Climate Change Authority

You are here

Modelling and approach to Costs and benefits of standards

B.1 Introduction

This appendix outlines the Authority’s approach to assessing the benefits and costs of implementing fleet-average CO2 emission standards for light vehicles in Australia. It includes an overview of the modelling commissioned from the CSIRO to investigate the impacts of standards. The results are used in Chapter 4.

The Authority has conducted an indicative assessment of the net private and net social benefits of these standards using a combination of commissioned modelling and additional economic evidence. This work aims to provide a good starting point for a full cost-benefit analysis required for any regulatory impact statement.

The Authority has not developed an Australia-specific estimate of the incremental cost of different standards. Instead, the cost estimate is based on international studies of the costs of fuel-saving technologies necessary to meet similar standards. These international estimates isolate the incremental costs of fuel-saving technologies from other vehicle features that contribute to driver utility. The estimates of fuel savings and emissions reductions from the standards have been calculated directly by the Authority using the modelling discussed here. These estimates necessarily involve making assumptions about a range of inputs, informed by the available evidence. Where a clear central estimate is not available, the Authority has attempted to err on the side of choices that would underestimate the benefits available from standards.

This appendix outlines:

  • the modelling commissioned to analyse standards, and the BAU and standards scenarios analysed (B.2)
  • details of the approach to estimating fuel savings and the impact on vehicle costs (B.3)
  • details of the approach to estimating the cost of emissions reductions from standards (B.4).

B.2 CSIRO modelling and scenarios

The Authority commissioned modelling from the CSIRO (Reedman and Graham 2013b) to explore the potential benefits of standards; in particular, fuel savings and emissions reductions. The starting point for the modelling is the BAU projection describing what would happen in the absence of standards. Six different standards scenarios are then modelled—a lenient, medium and strong standard starting in either 2018 or 2025. The results are compared with BAU to identify the benefit of each standard.

B.2.1 CSIRO modelling of light vehicle emissions standards

Reedman and Graham use the CSIRO’s Energy Sector Model (ESM) to investigate the impacts of standards. The ESM (see Box B.1) determines the least-cost fuel and vehicle mix to meet a given transport demand, subject to constraints such as policy, vehicle class preferences and vehicle stock turnover.

The analysis of standards forms part of a larger transport emissions projections exercise (Reedman and Graham 2013a) conducted for the Authority’s Targets and Progress Review, the assumptions for which were subject to public consultation at the start of 2013.

When interpreting the modelling results, the Authority has taken into account the difference between actual and ‘tested’ new light vehicle emissions intensity, which make the CSIRO projections for new light vehicle emissions intensity appear higher than in some other sources.

The CSIRO model estimates actual new light vehicle emissions intensity, calculated from public data on fuel consumption, vehicle sales and emissions intensity of fuels (Graham 2014). In contrast, ‘test cycle’ readings of new light vehicle emissions intensity, such as those published by the National Transport Commission, are the result of laboratory testing of new vehicles’ emissions intensity. The CSIRO’s analysis of the difference between its estimates and those measured by the test cycle over the last decade indicates the CSIRO’s estimates are, on average, about 5 per cent higher than test cycle intensity.

Because a mandatory standard would set a target level for test cycle rather than actual emissions intensity, the Authority has incorporated this adjustment when making findings on the level of a light vehicle emissions standard for Australia (Chapter 4). Throughout the report, the unadjusted CSIRO estimates of actual new light vehicle emissions intensity are described as ‘measured,’ and projections for new light vehicle emissions intensity intended to correspond to the results of vehicle testing are described as ‘tested’.

In this analysis, the Authority focuses on the projected impacts of standards to 2030. The CSIRO modelling contains projections of the impacts of standards commencing in 2018 and 2025 over the period to 2050. Over longer time horizons, it is increasingly difficult to project what technologies might exist, their rate of deployment in new vehicles, the relative cost and emissions performance of those technologies, and fuel prices that may influence fuel consumption and emissions outcomes.

For a full account of the modelling, see Reedman and Graham (2013b).

Box B.1: The CSIRO’s Energy Sector Model

To model the emissions reduction potential of light vehicle emissions standards, the CSIRO uses its Energy Sector Model (ESM). The ESM assumes vehicle owners make the least-cost vehicle choices to meet a given transport task. Consumers are assumed to purchase alternative fuel or engine vehicle technology if the discounted payback from the fuel savings offsets any additional upfront costs within five years. Inputs include projected rates of improvement in the fuel efficiency of internal combustion engines and consumer preferences about vehicle sizes. Outputs include the fuels consumed (such as petrol, diesel and LPG, and their associated spark or compression ignition engines types), and the drivetrains chosen, including internal combustion engine, hybrid, electric and fuel cell drive. In addition to the cost of alternative fuels and vehicles, ESM incorporates detailed fuel and vehicle technical performance characterisations such as fuel efficiencies and emission factors by vehicle type, engine type and age.

For this exercise, demand for road transport in the BAU scenario was determined in the Monash Multi-Regional Forecasting Model, taking into account population growth, projected output of industries and changes in the cost structure of road transport.

Demand in the standards cases was determined in the ESM by allowing changes in the overall cost of travel due to standards to affect the level of travel demand (that is, by incorporating a ‘rebound effect’). The value of the rebound in the ESM is 0.2, meaning that there is a 0.2 per cent increase in demand for every 1 per cent fall in the overall cost of travel. Estimates of the value of the rebound effect in road transport vary; the value used in the ESM is broadly equivalent to the mean of international estimates (NHTSA 2012, p. 853).

Fuel prices are the same across the BAU and standards scenarios. Australian retail prices are projected by applying a method for translating oil and gas paths into retail fuel prices, which includes assumptions about future excise rates by fuel (see Reedman and Graham 2013a, pp. 28–31). The oil price path is based on the IEA’s 2012 World Energy Outlook, which grows in real terms by 61 per cent over 2013–30 (Treasury and DIICCSRTE 2013, p. 59). Consistent with this outlook for oil prices, retail petrol prices are projected to increase by 24 per cent in real terms over the same period (Reedman and Graham 2013a, p. 30), taking into account the outlook for other components of the retail price, including fuel excise. The excise rates do not reflect the increases announced in the 2014–15 Budget. With higher real excise, fuel prices would be higher in all scenarios, and the fuel savings for consumers from standards would most likely be larger than the estimates provided here.

The ESM assumes a linear change in fuel consumption in response to changes in activity. There is an assumption that average activity per vehicle plateaus after 2030, after which demand for passenger transport grows in response to population growth. This approach implicitly accounts for a typical vehicle in Australian traffic conditions over time.

As with all models, there are limitations, including to assumptions for parameters that are in reality uncertain and in some cases evolving rapidly (for example, advanced biofuels and the cost and driving range of future electric vehicles). As the ESM considers cost as the only driver of consumer choice, it cannot capture behaviour driven by other factors. This could result in either underestimating or overestimating rates of adoption of some new technologies or fuels, depending on whether these non-price factors encourage or discourage adoption.

The way in which standards are introduced in the ESM is described in B.3.3. Further information on the ESM is provided in Reedman and Graham (2013a).

B.2.2 Assumptions about BAU reductions in emissions intensity

Projected emissions under BAU depend on the rate of improvement in new light vehicle fuel efficiency that would occur without standards. As discussed in Chapter 3, two factors complicate the projection of BAU improvements in new Australian light vehicles:

  • Recent rates of improvement have been rapid relative to Australia’s earlier history, but it is unclear whether these rates will be sustained.
  • Mandatory vehicle emissions standards in other countries will become increasingly ambitious over the period to 2025; this will likely make new light Australian vehicles more efficient but the extent of this influence is unclear.

The modelling assumes the BAU rate of reduction in the average emissions intensity of new light vehicles slows from the 2.8 per cent per year observed over the last eight years to 2.0 per cent per year over the period to 2020, and reduces further to 1.6 per cent per year to 2025. This results in average measured emissions intensity of approximately 197 g CO2/km in 2015, 178 g CO2/km in 2020 and 164 g CO2/km in 2025 (Graham 2014).

This projected annual improvement rate is similar to other recent estimates of BAU.

  • In 2010, the FCAI commissioned estimates of the BAU of the light vehicle fleet (PWC 2010). The work was based on confidential consultations with vehicle manufacturers operating in Australia to assess both the rate of technology uptake and consumer preferences. It projected that the change in average light vehicle CO2 emissions intensity would slow from the average of 2.1 per cent (4.3 g CO2/km per year) achieved from 2002–10 to about 1.9 per cent per year (2.3 g CO2/km per year) from 2010–20. The average tested emissions intensity of the new light vehicle fleet was projected to be about 195 g CO2/km in 2015 and 176 g CO2/km in 2020 under BAU.
  • More recently, ClimateWorks’s 2014 Briefing Paper Improving Australia’s Light Vehicle Fuel Efficiency drew on unpublished analysis by Rare Consulting. This used the same path for improvement as the FCAI/PWC analysis, and extended its projection to 2024. With the inclusion of 2011 data, a slightly higher historical rate of improvement of 2.2 per cent was assumed by Rare, and it projected this would slow to about 1.8 per cent per year from 2011–24. The average tested emissions intensity of the new light vehicle fleet was projected to be about 175 g CO2/km in 2020 and 165 g CO2/km in 2024 under BAU.
  • In its 2011 discussion paper on light vehicle standards, the Department of Infrastructure and Transport proposed a BAU annual improvement of 2.1 per cent or 2.5 per cent over the period to 2015 as a basis for analysing the effects of standards starting in that year.

Table B.1 and Figure B.1 compare the CSIRO, PWC and ClimateWorks projections. The rate of emissions intensity improvement in the CSIRO modelling is similar to the other sources, and all projections are slower than the rate achieved in the past decade.

In this analysis, the Authority has estimated the benefits of standards relative to the BAU rates of reduction in the CSIRO modelling. If BAU rates of improvement are faster than 2 per cent per year, the modelling will overestimate emissions and fuel savings from standards, but will also overstate the effort necessary to achieve any given standard. If BAU rates of improvement are slower than projected, the opposite will be true.

Table B.1: Comparison of projected rates of light vehicle emissions intensity improvement, three BAU scenarios

Source (year) Annualised rate of change Time period Rationale
PWC (2010) –1.9% 2010–20 Industry consultation on technology uptake and consumer preferences
ClimateWorks (2014) –1.8% 2011–24 Builds on the PWC estimate with updated 2011 data and extended to 2024
CSIRO (2013) –1.8% 2013–25 Driven by projected improvements in petrol internal combustion engines; some projected changes in preferences

Source: Climate Change Authority based on sources listed in table

Figure B.1: Historical and projected rates of improvement in light vehicle emissions intensity, three BAU scenarios

Figure B.1 is a line chart showing the emissions intensity of Australia’s light vehicles over time, with historical emissions from 2002 and projections to 2025. Projections are shown for three business-as-usual scenarios from the CSIRO, PWC and ClimateWorks. All three show a similar rate of emissions intensity improvement. However the CSIRO modelling starts at actual 2012 emissions, which are lower than projected emissions in the PWC and ClimateWorks scenarios. As a result, CSIRO projections remain at a lower level over the period to 2025. All projections assume slower improvement than the rate achieved in the past decade.

Note: CSIRO-projected BAU levels are converted to test cycle from measured emissions; other sources project test cycle emissions intensity. See Section B.2.1 for further discussion.
Source: Climate Change Authority (from sources listed in legend)

 

B.2.3 Standards modelled for the Authority

The CSIRO modelled a total of six standards scenarios with three different stringencies—lenient, medium and strong—and two different start years—2018 and 2025 (Reedman and Graham 2013b). These standard scenarios are implemented in the ESM by:

  1. Imposing an additional, fixed amount of improvement in the efficiency of petrol internal combustion engine efficiency (3.3 per cent a year, up from 1.3 per cent under BAU). This is assumed to be available with no additional upfront cost to vehicles. Reedman and Graham (2013b, p. 6) draw on analysis in the 2007 King Review for their assumption that there is a set of fuel-saving changes available to the mass market in the range of $150 to $1,000 for new vehicles using internal combustion engines. Assuming these lower cost fuel savings innovations are introduced in a gradual manner and as a priority over other product features, they conclude that real vehicle prices are not likely to be significantly changed.
  2. Allowing the most cost-effective deployment of alternative drivetrains and use of diesel vehicles to achieve the remainder of the emissions standard. To meet the required standards, the model ensures consumers adopt vehicles, even if the payback period is longer than the five years typically specified in the model as the basis for consumer choice.

The modelling assumes that standards have no effect on consumer preferences for vehicle size. Under both BAU and standards scenarios, smaller vehicles increase their share of the passenger vehicle market at the expense of larger and, to a lesser extent, medium vehicles (Reedman and Graham 2013b, p. 5).

Table B.2 shows average annual light vehicle emissions intensity associated with the BAU scenario and standards starting in 2018, along with the approximate corresponding test cycle level of emissions intensity.

All standards modelled by the CSIRO are illustrated in Figure B.2. The figure shows that all of the standards assume sustained improvement until the average measured emissions intensity of new light vehicles reaches 100 g CO2/km, after which no further reductions in emissions intensity occur. While it has no practical impact on the modelling, this may be a conservative limit—it is equivalent to a tested target of around 95 g CO2/km, which is the 2020 EU target for passenger vehicles. While it would be more difficult to meet this target for all light vehicles (rather than just passenger vehicles), the EU is considering a 2025 passenger vehicle target of between 68 and 78 g CO2/km (ICCT 2013c), suggesting average new light vehicle limits below 100 g CO2/km are feasible.

Table B.2: Standards modelled starting in 2018—average measured (and approximate test cycle) emissions intensity levels, new light vehicles, selected years

Scenario 2018 2020 2025
BAU (2 per cent 2013–20; 1.6 per cent 2021–25) 185 (176) 178 (169) 164 (156)
Lenient (3.5 per cent from 2018) 182 (174) 170 (162) 142 (135)
Medium (5 per cent from 2018) 179 (171) 162 (154) 125 (119)
Strong (6.5 per cent from 2018) 177 (168) 154 (147) 110 (105)

 

Note: Measured new light vehicle emissions intensities are estimated to be about 5 per cent higher than test cycle emissions intensities. See Section B.2.1 for further details.
Source: Reedman and Graham 2013b

Figure B.2: Standards modelled—average measured emissions intensity levels from new light vehicles, standards starting in 2018 or 2025

Figure B.2 is a line chart showing emissions intensity levels between 2013 and 2030 for seven scenarios: business-as-usual (no standard); lenient, medium and strong standards starting in 2018; and lenient, medium and strong standards starting in 2025. Lenient standards improve by 3.5 per cent per year compared to BAU; medium standards improve 5 per cent per year, and strong standards improve 6.5 per cent per year. Improvements continue until emissions intensity reaches 100 g CO2/km, when it plateaus.

Note: This graph differs from Figure 2.2 in Reedman and Graham 2013b; this version corrects the BAU rate of new light vehicle emissions intensity and levels for the standards cases. This figure shows measured new light vehicle emissions intensities that are estimated to be about 5 per cent higher than test cycle emissions intensities. See Section B.2.1 for further details.
Source: Climate Change Authority based on Reedman and Graham 2013b and Graham 2014

 

B.3 Estimating the net impacts of standards

B.3.1 Fuel savings from light vehicle emissions standards

The estimates of fuel savings in Chapter 4 for each standards scenario are calculated as follows:

  1. Determine total fuel savings each year by calculating the difference between the modelled total fuel spend under standards and BAU.
  2. Calculate the amount of total fuel savings in (1) that come from new vehicles of each model year subject to standards, by subtracting total fuel savings in the current year from the previous year. Note that if vehicles have an average life of 15 years in the stock, the first vehicles subject to standards would exit the fleet in 2033 on average, which is beyond the end of the first phase of standards. This means that it is acceptable to ascribe all of the annual change in fuel savings to the new vehicles subject to standards that entered the fleet that year, rather than to a combination of entry and exit.
  3. Calculate the average annual savings from the first year of ownership for vehicles from each model year by dividing the results in (2) by the number of new light vehicles purchased in each year.
  4. Calculate the present value of fuel savings for the first owner and the vehicle’s life for a vehicle bought in each year under standards, by:
    1. Growing average annual savings in (3) by the rate of real fuel price growth in each scenario (to adjust the fuel savings for rising real fuel prices over time).
    2. Taking the present value by discounting the stream of annual fuel savings in (a), and summing the discounted savings over three or five years (for the first vehicle owner) or 15 years (for the vehicle’s life).

For private net benefits, the calculations use the full retail fuel prices including excise; for social net benefits, these calculations exclude excise because this is a transfer at the economy-wide level (from consumers to government).

While vehicles may spend longer in the stock, the assumption of 15 years provides a conservative estimate of the fuel savings from standards; longer vehicle lives would mean higher fuel savings, if other things were equal. The discount rate of 7 per cent per year is the default discount rate for discounting private benefits in Commonwealth assessments of regulatory impacts (OBPR 2013).

The fuel spending in Reedman and Graham is in 2010 Australian dollars (Graham 2014). Along with all other monetary values in this report, fuel savings are in real 2012 Australian dollars unless indicated. The fuel savings were inflated to 2012 values using the RBA’s inflation calculator (inflation over the two years of 5.1 per cent (RBA 2014)).

B.3.2 Impact of standards on vehicle costs

The estimates of the incremental costs of US standards in Table 4.2 are the estimated incremental costs for passenger vehicles and light trucks (NHSTA 2012), weighted by the Authority to create an estimate of incremental costs for all light vehicles.

In the US, SUVs are classified as ‘light trucks’ (light commercial vehicles) while in Australia they are classified as passenger vehicles. Because the US incremental costs are used as an estimate of the incremental cost of meeting a similar standard in Australia, they are combined using weights that make some adjustment for this difference in classification—passenger vehicles receive a weight of 70 per cent, rather than their share of the Australian market according to Australian classifications (about 80 per cent). While SUVs make up about 30 per cent of the Australian market (Chapter 2), NHSTA estimates indicate costs for light trucks are lower than for passenger vehicles, so the Authority’s weights err on the side of overestimating the incremental costs.

The additional vehicle costs for the US and EU in Table 4.2 were converted to Australian dollars in two steps:

  1. Both the EU and US sources reported incremental costs in 2010 units of their respective currencies. These costs were converted to 2010 AUD using the average annual exchange rate for 2010 reported by the RBA: AUD$1=US$0.92 and AUD$1=€0.70.
  2. The converted figures were inflated to 2012 values using RBA 2014, and rounded to the nearest $10.

These US costs were then weighted as described above to generate the estimated incremental cost for all light vehicles. These costs are estimates of the incremental production costs associated with meeting standards and do not include smaller components of the overall cost of owning a new vehicle that might rise with higher vehicle purchase prices, such as insurance premiums.

As mentioned in Chapter 4, the increase in retail prices may be lower if not all of the cost of the increase in vehicle production costs resulting from production changes to meet the standard is passed through to consumers. Vehicle suppliers might absorb some of the increase over the short term to gain market share, or over the longer term if competition in vehicle markets was imperfect and suppliers could ‘price to market’ by adjusting vehicle prices in separate geographical markets to maximise overall profits. The proportion passed through to consumers in Australia would depend on a range of factors, including competition in the market and the extent to which a rise in vehicle prices will affect consumers’ purchasing decisions. In this context, it is worth noting that by international standards, the Australian new vehicle market offers a large number of models to consumers, increasing competitive pressure on suppliers. The RBA’s analysis of cost pass-through following changes in the exchange rate provides some evidence of pricing to market for Australian imports as a whole. While the proposition of full pass-through was not always rejected in statistical tests, the analysis suggests that exchange rate changes are passed through rapidly and, to a large but incomplete extent, into import prices. An estimated 80 per cent of a change in exchange rates is passed through, with the total effect occurring within one quarter (Cheung et al. 2011, pp. 10–11).

B.4 Emissions reductions from standards

B.4.1 Emissions reductions from 2018–25 standards

The cumulative emissions reductions in Figure 4.8 are for reductions from vehicles subject to the first phase of standards proposed by the Authority (2018–25). They are reported in carbon dioxide equivalent (Mt CO2-e) and include CO2, methane and nitrous oxide emissions. The CSIRO modelling assumes that standards continue past 2025 (see Figure B.2). The Authority has therefore calculated cumulative emissions reductions to 2030 from the first phase of standards by summing the cumulative emissions reductions from standards over 2018–25 and the average annual emissions reductions that vehicles from those model years would deliver over the period 2026–30.

B.4.2 Approach to estimating the value of emissions reductions

The cost per tonne of emissions reductions from standards discussed in Chapter 4 were determined using the Authority’s general approach to calculating the cost per tonne of emissions reductions—dividing the net present value of the incremental resource cost by the stream of resulting emissions reductions. In the case of vehicle standards this becomes:

Cost per tonne of emissions reductions for each model year ($/t) =

net present value of incremental costs from standards for model year ($) / stream of (undiscounted) emissions reductions from vehicles of model year over their life (t)

The net present value of the incremental costs from standards are equal to the incremental capital costs minus the present value of the fuel savings (excluding excise); these are taken from international evidence and the Authority’s calculations as described in B.3.1 and B.3.2, respectively. The incremental costs and fuel savings per vehicle calculated above are multiplied by the number of vehicles sold in each model year to obtain economy-wide costs for each year per model year.

The stream of undiscounted emissions reductions are Authority calculations from the CSIRO modelling. These are the product of:

  • the difference in new light vehicle emissions intensity between standards and BAU for each model year (g CO2/km)
  • the weighted average distance travelled per vehicle per year (vehicle kilometres per year)
  • vehicle life (assumed to be 15 years)
  • the number of vehicles sold each year.

The resulting estimate of –$580 per tonne of avoided emissions is the average of the cost per tonne over the model years 2020–25. Model years 2018 and 2019 are excluded from this average because the incremental capital costs are sourced from the US, and the Authority’s strong standard starting in 2018 is most similar to the US standard from 2020 onwards (see Figure 4.1.)

The Authority’s approach is conceptually similar to that of ClimateWorks in its cost curve analysis, with the difference that ClimateWorks looks at a particular year rather than computing net present values of the stream of costs and benefits. Its estimate of a –$350 per tonne private cost provides a ‘snapshot’ of the cost of emissions reductions in 2020 by dividing the net cost in 2020 by the emissions reductions in 2020 (ClimateWorks 2014).

There are some published estimates of higher positive costs of emissions reductions from standards. These are generally not estimates of the cost-effectiveness for society as a whole. For example, Frondel, Schmidt and Vance (2008, pp .8–9; cited by FCAI 2011c, p. 14) calculate the cost per tonne of emissions reductions from the EU standard as €100 to €200 per tonne for the standards to 2015, and €475 to €900 per tonne after 2015. The approach attempts to calculate the cost-effectiveness of standards for society as a whole from the cost per tonne to liable parties for non-compliance. In fact, the result is neither an upper bound on the compliance cost per tonne for liable parties, nor an estimate of the net benefit per tonne to society as a whole. It is therefore not informative about the potential costs of emissions reductions from vehicle standards in Australia.