Australia's Future Tax System

Final Report: Detailed Analysis

Chapter E: Enhancing social and market outcomes

E3. Road transport taxes

E3–4 How should roads be financed?

Recommendation 65:

Revenue from fuel tax imposed for general government purposes should be replaced over time with revenue from more efficient broad-based taxes. If a decision were made to recover costs of roads from road users through fuel tax, it should be linked to the cost of efficiently financing the road network, less costs that can be charged directly to road users or collected through a network access charge. Fuel tax should apply to all fuels used in road transport on the basis of energy content, and be indexed to the CPI. Heavy vehicles should be exempt from fuel tax and the network access component of registration fees if full replacement charges are introduced.

Recommendation 66:

The revenue-raising component of State taxes on motor vehicle ownership and use should be made explicit, and over time only be used to recover those costs related to road provision. The administrative costs of providing government services should be recovered through user charges where applicable. Quantity limits on taxi licences should be phased out.

The Review has been asked to consider taxation arrangements necessary to deal with the demographic, social, economic and environmental challenges of the 21st century. The best structure of road-related taxes depends on whether roads are funded from general taxation revenue, or whether road costs should be recovered directly from road users.

The congestion and road damage charges suggested earlier in this section would not cover the costs of the entire road network, particularly if the revenue from congestion charges on existing roads is invested in public transport infrastructure, or some other form of community compensation. Because most of the road network is not congested, and the road-wear costs of cars are negligible, efficient usage charges do not generate enough revenue to finance road building.

In addition, there are substantial operating costs that cannot be attributed to heavy vehicle road-wear, including time-related pavement deterioration, routine maintenance (cutting grass, maintaining roadside furniture, clearing drains), traffic management, regulatory and policing costs, and accident-related costs that are not paid for by road users either directly or through insurance (see Box E3–6).

Uncongested roads as public goods or cost recovery

Car travel on uncongested roads has the public good characteristic of non-rivalrous consumption — that is, an additional car has negligible impact on other road users and causes no pavement wear (see Section E1 User charging). A price for using a given road deters some people from travelling on the road. Costs of pavement wear associated with weathering are non-rivalrous to all road users, trucks as well as cars. The inefficiency of tolling something that costs nothing to use was first identified by Dupuit (1844), who observed some people going out of their way to avoid a toll-bridge across the Seine River in Paris.

Private network industries (such as electricity, gas and communications) finance the supply of infrastructure by making access to it excludable. They are funded by block tariffs or two-part tariffs (a fixed and a variable charge) set to reflect the financial costs of supply. Most Australian roads, on the other hand, are not specifically excludable, and have been funded over many decades from general tax revenue. Only a few major projects have been funded from specific tolls. The entire road system, however, is excludable, through the requirement that motor vehicles be registered.

New roads, bridges or tunnels built in urban areas are likely to become immediately congested if they are unpriced. Charging a variable congestion toll on these roads would be an efficient way to manage demand for a road, and could also make a contribution to the capital cost of the road itself. Similarly, new roads in development areas might be financed by infrastructure charges, which are discussed in Section E4 Housing affordability.

However, for large parts of the road network, the social opportunity cost of letting another car enter the road is negligible, because there is almost always significant spare capacity. The explanation for this is the nature of road investment — a road must be built or expanded at least one lane at a time, even if it is used by only one car an hour. If roads and cars were perfectly divisible, it would be more efficient to build a cheaper road that is 1/60th of a lane wide but in use almost all the time! Moreover, improvements to road quality to improve safety or travel times (such as straightening roads) also increase their redundant carrying capacity.

Box E3–6: Efficient road pricing and capital costs

Prices that lead to efficient use of roads are unlikely to lead to full capital cost recovery. Economic models suggest that revenue from congestion charges set at efficient levels would approximately cover the full costs of the roads to which they apply. However, for most of the road system there is no congestion most of the time and hence no warrant for congestion charges. This is the case for rural roads, suburban streets and major highways outside of cities. If only congestion and road-wear are priced, over large parts of the road network cars would pay nothing and trucks would be charged only for road-wear.

Much of the explanation lies in the economies of scale in road provision. With efficient investment in capacity, short-run marginal cost and long-run marginal cost become equal (or with time-variable pricing, the sum of short-run marginal costs over an entire cycle equals long-run marginal cost for the cycle). Economies of scale mean that long-run marginal cost, and hence the economically optimal price, lies below long-run average cost. If investment in capacity were restricted to the point where the short-run marginal cost price was as high as the long-run average cost, the economic benefit of increasing capacity would exceed the costs of doing so.

Costs of infrastructure along the sides of roads (shoulders, signs, guide posts, drainage ditches) are a source of economies of scale for non-urban roads. For example, a four-lane road requires the same shoulder widths as a two-lane road. Also, because of the greater passing opportunities, a four-lane road has more than twice the capacity, giving rise to economies of scale (Hau 1992).

Economies of scale are exacerbated by capacity–quality interactions. Investment to improve road standards by building a wider, smoother, straighter road with more passing opportunities is often found to be economically warranted based on the value of the time, vehicle operating cost, and crash cost savings to road users. However, these improvements also add to capacity, keeping any congestion price to practically zero (Walters 1968).

For major urban roads, economies of scale are offset by diseconomies of scale. For a network in a given area, the number of intersections increases faster than the number of lane-kilometres of roads. Intersections are land-intensive and often require traffic signals or grade separation (Hau 1992).

There are enormous economies of scale in strengthening road pavements (Harvey 1999). For flexible pavements, the rule-of-thumb is that a 10 per cent increase in pavement thickness results in a doubling of the traffic loading required to produce a given amount of wear. A compounding factor is that a large component of pavement deterioration is due to weather. It is the same regardless of vehicle usage.

Lumpiness in investment is another factor that can inhibit cost recovery with efficient prices. When capacity can be changed only in discrete jumps, long-run marginal cost cannot be finely adjusted to equal short-run marginal cost. The basic two-lane road with shoulders and drainage ditches provides ample capacity for most non-urban roads and so becomes a fixed cost (Hau 1992; Productivity Commission 2006).

In summary, short-run marginal cost pricing on congested urban roads is expected to yield sufficient revenues to cover the full costs of the roads concerned, and may generate additional revenue. However, for most of the road network, short-run marginal cost pricing will lead to major under-recovery of costs.

This means that imposing an additional toll to recover capital costs on each trip, where there is significant redundant capacity, would waste the existing resource. For these parts of the network, there are two main models to recover the capital costs of the road — cost recovery from road users, or financing from general tax revenue.

If the fixed costs are to be recovered from road users, this can be done through a tax. This can be enforced by excluding cars that do not pay, for example through motor vehicle registration charges (a fixed charge) or by imposing a tax related to distance travelled (a variable charge).

Finding

The road system as a whole has historically been excludable on the basis of motor vehicle registration requirements. In the future, specific roads or road systems may also be excludable using new technology.

Charges designed only to encourage the most economically efficient use of roads would not recover their full costs. If governments intend to recover the cost of building, operating and maintaining roads from road users, it would be necessary to impose a combination of additional fixed or variable charges above short-run marginal cost. The efficiency costs of specific cost recovery taxes or charges should be weighed against the efficiency cost of raising revenue from general taxation.

Arguments for and against cost recovery for roads

Market test

If road users do not pay the full cost of a road or network of roads, there is no direct feedback from the market about whether they are willing to pay for the infrastructure and hence whether the road or network should be provided at all. It is extremely difficult to make accurate estimates of willingness to pay because it requires knowledge of consumer demand at price levels that have not been observed.

If total revenue exceeds total costs, we can conclude that users' total valuation of the road exceeds total costs. The converse, however, is not necessarily true. In practice, a user charge cannot be designed that perfectly mirrors road users' entire willingness to pay.

Often, the important question is not whether the road should exist, but at what standard it should be maintained. Each improvement in road standard provides existing road users with savings in time, vehicle operating and crash costs, as well as generating new users.

Cost–benefit analysis techniques value these savings and compare them with the costs of building, maintaining or upgrading a road. The revenue impacts of the road upgrade, whether revenue from network access charges or from variable charges, provide little or no useful information. An upgrade to a single road would have negligible impact on revenue from network access charges and any increase in revenue from variable charges would relate to generated traffic only. It would not reflect the gains to existing users.

Relative economic efficiency costs

Charges above short-run marginal costs, whether access charges or variable charges, impose economic efficiency costs. The alternative of funding road infrastructure from general tax revenue also has economic efficiency costs. In principle, the two costs can be compared to determine which policy, or policy mix, is preferable in terms of efficiency.

Raising the necessary revenue from an increase in general revenue-raising taxes may have a lower efficiency cost than a variable road user charge, depending on how it operates. The economic efficiency costs of an access charging system depend on how well it discriminates between road users whose willingness to pay differs, so as to have minimal impact on people's decisions.

Financial

In some cases, funding constraints may prevent road agencies from undertaking economically warranted investment and maintenance. In these cases, greater certainty about future funding levels could improve the capacity of road agencies to plan for the future. An allocated funding stream might overcome some of this uncertainty.

The downside of allocating revenues from road user charges to road providers is that providers face few natural incentives to control costs they can pass on to users. Processes would need to be put in place to ensure accountability. These could include transparent and consistent cost–benefit analyses and independent post-build project evaluations to ensure that road users are charged only for costs that have been efficiently incurred.

A key issue is that Australia's roads belong to many different jurisdictional owners. There is little or no link between road revenues and the road owners. The road owners do not receive the economic rewards from road investment. As a result, road investment is largely determined by the competition for the use of tax revenues rather than efficiency criteria. A key issue for future consideration is whether there would be benefits in linking road revenues to road providers, and on what basis.

Uneconomic road spending (for example, investment projects with benefit–cost ratios below one) that is undertaken for broader social purposes should be transparently funded by government through explicit community service obligations.

Principle

Road investment and maintenance decisions that are taken for reasons of social policy, and are shown by cost–benefit analysis to be uneconomic, should be transparently identified as community service obligations and funded from general tax revenue.

Equity

Cost recovery is consistent with the user-pays concept of equity, which aims to ensure that those who receive the benefits from a government-provided service also pay for it. However, in the case of roads it is unclear that the group receiving benefits from road use can easily be distinguished from the entire Australian population. For example, of the 7.1 million dwellings that participated in the 2006 Australian Census (ABS 2006b), only 9.5 per cent did not have a motor vehicle.

On the other hand, there may be greater differences between the population of road users and, say, income tax payers. There may also be large differences between the users of high-cost roads and general taxpayers, or differences between road users in different jurisdictions.

Given the long life of road investments, there is also a question of intergenerational equity. Current road users obtain the benefit of past investment. They may reap only a small part of the benefits of current investment in roads, while people who no longer use the roads directly may nevertheless have paid in the past for investment that benefits road users today. If the road network was a constant size and required stable ongoing investment, these factors would cancel each other out. However, if investment in road networks is directed to meet anticipated future needs, then debt, to be repaid by future generations, might be a more equitable source of finance than charges imposed on today's users. However, there may be other macro-economic reasons for limiting desirable debt financing by governments.

The user-pays concept of equity ceases to apply where redistribution of income to particular groups is considered desirable or where there are 'merit good' considerations. Stanley and Starkie (1983) argue that the basic access characteristic of rural local roads is regarded as a 'merit good'. This goes some way to explaining why spending on some rural roads exceeds what might be justified under strict economic criteria.

The user-pays concept of equity suggests that much of the cost of local roads used for access to properties could continue to be paid for out of local government revenue from rates. Even households who do not own a car value well-maintained paved roads linking their property with the rest of the road network. Where heavy vehicles traverse local roads, their road-wear charges should be directed to the relevant local government as compensation.

Finding

There are arguments for and against recovering the total costs of the road system from road users. The social opportunity cost of the existing network is in general not subject to charging. Existing users could be charged explicitly for operating and maintenance costs, and for network improvement and expansion. The efficiency loss from raising the required revenue from income or other taxes must be compared with the efficiency loss from the most efficient, practical system of access and variable charges. The full information required to make all these assessments is not presently available.

Cost recovery through network access charges

Existing car registration charges can be thought of as a fixed network access fee, as the annual charge gives access to the entire road network, apart from toll roads. In 2006–07, expenditure on roads by the Australian and State and Territory governments (excluding spending by local governments) was almost $9 billion (BITRE 2009). Excluding an estimated $3 billion for road damage costs, which should be recovered directly through road-wear charges, these costs could be recovered by an average charge of around $500 per vehicle per year.

While these charges have the advantage of not influencing choices about how much and where to drive a vehicle that is already registered, the existence of a fixed charge may discourage some people from owning a car. This is likely to be low-income or low-wealth people and those who travel relatively few kilometres per year.

This effect can be reduced by varying the access charges between groups of users with different willingness to pay. For example, existing car registration charges are discounted for many groups, including pensioners. In some cases, this may reflect a lower than average willingness (or ability) to pay for access to the road network. However, it is doubtful that registration charges that increase with the number of the car's cylinders are closely correlated with the car owners' willingness to pay.

The distance people travel per period of time is related to willingness to pay. People who drive few kilometres per year are more likely to be deterred from owning a car by a flat access charge than people who drive long distances. The most efficient charge for cost-recovery purposes may be a combination of a distance-related charge and an access charge (two-part pricing).

Some variable charging might also be justified on the basis of recovering otherwise unpriced variable costs. The existing fuel tax is an example of a tax that varies with distance travelled. Subject to technology, direct charges based on distance travelled would be an alternative.

The future of fuel tax

Consistent with the principle that transport-specific taxes should be imposed only where they improve social or market outcomes in transport markets, fuel tax as a source of general government revenue should be phased out. However, some fuel tax might be retained as a simple variable charge for variable costs of the road network that cannot be priced directly.

The tax rate would be set by reference to costs efficiently incurred. To ensure its value is not eroded by inflation, the per litre tax rate would be indexed to the consumer price index. Fuel tax credits for off-road use would be retained, to ensure that only on-road use is subject to a charge.

To be an efficient user charge, fuel tax would need to apply to all energy sources used for road transport. This means extending tax to those fuels that are effectively tax-free under current arrangements. These include liquefied petroleum gas, liquefied natural gas, compressed natural gas, biodiesel and domestically produced ethanol. If alternative energy sources for road transport were developed (such as electricity), they would also need to be taxed in their on-road applications. As the energy density of these fuels varies, it would be necessary to tax fuel on an energy-content basis as this is more closely related to vehicle distance travelled (see Recommendation 65).

Different greenhouse emission costs associated with different fuels could best be addressed by price differentials under the Carbon Pollution Reduction Scheme rather than different rates of fuel tax. Similarly, policy relating to energy security or fossil fuel depletion should be more appropriately addressed through specific targeted measures, rather than fuel tax.

Variable charging through fuel tax would not be necessary in cases where technology provides ways to measure usage more directly. For example, as road user charging mechanisms based on mass-distance-location monitoring technology becomes widespread in heavy vehicles, it would be possible to charge heavy vehicles on this basis, and provide full exemption from fuel taxes and the network access component of registration charges (see Recommendation 65).

Other road taxes and charges should be phased out

Governments in Australia impose motor vehicle registration and transfer charges. Some charges relate to the costs of providing government services. In these cases, the charges are likely to have the potential to improve efficient allocation of resources, and should be maintained. However, other taxes impose charges that are disproportionate to any costs that the government incurs in providing the service. Reliance on revenue from these taxes should be phased out, and replaced with more efficient sources of revenue (see Recommendation 66).

COAG has asked the Review to consider the merit of financial incentives for the purchase of fuel-efficient cars and assess the merits of differential stamp duty and registration regimes linked to environmental performance. The use of these taxes for environmental, rather than revenue purposes, is discussed in Section E2 Taxes to improve the environment.

Stamp duty on cars is a highly inefficient source of revenue

State governments do not have the legal power to impose duties of excise, but do tax certain transactions related to goods. In the case of transport, State governments tax motor vehicles by imposing an additional stamp duty whenever a new car is registered, or by requiring buyers of second-hand cars to pay a registration charge on transfer between owners (see Chart E3–3). As stamp duty on housing leads to a misallocation of the housing stock (see Section C2 Land tax and conveyance stamp duty), so vehicle stamp duties impede efficient allocation of vehicles.

Chart E3–3: Stamp duty payable on purchase of new $40,000 car(a)

Chart E3–3: Stamp duty payable on purchase of new $40,000 car(a)

  1. Assumes car is V6 (QLD) and rated between 3.5–4.5 on Green Vehicle Guide (ACT).

Source: Treasury estimates.

Rates of stamp duty vary between States, based on car value and, in some cases, vehicle size or type. Some States apply different rates to passenger and non-passenger vehicles, and some apply tax rates that rise with the value of the vehicle (Australian Government 2008b, p. 86).

These taxes mean that people purchase new vehicles and scrap old vehicles less often, and reduce the overall demand for cars.11 They mean that some people will continue driving vehicles not suited to their present needs. For example, an older couple whose children have left home might delay getting a smaller car. Alternatively, a young couple may delay upgrading to larger family car when they have children, because of the additional cost.

Finding

Stamp duty on the transfer of motor vehicles is a highly inefficient revenue source.

Restrictions on the number of taxi licences should be removed

Taxis form a small but important component of the transport system, providing flexible mobility when other forms of transport are not viable. But taxi fares are up to 25 per cent higher than they need be because of State government taxes (Productivity Commission 1999, p. 16).12 These taxes are imposed in a relatively unusual way. The States limit the number of taxis then make money by selling licences.

These restrictions are beyond those necessary to maintain safety or service standards and are simply used to raise revenue. For these reasons, accounting standards and economic reasoning recognise the revenue as tax revenue. Taxi licences now sell on secondary markets for up to $470,000.13 The purchaser expects that future restrictions on the number of taxis will allow this money to be recouped from future taxi customers — with interest.

The total value of taxi licences in Australia is around $6.5 billion (Clarke & Prentice 2009). Because of the tax, taxi services are more costly and waiting times are longer. This has a number of adverse impacts on Australian society. Businesses pay the taxi tax, or they must use alternative transport less fit for purpose, driving the cost of goods and services up for all Australians.

The taxi tax reduces the ability of Australians to effectively 'share' a motor vehicle by taking taxi trips instead of owning their own car. This means the tax falls disproportionately on those who do not drive (perhaps because of cost), should not drive (due to inebriation) or cannot drive (because of disability). The poorest 20 per cent of the community spends more than twice as much as a proportion of income on taxis as other Australians (ABS 2006c). In regional areas, taxis often replace other urban transport systems, such as rail or bus services that are not available. Taxis are therefore one important means by which people continue to participate in society. The tax impacts on some of the most vulnerable in society, either by reducing their incomes or dissuading them from taking a taxi.

Why does such a bad tax persist? Abolishing the tax would mean retaining only those restrictions on taxi licences that relate to safety and service. This would see the value of plates fall nearly to zero. Existing plate holders — who are often not the actual drivers — would lose all the value of their investment. By selling the plates for revenue up front, governments have effectively created a group of people with an interest in maintaining the asset value of taxi plates so they can recoup their investment.

Some of the financial return plate holders earn reflects the risk that current arrangements may change, so whether they should be compensated at all is an open question. There is no doubt, however, that it would be far better for society to cash plate holders out using revenue from other taxes rather than to retain the highly inefficient taxi tax.

Finding

Quantity restrictions on taxi licences are an implicit tax on taxi users, from which additional revenue flows to existing taxi plate holders and State governments.

Quantity restrictions on taxis should be phased out (see Recommendation 66). This could be done by imposing a price ceiling on the price of new taxi licence plates that reduces over time, triggering the automatic release of new plates. Alternatively, existing taxi licences could be bought back by the government. While for efficiency reasons this should funded from general tax revenue, for equity reasons it could be done by replacing the existing implicit tax on taxi fares with an explicit levy on taxi users. This would be imposed for a limited time until the revenue needed to fund the transition has been raised.


11 Tcha and Kuriyama (2003) estimate an own-price elasticity of demand for cars of -0.43, suggesting that a tax that increases the price of a car by 3 per cent could reduce demand for cars by around 1.3 per cent.

12 More recently, IPART (2009, p. 21) found that in NSW 'Plate lease costs contribute 13 per cent of the overall cost of providing taxi services in urban areas, and around 11 per cent of this overall cost in country areas.'

13 On the BSX Taxi Market, plate prices averaged $430,000 in 2008–09.