benefits—approximately only 10% of the benefit in $/person‐km achieved in

the peak.

8. The great majority of travellers who elected to switch from Sydney Ferries

services in the event of a price increase would go to other public transport

modes, such as bus or rail, rather than car. These modes would have

superior external benefits in terms of emissions and similar congestion relief

benefits compared to ferries. As a result, it is possible if not likely that a

switch away from ferries could actually be welfare‐enhancing.

9. Optimal ferry fares would be close to Long Run Marginal Cost because the

external benefits associated with ferry use are relatively minor. This is

particularly so for tourist demand for ferry services.

10. The total external benefit from Sydney Ferries is orders of magnitude

smaller than the current level of Government financial support. This

conclusion is robust to alternative assumptions about uncertain input values

concerning the value of travel time savings and the carbon price.


References

Auditor-General’s Report to Parliament (2008), Volume Five, 295-300.

Australian Automobile Association (2008), ‘Motoring matters factsheet: Fuel

Tax Reform and Road Pricing.’

Australian Bureau of Statistics (ABS 2003), ‘Survey of Motor Vehicle Use’.

Australian Transport Council (2006), National Guidelines for Transport System

management in Australia, Volume 3: Appraisal of initiatives.

Beer, T. (August 2002), ‘Valuation of pollutants emitted by road transport into

the Australian atmosphere.’ Proceedings of the 16th International

Clean Air & Environment Conference. Christchurch, New Zealand: 86-

90.

Bureau of Infrastructure, Transport and Regional Economics (2009), Information

sheet 37: ‘Public road-related expenditure and revenue in Australia

2009.’

Bureau of Transport Economics, (2000), ‘Report 102: Road crash costs in

Australia’.

De Borger, B., Mayeres, I., Proost, S., and Wouters, S. (January 1996), ‘Optimal

Pricing of Urban Passenger Transport.’ Journal of Transport Economics

and Policy. 31-54.

Freebairn, J. (2002), ‘Opportunities to Reform State Taxes.’ Australian

Economic Review, vol. 35, no. 4, pp. 405-22.

Gabbitas, O., and Eldridge, D. (May 1998), ‘Directions for State Tax Reform.’

Productivity Commission Staff Research Paper, AusInfo, Canberra.

Glaister, S. and D. Lewis (1978), ‘An Integrated Fares Policy for Transport in

London’. Journal of Public Economics, vol.9,pp.341-55.

Hensher, D. (2009), ‘Attribute Processing, Heuristics, and Preference

construction in Choice Analysis,’ First International Conference on

Choice Analysis, Harrogate, UK, March 29 – April 3 2009.

Hensher, D. and Raimond, T. (1996), ‘Evaluation of Fare Elasticities for the

Sydney Region,’ Institute for Transport Studies, University of Syddney.

IPART Research Paper No. 7 (1996).


Jansson, J.O. (January 1994), ‘Accident Externality Charges.’ Journal of

Transport Economics and Policy. 31-43.

Karpouzis,G. (2007), ‘Value of CityRail to the NSW community.’

Newbery, D. (1987), ‘Road user charges in Britain’. Discussion Paper 174,

Centre for Economic Policy Research.

Parry, I. and K. Small (2007), ‘Should Urban Transport Subsidies Be Reduced?’.

Discussion Paper RFF DP 07-38, Resources for the Future,

Washington D.C.

Peirson, J., Skinner, I., and Vickerman, R. (1998), ‘The Microeconomic Analysis

of the External Costs of Road Accidents.’ Economica 65. 429-40.

Small, K. (1983), ‘The Incidence of Congestion Tolls on Urban Highways’.

Journal of Urban Economics, vol.13, pp. 90-111.

Small, K. (1992), ‘Urban Transportation Economics’. Fundamentals of Pure and

Applied Economics, Volume 51, Harwood Academic Press, Chur,

Switzerland.

Smart, M. (2008), ‘An empirical estimate of CityRail’s marginal costs and

externalities’. Final report prepared for IPART.

http://www.ipart.nsw.gov.au/files/Consultancy%20Report%20-%20LECG%20Report%20CityRail%20externalities%20and%20marginal%20costs%20final%20-%2020%20November%202008%20-%20WEBSITE%20DOCUMENT.PDF

Smart, M. (2009), ‘Value of Sydney bus externalities and optimal Government

subsidy’. Draft report prepared for IPART.

http://www.ipart.nsw.gov.au/files/Consultancy%20Report%20-%20LECG%20Draft%20report%20on%20Value%20of%20Sydney%20bus%20externalities%20and%20optimal%20Government%20Subsidy%20-%2012%20May%202009.PDF

Sydney Ferries (2011) Annual Report.

Vickrey, W. (1968), ‘Automobile accidents, tort law, externalities and insurance:

an economist’s critique’. Journal of Contemporary Law and Problems,

Summer, 464-84.

Viton, P.A (1980), ‘The Possibility of Profitable Bus Service’. Journal of

Transport Economics and Policy, vol. 14, no.3, pp.295-314.


Viton, P.A (1983), ‘Pareto-Optimal Urban Transportation Equilibria’. In Keeler,

T.E. (ed) Research in Transportation Economics, Vol.1, pp.75-101.

Greenwich, Connecticut: JAI Press.

Watkiss, P. (2002), ‘Fuel Taxation Inquiry: The air pollution costs of transport in

Australia,’ A Report for the Commonwealth of Australia, March.


Appendix 1: Technical formulation of fare optimisation problem

Important foundations for the estimation of socially optimal public transport fares were established by Glaister and Lewis (1978), Small (1983) and Viton (1980, 1983). More recently, De Borger et. al. (1996), building on these foundations, set out optimal public transport pricing rules in a partial equilibrium model of the transport sector. Following De Borger et. al., I adopt a Bergson‐Samuelson type consumer surplus function W, which depends only on the indirect utility functions of individuals comprising the society.

The indirect utility of an individual Vh is a function of that individual’s full13 income Yh,

the out‐of‐pocket price for each of n transport services pi, the average speed for each transport service, the level of environmental pollution by transport services in total, and the number of transport accidents for each transport service.

Normally one would think of travel by car, train, bus and ferry as separate transport services. This could be taken further. Peak and off‐peak travel by each of these modes could be treated as separate transport services. Subscripts h refer to individuals. Superscripts i and j refer to different transport services.

The optimisation problem is to select a set of prices for all transport services that maximises the sum of consumer and producer surplus. Where a public subsidy is needed to fund the public transport deficit, the producer surplus is negative. For each set of transport prices, I assume that individuals will maximise their own utility. The overall welfare function that we wish to maximise is:

F = MAX W(V1, …Vh, …VH) + (1 + λ) [Σ (piXi – Ci) – FC] (1)

The first term, MAX W, is consumer surplus maximised by the decisions of individuals, subject to the constraints posed by transport prices. The second term is the producer surplus for transport providers as a whole. For public transport services this is expected to be negative, reflecting a deficit that must be funded from taxation. λ represents the marginal excess burden of taxation (a measure of

13 Full income refers to the sum of monetary and non-monetary consumption ability of an individual.


the economic inefficiency of the state taxes used to raise the funds). Xi is the total number of passenger kilometres travelled by all individuals on transport service i. Ci is the total cost of transport service i at patronage level Xi. FC is the fixed cost of all transport services combined. The summation in the second term of equation (1) is over transport services i=1..n.

The partial derivatives of F with respect to each of the transport prices pj will be zero at an interior optimum (the “first‐order condition” for optimality). It remains to be seen whether these interior optimum points exist and whether they do indeed correspond to welfare maximising prices. I will examine that question later in this report, once the input values have been calculated. The first‐order conditions can be written as follows:

∂F/∂pj = 0

= ∑h (∂W/∂Vh)(∂Vh/∂Yh)(∂Yh/∂pj) + (1 + λ) [Xj + ∑i (p

i – ∂Ci/∂Xi) (∂Xi/∂pj)] (2)

Equation (2) can be simplified by noting the following facts.

(∂W/∂Vh)(∂Vh/∂Yh) represents the marginal social utility of income to individual h. For simplicity, I assume that this quantity is equal to one for all individuals. In reality, lower‐income individuals are likely to have a higher social utility of income than high‐income individuals. However it would add unmanageable complexity to the fare calculation to attempt to take account of this effect here.

The elasticity of demand for transport service i with respect to the price of transport

service j is eij = (∂Xi /Xi) / (∂pj /pj). When i=j this quantity is the own‐price elasticity.

Otherwise it is a cross‐price elasticity. Rearranging terms,

(∂Xi/∂pj) = eij Xi /pj

∂Ci/∂Xi is the marginal cost of transport service I, which I shall call ci.

Applying these simplifications to equation (2):

∑h (∂Yh/∂p

j) + (1 + λ) [Xj + ∑i (pi – ci) eij Xi /pj ] = 0 (3)

(∂Yh/∂pj) is the change in an individual’s full income with respect to a change in the

price of transport service j. It has two elements: direct and indirect. The direct effect is on the individual’s own travel costs: pj xj

h , where xjh is defined as the

number of kilometres individual h travels on service j.

The indirect effect comes about because a change in the price of transport service j will affect Xj. In turn, a change in Xj will influence individual h’s utility through its effect on transport speeds, environmental pollution, and transport accidents—the


externalities, in other words. Quantitatively, this influence on individual utility with a given income is equivalent to the influence on the same individual’s full income with a given utility level, as a result of the duality between Marshallian and Hicksian demand.14 Given these observations,

(∂Yh/∂pj) = – xj

h – ∑i mecih (∂Xi/∂pj) (4)

where “mec” refers to the marginal external cost imposed on individual h by transport service i.

Combining equations (3) and (4), rearranging to group like terms and simplifying:

∑i e

ij Xi [(1 + λ)(ci – pi)+ ∑h mecih] = λ pjXj (5)

Note that equation (5) is actually a system of simultaneous equations. There is one equation for each of the n possible values of j. Each one must be satisfied for the price set to be optimal (unless there is a corner solution).

In the case where the marginal excess burden of taxation (λ) is zero, equation (5) can be reduced to a more intuitive form:

∑i e

ij Xi [ci – pi+ ∑h mecih] = 0 (6)

This version of equation (5) implies that when all prices are equal to the sum of marginal costs and marginal external costs for their respective transport services, prices are optimal. It is important to recognise, however, that if price does not equal the sum of marginal cost and marginal external cost for any one of the transport services (for example cars), then it is not optimal for the remaining prices (i.e., public transport fares in this example) to equal the sum of their own marginal costs and marginal external benefits. Rather, the imbalance in car prices (the lack of sufficient road pricing) implies that public transport fares would need to be subsidised to achieve a welfare‐optimal modal split.

Equation (6) can be rearranged to solve for optimal ferry prices, on the assumption that prices for other transport services are fixed at their current values. Let the superscript i = f for ferries.

14 Marshallian demand is the consumption bundle that maximises individual utility given prices and an income constraint. Hicksian demand is the consumption bundle that minimises individual expenditure, given prices and a minimum utility constraint. These two demands must be equal.


pf * = cf + ∑h mecfh + ∑i<>f [ci – pi+ ∑h mecih] (e

ij Xi / efj Xf) (7)

Recalling that eij = (∂Xi /Xi) / (∂pj /pj), the expression eij Xi / efj Xf can be simplified

further:

eij Xi / efj Xf = (∂Xi / (∂pj /pj)) *((∂pj /pj)/ ∂Xf )) = ∂Xi / ∂Xf (8)

Therefore, the optimal ferry price when λ = 0 and other prices are fixed is:

pf * = cf + ∑h mecfh + ∑i<>f [ci – pi+ ∑h mecih] (∂X

i / ∂Xf) (9)

The intuition behind equation (9) is that when the marginal excess burden of taxation is zero, the optimal ferry price is the sum of marginal cost and marginal external costs of ferry service plus the weighted sum of the difference between price and marginal social costs for each of the other transport modes. The weights are the marginal rate of technical substitution between ferry and each of these other modes. To the extent that other modes are under‐priced relative to their own social costs, optimal prices would be reduced below marginal social costs for ferries.

In the case where the marginal excess burden of taxation (λ) is not zero, equation (5) can be reduced to a somewhat more complicated form:

( cf – pf *) (1+λ) + ∑h mecfh + ∑i<>f [(ci – pi) (1+λ) + ∑h mecih] (∂X

i / ∂Xf)

= (λ pjXj/ efj Xf) (10)

Equation (10) must be valid for all values of j, including f. Where j=f, equation (10) can be reduced to:

( cf – pf *) (1+λ) + ∑h mecfh + ∑i<>f [(ci – pi) (1+λ) + ∑h mecih] (∂X

i / ∂Xf)

= λ pf */eff (11)

Here, eff is the ferry own‐price elasticity. Gathering the p* terms, the optimal ferry ticket price in the presence of non‐zero marginal excess burden of taxation is:

pf * = {(cf (1+λ) + ∑h mecfh + ∑i<>f [(ci – pi) (1+λ) + ∑h mecih] (∂X

i / ∂Xf)}

/ (1+λ+ λ/eff) (12)


Appendix 2: Input data derived from the literature

9.1 Value of travel time

In order to convert the STM outputs into dollar values of marginal external benefit it is necessary to establish values of travel time, and then apply them to the passenger hours for inframarginal users calculated for each mode in each model run.

The range of values of travel time used in sensitivity analysis was:

A low value of $9.23/hr, representing the value per occupant of travel time

for private use of a car;15 and

A high value of $22.60/hr, representing a weighted average of business and

private travel in passenger cars in urban areas.16

Both reference sources cite a 2004 Austroads publication as the primary source.17

In order to compare these values with hourly rates of pay, I note that, according to the ABS catalogue number 6306.0, “Employee earnings and hours, Australia, May 2006,” the average hourly rate of pay across all full‐time employees, for ordinary time was $26.00/hr. Ordinary time best matches the peak commuter travel profile. ABS catalogue number 6302001 indicates that average weekly earnings for persons in full‐time work during ordinary hours increased by 7.7% between May 2006 and February 2008, suggesting that the February 2008 hourly rate of pay had increased to $28.01/hr. ABS catalogue number 63020011a permits an inference to be made of the NSW average weekly earnings compared to the Australian average weekly earnings in both May 2006 and February 2008. Putting this information together, a

15 Centre for International Economics (August 2006), “Business costs of traffic congestion,” Prepared for Victorian Competition and Efficiency Commission, Table 4.1, p. 20.

16 Marschke, K., L. Ferreira, J. Bunker (2005), “How should we prioritise incident management deployment?,” Proceedings 28th Australasian Transport Research Forum, Sydney, Table 4, p. 7.

17 Austroads (2004). Guide to Project Evaluation Part 4: Project Evaluation Data. Sydney.


February 2008 NSW average hourly rate of pay for persons in full‐time employment during ordinary hours of $28.80/hr is derived. The ABS does not routinely collect city‐specific data on hourly wages or weekly earnings, so it is difficult to make this figure more geographically specific than NSW.

The low time valuation of $9.23/hr would be approximately 32% of this $28.80 hourly wage figure, and the high time valuation of $22.60/hr would be approximately 78% of the hourly wage. It is relatively common practice to link the value of travel time to the prevailing hourly wage, however the literature reveals considerable dispersion in the measured ratio of value of time to hourly wage. For example, BTE Occasional Paper 51 calculates and presents the ratio of value of travel time to average wage rate implicit in the travel time valuations contained in a range of studies.18 Table 8.1 in that paper presents the ratio for business values of travel time. Of the 27 references cited there that are not assumed values, the mean ratio is 83.8%, the median ratio is 76%, and the standard deviation is 62.7%. Table 8.3 of the BTE paper presents the ratio for commuter values of travel time. Of the 71 references cited there that are not assumed values, the mean ratio is 43.5%, the median ratio is 35%, and the standard deviation is 25.8%.

For business travel, the median ratio applied to the $28.80/hr wage would be $21.89/hr. For commuter travel, the median ratio applied to the hourly wage would be $10.08/hr. There is necessarily a degree of imprecision in these ratios. Rather than attempt to refine the estimates further, I adopt a central case value of time of $15.80/hr, which lies approximately midway between the median ratios for business and commuter travel applied to the hourly rate. For sensitivity testing I retain the range mentioned above: low valuation of $9.23/hr and high valuation of $22.60/hr.

Separate values of time for motorists, bus passengers and rail passengers19 have not been adopted, but the analytical framework set out here could easily be adapted to reflect mode‐specific values of time.

9.2 Fuel consumption

Fuel consumption was estimated as follows. The web site:

18 “The Value of Travel Time Savings in Public Sector Evaluation,” BTE Occasional Paper 51, AGPS, Canberra, 1982.

19 There is some evidence that automobile commuters tend to have higher valuations of travel time than public transport commuters, possibly because average incomes are higher among motorists.


http://www.climatechange.gov.au/cgi‐bin/transport/fuelguide/fuelguide.pl?querytype=advancedquery&min_cons=&max_cons=&manufacturer=any&year=2003&transmission=any&fuel=any&vehicletype=any&model=&minenginesize=&maxenginesize=&mincityfuel=&maxcityfuel=&minhighwayfuel=&maxhighwayfuel=&sort1=manufacturer&sort2=year

contains highway and city consumption figures for each of approximately 980 different 2003 models of passenger cars in use in Australia. The simple average of highway consumption of these vehicles was 7.2 litres per 100 km. The average of city consumption was 10.8 litres per 100 km.

I assumed that the city consumption figure applied to the speed band between 30 and 35 km/hr,20 and that the highway figure was relevant to the speed band between 80 and 85 km/hr. Fuel consumption rates for intermediate speed bands was calculated by linear interpolation between these points. The fuel consumption rate was assumed to remain constant for speeds above 85 km/hr. The rate of fuel consumption was assumed to rise as speed dropped below 30 km/hr. The resulting fuel consumption rates are shown in Table 8.1 below.

Table 8.1 Assumed rates of automobile fuel consumption as a function of speed

20 This figure is roughly consistent with average automobile speeds predicted for STM model runs.


9.3 Cost of greenhouse gas emissions

The assumed relationship between fuel consumption and the quantity of CO2 emitted was 2.64 kg CO2 per litre of fuel consumed. That figure is between the fuel conversion rates cited by

www.nqclimatealliance.org.au/Business_Travel_ServiceSector_v2.0_Final.xls

for petrol (2.34) and diesel (2.68).

For the analysis I have adopted the Federally established carbon price of $23/tonne CO2.

Appendix 3: STM ferry modelling issues

Req11056 – Ferry Externality Study

Prepared by Blake Xu on 5 April 2011

min max0 5 0.321 5 10 0.285

10 15 0.250 15 20 0.215 20 25 0.179 25 30 0.144 30 35 0.108 35 40 0.104 40 45 0.101 45 50 0.097 50 55 0.094 55 60 0.090 60 65 0.086 65 70 0.083 70 75 0.079 75 80 0.076 80 85 0.072 85 90 0.072 90 95 0.072 95 100 0.072

100 105 0.072 105 110 0.072

Speed band km/hrlitres fuel consumed by cars / vkm


Reviewed by Peter Hidas on 5 April 2011

1 Introduction

This technical note summarises the key points of the methodology developed under the objectives of this project, which was to use the STM to predict changes of transport usage by mode and purpose, from business as usual to variations of availability of the Sydney ferry services operated by government in 2011. Private ferry services were excluded during this study.

2 Scenarios

Seven scenarios for 2011 were developed to represent the different levels of availability of the Sydney ferry services as specified by the client:

2011_s1: Business as usual, including all 2011 Sydney Harbour ferry services

2011_s2: Complete absence of the Sydney Harbour ferry services

2011_s3: No Manly route ferry services

2011_s4: No Parramatta River route ferry services

2011_s5: No Taronga Zoo, Mosman, and Neutral Bay route ferry services

2011_s6: No Watsons Bay route ferry services

2011_s7: No Darling Harbour, Balmain, Woolwich and Cockatoo Island route ferry services.

The ferry services in STM in terms of itineraries were developed from 2011 ferry timetables published in the Sydney Ferry website, using the AM peak hour services. The scenarios were established through disabling different parts of the ferry itineraries according to the scenario definitions.

3 Ferry Modelling in STM

It is known that standard STM runs generate less ferry demand than expected. There are three reasons for this:

1. Sydney ferry services attract a large proportion of patronage from tourists. But STM models trips / tours generated by Sydney local residents only, and the trips generated by non-private dwellings such as tourists are excluded.

2. The low number of OD-pairs with a ferry option in the STM mode-destination choice set which is related to the STM public transport generalised cost calculations. The key components of the generalised cost


are public transport fares, access time, waiting time, boarding time and in-vehicle time. The inclusion or exclusion of the ferry option for a particular OD pair depends on the comparable generalised cost values of the ferry vs. other transport modes between the same OD pair.

3. The “main mode” definition adopted in STM which refers to the mode hierarchy of rail, light rail and ferry. That is, rail is prioritised over the other two modes in allocating demand when multi-modes of rail, light rail and ferry were used. For example, a trip using ferry and rail during its journey would be counted as rail demand, instead of ferry demand.

This project aimed to address the above points 2 and 3 because the point 1 is beyond the scope of this project.

Firstly, to attract adequate OD pairs to be considered in the STM modal destination choice set, the boarding and waiting factors for ferry and light rail were lowered, to improve the ferry generalised cost values in comparison with those of other modes. Zero boarding time factor and 0.01 waiting factors were adopted for ferry and light rail, while standard STM factors (5 boarding factor and 1 waiting factor) were used for rail and bus. As a result, public transport line and node specific parameters were created and adopted during this project.

Secondly, to restore the ferry demand which might be allocated to rail or light rail demand according to the existing mode hierarchy, a new hierarchy was adopted for this project, which follows the order of ferry, rail and light rail. The investigation shows that the change of the hierarchy led to observable improvements of the STM raw ferry demand.

In the first set of model runs rail was also coded with lowered boarding and waiting factors, which caused some unrealistic demand split between rail and bus. It was observed for example, that when Manly ferry services were disabled, some travellers from the vicinity of Manly travelling to Wynyard, would ride a bus to Milsons Point, transferring to rail to Wynyard (zone 110), although the bus service continues from Milsons Point to the City. The issue was largely solved in the second set of model runs when the rail boarding and waiting factors were not lowered at all. However, the issue remains for other zones such as the travel from zone 2871 (near Manly beach) to Wynyard (zone 110). This issue is more related to the bus route network adopted in STM. The consequence of the unexpected transfer is that some of the diverted ferry users due to unavailability of Manly services are counted as new rail users instead of bus users although the rail part of the entire journey is relatively short.

When the above two STM adjustments adopted for this project are in place, “detailed analysis” shows the model can produce reasonable mode choice and ferry demand when part of ferry services disappears based on the needs of the scenarios. However, it should be understood that the results are still incomplete without the tourist demand that is not modelled in the STM.

4. Road Traffic


Another aspect of the investigation during this project was to look into the traffic variations through Sydney Harbour Bridge and Harbour Tunnel across the ferry scenarios. The traffic volume was separated by car driver and car passenger. Car passenger demand was derived in addition to car driver demand from STM runs and path-based multi-class assignment was used to assign the car driver and car passenger demand matrices for the AM peak period. The traffic flows at the other two locations (Ryde Bridge and Gladesville Bridge) were also analysed and compared. Very little variations were observed between the scenarios.

5. Results

Key outputs prepared for this project include:

- For average weekday * trips by mode and purpose

* person km by mode and purpose

* person hour by mode and purpose

* rail, bus, light rail and ferry service statistics for AM peak 1 hour

* average rail, bus, light rail and ferry trip distance and time

- For AM peak 2 hours in average weekday

* traffic flows by car driver and car passengers at Sydney Harbour Bridge and Tunnel, Ryde Bridge and Gladesville Bridge.

Documents

benefits of Sydney Ferry report to IPART - … · External benefits of Sydney Ferry ... Mike Smart is a director of Sapere Research Group in Sydney. He advises industry leaders in