Glossary - Department of Industry, Tourism and …



report What existing economic studies say about Australia’s cost of abatementAnd what are the most fruitful areas for future modelling?Prepared forDepartment of the Environment and Energy31 July 2019The Centre for International Economics.aurighttop00The Centre for International Economics is a private economic research agency that provides professional, independent and timely analysis of international and domestic events and policies.The CIE’s professional staff arrange, undertake and publish commissioned economic research and analysis for industry, corporations, governments, international agencies and?individuals.? Centre for International Economics SAVEDATE \@ "yyyy" \* MERGEFORMAT 2020This work is copyright. Individuals, agencies and corporations wishing to reproduce this?material should contact the Centre for International Economics at one of the following addresses.CanberraCentre for International Economics Ground Floor, 11 Lancaster Place Majura ParkCanberra?ACT?2609 GPO Box 2203 Canberra?ACT?Australia?2601Telephone+61 2 6245 7800 Facsimile +61 2 6245 7888 Emailcie@.auWebsite.auSydneyCentre for International Economics Level 7, 8 Spring Street Sydney?NSW?2000Telephone+61 2 9250 0800 Emailciesyd@.auWebsite.auDisclaimerWhile the CIE endeavours to provide reliable analysis and believes the material it?presents is accurate, it will not be liable for any party acting on such information.Contents TOC \t "Heading 1,1,Heading 1 NotNumbered,1,Heading 2,2,Heading 6,1,Part Title,4" Glossary PAGEREF _Toc15473386 \h 1Summary PAGEREF _Toc15473387 \h 21Introduction PAGEREF _Toc15473388 \h 92Models and key characteristics PAGEREF _Toc15473389 \h 12Summary of model types PAGEREF _Toc15473390 \h 183Overview of key model outcomes PAGEREF _Toc15473391 \h 19Modelling strategies PAGEREF _Toc15473392 \h 19Key model mechanisms PAGEREF _Toc15473393 \h 19Abatement and the implied carbon price PAGEREF _Toc15473394 \h 22Macroeconomic impacts PAGEREF _Toc15473395 \h 27Comparison with international outcomes PAGEREF _Toc15473396 \h 29Botttom-up estimates of marginal abatement cost (MAC) curves PAGEREF _Toc15473397 \h 30The importance of international abatement PAGEREF _Toc15473398 \h 34The cost of international action to Australia PAGEREF _Toc15473399 \h 36Structural change and adjustment PAGEREF _Toc15473400 \h 37Relative merits of different policy instruments PAGEREF _Toc15473401 \h 45Composition and change in the energy market PAGEREF _Toc15473402 \h 46Energy price changes PAGEREF _Toc15473403 \h 494Sources of model input information PAGEREF _Toc15473404 \h 50Data for underlying baseline construction PAGEREF _Toc15473405 \h 50Data for simulations PAGEREF _Toc15473406 \h 51Key parameter choices PAGEREF _Toc15473407 \h 52Dealing with uncertainty PAGEREF _Toc15473408 \h 525Dependence of results on underlying assumptions PAGEREF _Toc15473409 \h 53Price pathway sensitivities PAGEREF _Toc15473410 \h 53International abatement cost sensitivities PAGEREF _Toc15473411 \h 54Technology sensitivities PAGEREF _Toc15473412 \h 59Evolution of renewable costs PAGEREF _Toc15473413 \h 61Integration costs PAGEREF _Toc15473414 \h 64Structural change sensitivities PAGEREF _Toc15473415 \h 68Potential use of Integrated Assessment Models PAGEREF _Toc15473416 \h 686Lessons for future modelling PAGEREF _Toc15473417 \h 71Lessons from modelling to date PAGEREF _Toc15473418 \h 71The value of economic models PAGEREF _Toc15473419 \h 72Recommended ‘style’ of modelling PAGEREF _Toc15473420 \h 73Areas for further analysis PAGEREF _Toc15473421 \h 74Specific modelling suggestions PAGEREF _Toc15473422 \h 75Boxes, charts and tables TOC \t "Caption" \c 1.1Overview of key studies covered PAGEREF _Toc15473423 \h 102.1Features of selected IAMs PAGEREF _Toc15473424 \h 172.2Summary of model types PAGEREF _Toc15473425 \h 183.1Carbon price and abatement outcomes for Australia Real $2010 PAGEREF _Toc15473426 \h 253.2Carbon price and abatement outcomes for Treasury, BAEconomics and G-Cubed PAGEREF _Toc15473427 \h 263.3Reduction in GNP relative to abatement PAGEREF _Toc15473428 \h 273.4GNP and abatement including G-Cubed and BAEconomics results PAGEREF _Toc15473429 \h 283.5GNP and abatement including two ANO scenarios PAGEREF _Toc15473430 \h 293.6Comparison with AR5 WGIII PAGEREF _Toc15473431 \h 293.7McKinsey Australian Abatement Cost Curve PAGEREF _Toc15473432 \h 303.8ClimateWorks Australia Abatement Cost Curve PAGEREF _Toc15473433 \h 313.9Energetics Australia Abatement Cost Curve PAGEREF _Toc15473434 \h 313.10Carbon price and abatement outcomes: CGE models versus bottom-up MAC curves PAGEREF _Toc15473435 \h 323.12Comparative cost of abatement: Australia versus world PAGEREF _Toc15473436 \h 353.13Domestic versus purchased abatement in Treasury modelling PAGEREF _Toc15473437 \h 363.14Structural change illustrated PAGEREF _Toc15473438 \h 373.15Structural change PAGEREF _Toc15473439 \h 393.16Declining Australian industries in 2050 relative to baseline PAGEREF _Toc15473440 \h 403.17Expanding Australian industries in 2050 relative to baseline PAGEREF _Toc15473441 \h 413.18Ratio of consumption emissions to production emissions by broad sector PAGEREF _Toc15473442 \h 423.21Compositional change in the electricity market PAGEREF _Toc15473443 \h 463.22Black coal shares and abatement PAGEREF _Toc15473444 \h 473.23Different pattern of renewables PAGEREF _Toc15473445 \h 483.24Electricity prices and abatement PAGEREF _Toc15473446 \h 495.2Australian marginal cost of abatement curve implied by 2013 Treasury analysis PAGEREF _Toc15473447 \h 555.4Effects on abatement prices removing sellers of abatement PAGEREF _Toc15473448 \h 585.5Technology sensitivities. Changes in deviation of GNP from baseline PAGEREF _Toc15473449 \h 595.6Global technology shares in electricity generation, the role of CCS PAGEREF _Toc15473450 \h 605.7Projected cost of wind: modelling studies and CSIRO projections PAGEREF _Toc15473451 \h 615.8Projected cost of solar: modelling studies and CSIRO projections PAGEREF _Toc15473452 \h 625.9Projected cost of geothermal: modelling studies PAGEREF _Toc15473453 \h 625.10Rate of cost decline to 2050 (per cent per year) PAGEREF _Toc15473454 \h 625.11IRENA suggestions of innovations to deal with VRE integration issues PAGEREF _Toc15473455 \h 665.12Incremental cost of abatement: renewables versus coal-CCS Electricity market. PAGEREF _Toc15473456 \h 67GlossaryAEMOAustralian Energy Market OperatorALPFA Low Pollution Future, Australian Government reportAR5Refers to the IPCCs Fifth Assessment ReportBAUBusiness as usualCCMSClimate Change Mitigation Scenarios, Australian Government reportCCSCarbon capture and storageCGEComputerised General EquilibriumGNPGross National ProductIAMIntegrated assessment modelIPCCIntergovernmental Panel on Climate ChangeROWRest of worldSGLPStrong Growth Low Pollution, Australian Government reportVREVariable Renewable EnergySummaryThis reportThis report uses a variety of studies published in the past 10 years to draw out implications for the cost of greenhouse gas abatement to Australia.The report has a particular emphasis on how the estimates of abatement cost and other economic adjustments are sensitive to underlying model specifications, simulation design, technology cost assumptions and overall scenario construction. The report also considers what form of additional modelling could be used to build on the understanding to date and provide additional useful insights into the cost of abatement and the structural challenges that might emerge from mitigation actions in Australia and the rest of the world.Key results from past modellingOverall (except in a small number of scenarios) the cost of abatement increases steadily as the amount of abatement increases.When expressed in terms of carbon prices, this increase appears almost exponential, however when expressed in terms of deviations in national income, the increase appears more linear.There are a variety of cost of abatement estimates, varying by model and scenario construction. For low levels of abatement, there is rough agreement about the cost, however estimates diverge as the amount of abatement increases.Abatement of around 30 per cent (relative to business as usual) is expected to result in between 1 and 3 per cent reduction in national income (GNP) relative to what it would otherwise have been.(It is important to note that this is a reduction in the level of GNP compared with what it would otherwise have been at some point in the future, not an absolute reduction relative to today. That is, in all the modelling of abatement, the economy continues to grow, but at a slightly lower rate).Abatement of around 50 per cent (relative to business as usual) is expected to result in between 2 and 5 per cent reduction in national income.Higher abatement, around 80 per cent, is estimated to results in between 4 and 8 per cent reduction in national income.Higher cost of abatement is generally further in the future as most model simulations have abatement increasing over time.The optimal timing of abatement — how to minimise the total cost of abatement over time —is usually simulated in the models by assuming that the carbon price increases at the discount rate.The models provide important information about the structural change likely to emerge as a result of mitigation. In most of the reports, this is directly a consequence of increases in the price of energy and the subsequent economy wide effects of this price increase.In this regard, projections of structural change depend very much on underlying model assumptions.The results are sensitive to assumptions about technologyModelling undertaken to date clearly shows that economic costs vary between models as a consequence of different model assumptions and parameters settings. Importantly, all model results are very sensitive to assumptions about the cost and availability of low carbon energy alternatives, including both the absolute level of costs, and the rate at which costs change over time.Results are particularly sensitive to assumptions about carbon capture and storage technology — either its availability or overall cost.As yet there is no single agreed narrative about the specific details of a low carbon technological future. There are a range of different possibilities with different potential outcomes and different implications for overall cost and structural change. There is, however, a common view that abatement will require considerable change in the structure of the energy sector, particularly electricity.The results are sensitive to assumptions about the rest of the worldAnalysis of past studies indicates that abatement actions in the rest of the world — and the policies and technologies that drive them — are very important for Australian economic outcomes. There are a number of reasons for this.First, the overall economic cost of abatement estimated in a number of studies is closely related to Australia’s ability to purchase international abatement. Without this, costs of abatement are considerably higher.Second, abatement actions in the rest of the world directly affect Australia through the demand for raw materials of various kinds.The effect can be negative, in the case of reduced demand for coal and other fossil fuelsThis effect can also be positive through demand for copper, lithium and other minerals associated with renewable energy or with electrification.Third, abatement actions, particularly adoption of renewable energy, around the world directly affect the cost structure of renewable energy as it appears to Australia which, as noted above, affects technology costs.Fourth, specific policy actions — in particular their timing and coverage — affect the results of particular Australian abatement policies. This is especially true in the well known case of emissions intensive and trade exposed industries.Global abatement will significantly change investment and trade flows, so it is not surprising that as a small, trade exposed, economy, Australian results are sensitive to international outcomes.Opportunities for future modellingGiven the range of sensitivities identified, there is considerably opportunity to build on the work already done to further understand potential outcomes of Australian, and world, mitigation.The value of economic modelsWorking through the future implications of climate mitigation — in terms of overall economic costs or the threats and opportunities related to structural change within the economy — essentially amounts to understanding the implications of a wide variety of technological futures.Models are particularly useful for exploring the implications of different assumptions about the future, along with different assumptions about future policy actions. They can be used to assess the detailed implications of scenarios for technological and policy futures.Recommended ‘style’ of modellingIn considering the potential role of future modelling, it is worth noting a distinction between two very different ‘styles’ or ‘purposes’ of modelling.Consolidative modelling is an approach that brings together known facts into a single package which is used as a surrogate for the real world and then used to predict particular outcomes. This is the approach taken by most climate economic modelling.In contrast, in exploratory modelling, models are not used to generate predictions or to elicit ‘answers’ to explicit initial questions, but are used to generate new information helpful in deriving informed decisions. Under exploratory modelling, simulations are used to explore a very wide range of possible outcomes and to understand the broad properties of these outcomes. Exploratory modelling is much more suited to situations in which there is fundamental uncertainty about key parameters or key exogenous assumptions. This is particularly the case, for example, regarding future technology.Useful areas for further analysisWhile the studies undertaken to date are comprehensive and have all involved considerable analytical and development work, there remains considerable scope for further exploration of the implications of future technology and policy scenarios.In particular, we recommend that future analysis should focus on understanding the implications of two areas of uncertaintyFirst, the implications of the abatement actions of other countries on AustraliaSecond, the implications of a range of different technology scenarios for Australia.There is also an important interaction between these two areas in that the implications of other countries actions also include the implications of technological development overseas. More analysis with global modelsA large number of the issues noted above could be fruitfully examined with careful analysis using one, or a number, of the available global economic models. Global models are key because of the need to capture a number of levels of interaction between Australia and the rest of the world, as outlined above. Further, it would be more valuable if the global models were able to capture capital flows as carbon policy particularly has significant implications for returns to capital and hence the availability of capital within the Australian economy.The nature of the Paris Agreement — involving bottom up and highly variable targets and policies proposed by countries themselves — means that variation across countries needs to be captured, and the implications for Australia well understood.Analysis of international developmentsRealistic modelling of country policies to understand the global cost of abatement to Australia will provide important understanding of cost and structural change pressures within the Australian economy.The existing studies clearly indicate that international actions, particularly relating to abatement, have important implications for the cost of abatement in Australia.One of the limitations of the existing modelling is that it uses an idealised representation of global abatement policies. This is usually implemented through an assumption of uniform global prices.Real policy and abatement action is likely to be considerably more fragmented and unlikely to be uniform across all sectors within a given country.Given that Australian outcomes will be determined by international actions through a variety of channels, further exploration of realistic policy scenarios is likely to provide important insights into the evolution of costs for Australia.There are a number of components to realistic modelling of country policies.First, it is important to recognise that policies are rarely applied on a truly economywide basis and that there are very often exemption or rebates for particular industries.Second, it is important to clearly distinguish between ‘taxes’ and ‘subsidies’ in international climate policies. Each of these has different impacts on global trade and investment.Third, it is important to recognise that the headline tax or subsidy rate is not necessarily the rate actually imposed. The extent of the difference will depend on the nature of the policy.For example, in some country emissions trading schemes there is often a substantial overallocation of permits for particular industries which means actual costs paid are lower than the reported permit price.Given recognition of each of these points, it should be within the capacity of existing global models to take these various factors into account, assuming that the existing model has sufficient sectors to distinguish between different applied policies.Most of the information required to do this analysis is available either from country sources or through information collected by the World Bank or the UNFCCC.International policy developments captured in this way will allow understanding in more detail of the ways in which international developments will influence:The demand for particular Australian products (for example, coal versus other mineral products)The availability of ‘international permits’ that could potentially be purchased by AustraliaThe competitive pressures faced by particular Australian energy intensive industriesMore analysis with integrated CGE/energy sector modelsSome of the rapid changes in expectations about renewable costs illustrated in this report — along with the various technical developments required for integration — indicate that careful scenario analysis is required to fully explore the potential implications of higher renewable uptake.These scenarios need to go beyond the simple renewable uptake story simulated to date and capture other aspects of structural developments associated with renewable uptake.Because of the close linkages between the energy sector and the rest of the economy, it would be preferable if these scenarios were analysed using a modelling tool that incorporated CGE and energy markets within the same model. While the past strategy of iterating between a detailed dispatch energy model and a CGE country model has been fruitful, this approach has limitations and runs the risk of missing key interactions that should be kept endogenous.Analysis of technology scenariosAnalysis of rich technology scenarios — including scenarios relating to the costs of integrating a large proportion of renewables — will provide useful insights into adjustment pressures in Australia.This analysis could consider a range of ‘integration’ issues including understanding the variety of technologies needed to maintain a stable and effective electricity grid.This would also include analysis of more complex technology scenarios where energy across different sectors (electricity and transport, for example) is integrated. An example is the potential integration of electric vehicles into the electricity grid (as a source of storage).It may also include implications of more remote developments in solar technology that move beyond current silicon based PV techniques.Some of the technology scenarios noted above will require particular ‘pathways’ for development; a particular sequence of investments.This would explore whether particular sequences result in lower cost overall abatement if they are implemented correctly.The other side of this is the potential cost of ‘lock in’ to ultimately inappropriate technologies, or unintentionally ‘locking out’ other technologies.There is a large amount of information currently available on integration issues, renewable developments and the implication of electric vehicles and other energy developmentsThe key challenge for this analysis will be constructing a sensible set of scenarios that can usefully guide analysis of future outcomes. Such scenarios, once developed, could easily be simulated with existing energy sector specific models, and it is likely that this would be a useful initial stage of analysis to understand exactly how different scenarios lead to different energy cost outcomes.As noted above, however, it is likely that good combined energy sector/CGE modelling will allow the capture of a richer set of interactions than allowed by an energy model alone or allowed through sequential analysis using separate energy sector and CGE models.Analysis with a combined energy/CGE model is likely to be more demanding and resource intensive as development work on such a model will probably be necessary.An important step in this regard will be consulting with existing model developers (within Government and elsewhere) to assess capabilities.Australia’s terms of trade under different technology futuresAs noted in this report, one of the channels of abatement cost to Australia is through a reduction in traditional energy based export (such as coal) as other countries go about abating.However, new energy technologies will require alternative sets of mineral inputs (lithium and copper, for example) and so it is possible that there will be increased exports from these products.This analysis would explore the extent to which this is already implicit in the economic modelling or whether there is an additional economic effect to be captured.Understanding higher order structural changeRelated to this is analysis of new structural relationships within the Australia economy that may emerge through the development and implementation of new energy technologies IntroductionThis report undertakes a review of what recent modelling exercises (roughly in the past 10 years) have said about the cost of abatement to Australia; that is, the cost of meeting various greenhouse gas reduction targets. In the language of the IPCC this is the ‘economic implications of transformation pathways’. The studies broadly covered here are summarised in table 1.1The past decade has seen the publication of a large number of modelling analyses including a sequence produced within the Treasury (in association with other groups including the Garnaut Climate Change Review) along the supporting studies associated with those modelling studies. In each case, these studies were associated with the construction or review of a particular policy framework.Over the same period, there have been a number of modelling reviews of signature energy policies, and in particular the Renewable Energy Target (RET) in its various forms. These reviews have usually involved some detailed modelling analysis, often using very specific electricity sector models.The models used for the analyses range from detailed single country CGE models, to multicountry international CGE models to energy specific sector models.The relatively large amount of activity in Australia mirrors the very large amount of simulation work that has taken place internationally and summarised in some detail as part of the IPCCs AR5 Working Group III reports. There are a large number of international models currently in use.The approach taken in this report is ‘thematic’ in that it uses a variety of studies to make overall points about the whole body of literature. In much the same way that the IPCC has summarised a variety of studies this report tries to draw out themes and implications without necessarily going into the full detail of each individual study. This choice is largely pragmatic and reflects the fact that many of the models are part of what has turned out to be a sequence of studies. IF 1 = 1 " IF STYLEREF "Heading 6" \l \n 0 = 0 " IF STYLEREF "Heading 1" \l \n 1 = 0 "" " STYLEREF "Heading 1" \l \n 1." 1." "B." 1. SEQ Item \* ARABIC \s IF STYLEREF "Heading 6" \l \n 0 = 0 " IF STYLEREF "Heading 1" \l \n 1 = 0 "" "1" 1" "6" 1 1" "" 1.1Overview of key studies coveredYearReportType of model used for analysis2019Fisher, B Economic consequences of alternative Australian climate policy approaches. BAEconomics, March. Referred to as BAEconomics.BAEGEM CGE model of world economy2019Liu, McKibbin, Morris and Wilcoxen Global Economic and Environmental Outcomes of the Paris Agreement Brookings Climate and Energy Economics Project. Referred to as Liu et al in this report.G-Cubed multicountry intertemporal CGE model.2018ACIL and Frontier Economics analysis for National Energy GuaranteeProprietary energy market simulation models2018Kompas, Ha and Che ‘The Effects of Climate Change on GDP by Country and the Global Economic Gains from Complying with the Paris Climate Accord, Earth’s Future 6.Enhanced version of GTAP multicountry model2017Finkel Review of Future Security of National Energy MarketDraws on a range of models including energy sector models2017CSIRO Low Emissions Technology RoadmapDraws on a range of methodologies2017Energy Networks Australia and CSIRO Electricity Network Transformation RoadmapDraws on a range of methodologies2016Energetics Modelling and analysis of Australia’s abatement opportunitiesBottom up estimate of the cost/benefit of particular approaches to abatement2015McKibbin Software Group, Report 1: Economic modelling of international climate action under a new global climate change agreement and Report 2: Economic modelling of Australian action under a new global climate change agreement, Referred to as DFAT G3 in this report.G-Cubed multicountry intertemporal CGE model.2015CSIRO Australian National Outlook. Includes technical reports and Nature publication as well as database of scenario outcomes. Referred to as ANO in this report.Used versions of GTAP and MMRF and CSIRO modes 2014Deloitte Assessing the impact of the Renewable Energy Target. Report prepared for ACCI, BCA and MCA as part of the RET Review.Proprietary models of electricity market2014ACILALLEN RET Review Modelling: Market Modelling of Various RET Policy Options, Report to RET Review Expert PanelProprietary energy sector model.2014ClimateWorks Australia, ANU, CSIRO and CoPS 2014, Pathways to Deep Decarbonisation in 2050: How Australia can prosper in a low carbon world: Technical report, ClimateWorks AustraliaVariety of analysis including MMRF model,2013-14Modelling conducted for the Climate Change Authority’s Special Review (2014-15) and Targets and Progress review. Includes: Treasury and DIICSRTE Climate Change Mitigation Scenarios and ACIL ALLEN Electricity Sector Emissions. Referred to as CCMS in this bination of MMRF, GTEM and energy sector models2013Jiang et al Modelling the Trade Implications of Climate Mitigation Policy RIRDC Publication no 12/104Modified version of GTAP2012SKM Modelling the Renewable Energy Target, Report for the Climate Change Authority. SKM proprietary energy model2011Australian Treasury, Strong Growth Low Pollution. Referred to as SGLP in this bination of MMRF, GTEM and energy sector models2010ClimateWorks Low Carbon Growth Plan for AustraliaBottom up estimate of the cost/benefit of particular approaches to abatement2009McKinsey An Australian Cost Curve for Greenhouse Gas Reduction.Bottom up estimate of the cost/benefit of particular approaches to abatement2009Frontier Economics, The economic impact of the CPRS and modifications to the CPRS. Report for the Coalition and Senator XenophonFrontier proprietary models2008Australian Treasury, Australia’s Low Pollution Future. Referred to as ALPF in this bination of MMRF, GTEM and energy sector models2008Garnaut Climate Change ReviewCombination of MMRF, GTEM and energy sector modelsThe report begins with an overview of the key features of the models used (chapter 2) and then proceeds to summarise key model outcomes (chapter 3). Wherever possible, results from different models are expressed in a common basis (usually in terms of abatement relative to BAU).Chapter 4 summarises the key sources of model input data, while chapter 5 tries to summarise how the various results are sensitive to underlying assumptions.Chapter 6 draws some broad conclusions and makes some suggestions about what future avenues of model investigation may be useful.Models and key characteristicsThere are, in effect, four broad levels of models have been used to assess the cost of abatement in Australia (and elsewhere in the world).Global economic modelsThe first level is a global economic model that covers a number of individual countries or regional groupings of countries along with a number of sectors within each country or region (for convenience we will refer to countries in what follows, under the understanding that this may also refer to regions in some cases). The global models track trade and investment flows between countries and have a treatment of greenhouse gas emissions as well as a distinction between carbon and non-carbon sources of energy. Global models cover other sources of emissions to varying degrees. In Australia the key global models used have been GTEM and G-Cubed (see below).Examples of studies using global models included in this report are the 2019 studies by Fisher and by Liu et al (see table 1.1); the 2015 study by the McKibbin Software Group; and a portion of the each of the sequence of Treasury studies (ALPF, SGLP and CCMS).The 2018 analysis by Kompas et al is also an example of the use of a global economic model based around the GTAP project. The Kompas model involves a considerable enhancement of the standard GTAP model.Detailed country modelsThe second level is a detailed country economic model, usually a general equilibrium model that contains a detailed picture of the economic structure of a single country. These models usually have considerably more industry and commodity detail than the global model and often have very detailed treatments of other aspects of the economy such as fiscal structure and government accounts. In Australia, the most frequently used country model is MMRF (see below). Example of studies using this model include the sequence of Treasury studies (ALPF, SGLP and CCMS).Detailed sectoral modelsThe third level are detailed sectoral models that focus on a specific source of emissions in detail. These include, for example, detailed electricity sector models that track all types of generation activity in considerably more detail than is possible in the GE economic models. A number of electricity sector models have been used in Australia including those used by ROAM (now EY), SKM MMA (now Jacobs), Frontier Economics and ACIL ALLEN.Each of these models was used, for example, in the sequence of Treasury studies (ALPF, SGLP and CCMS) as well as in individual studies looking at the RET (including, for example, the 2014 ACILALLEN RET review).Integrated assessment modelsThe final level or type of model are integrated assessment models (IAMs). These are an integrated combination of economic and climate models that seek to couple economic activity and climate outcomes within the one modelling framework. IAMs usually contain a reduced form representation of both the economy and the climate in order to make solution of the model tractable. The most well know IAM is Nordhaus’ DICE model (the development and use of which contributed to Nordhaus’ Nobel Prize in economics). IAMs have different levels of complexity and have been widely used in the international literature.The 2018 analysis by Kompas et al reflects the use of a global economic model to assess the implications of temperature increases as a results of climate change. Strictly speaking, however, this model does not incorporate a climate sector and so is not an IAM in the sense noted above.Integrating models at different levelsA typical approach in Australia has been to integrate the use of the first three levels of model into a single study. The global model is used to understand interactions between Australia and the rest of the world, the country model is used to understand detailed economic impacts within Australia and the sector models provide detail not otherwise available in the global model.One of the major challenges in using three levels of model is dealing with the ‘double endogeneity’ problem. Essentially, all three levels of model determine some of the same variables, and ensuring consistency between models is crucial in order to obtain sensible results. For example, the global and the Australian models both determine Australian GDP, output by sector, consumption, investment and so on. Similarly, the Australian country model and the detailed electricity sector models both determine electricity output and generation by fuel type. This problem is generally resolved by iterating between models until the doubly determined variables a consistent across models.GTEMThe Global Trade and Environment Model (GTEM) was developed by the Australian Bureau of Agricultural and Resource Economics (ABARE) and has been used frequently for climate change policy analysis. It is used in all three sequences of Treasury modelling and also forms the basis of the BAEGEM model used in the 2019 Fisher study. GTEM has also been used as part of international model comparison exercise. GTEM is a global model that provides insights into what happens to Australia’s major international trading partners. However, it has less industry detail than found in the MMRF model. Documentation of GTEM is available online.GTEM is not currently available in the public domain, however much of its structure is similar to that of the GTAP model which is readily available and used by a large consortium of modellers. GTAP has regular training courses and conferences. (See ). G-CubedThe G-Cubed model, developed by Warwick McKibbin and Peter Wilcoxen, is also a global model. Current versions of G-Cubed are based around the GTAP database and actual applications of the model vary in the sectoral and country detail actually applied. While G-Cubed model potentially has the same regional and industry detail as GTEM, it’s macroeconomic linkages are considerably more developed. For instance, unlike GTEM, G-Cubed can provide estimates of what happens to inflation, and contains elements of forward-looking behaviour which has important implications for the effects of mitigation policy. G-Cubed also contains a detailed treatment of capital flows between countries, which has also proved important in understating climate policy. The theoretical structure of G-Cubed is described in McKibbin & Wilcoxen (1998).G-Cubed has been used in a wide variety of studies and model comparison exercises.MMRF/VURMThe Monash Multi Regional Forecasting (MMRF) model, developed by the Centre of Policy Studies (CoPS) at Monash University (now at Victoria University), is a detailed model of the Australian economy that gives results for all eight States and Territories. MMRF has rich industry detail (with 58 industrial sectors).Documentation of MMRF (along with a rich database of other working papers) is available at . The MMRF model has a very strong pedigree, evolving out of a long Australian CGE work program supported by a variety of organisations. Indeed, in many ways the team behind MMRF have been at the leading front of CGE modelling globally for many years. The model, or variants of it, are relatively easy to access, and CoPS undertakes a wide variety of educational activities.While the CoPS model used for past analysis was termed MMRF, the current operational version is VURM (Victoria University Regional Model). Documentation for this version of the model is available at . Energy sector modelsGlobally, there is a very large number of electricity market models used for a variety of analyses including greenhouse abatement. Ringkjob et al 2018 provide an overview of 75 modelling tools available globally. Energy sector models are generally large linear programming models that seek to find the best combination of generation capacity that satisfies system constraints ACIL ALLENFor much of the analysis referred to here, ACIL ALLEN use an electricity model called PowerMark LT which is a dynamic least cost model which optimises existing and new generation operation and investments. The model is a large linear program and is solved over a chosen time horizon subject to a range of input assumptions (generators costs, development costs, interconnectors and so on), system constraints and government policy settings. PowerMark LT is based around a very detailed generator database and contains rich possibilities for specifying future scenarios.ACIL ALLEN also have a short term version of their model called PowerMark ST which is more detailed simulation model designed to replicate the market operators dispatch engine at half hourly or hourly frequency.Both PowerMark LT and PowerMark ST are commercial models and are not in the public domain.SKM MMASKM MMA (now Jacobs) use a variety of customised electricity as gas sector models to simulate dispatch in the electricity market. The overall modelling framework is provided by the Strategist modelling tool. While there does not appear to be documentation available for this, it effectively operates as a LP dispatch model.ROAMROAM consulting (now part of Ernst and Young, EY) used a proprietary electricity market modelling package called 2-4-C. 2-4-C was built to match as closely as possible the operation of the AEMO Market Dispatch Engine used for real day-to-day dispatch in the NEM. 2-4-C implements the highest level of detail and bases dispatch decisions on generator bidding patterns and availabilities in the same way that the real NEM operates. Simulations include modelling of generator outages (full, partial and planned), intermittent or inflexible generators and inter-regional transmission capabilities and constraints. Generator bidding strategies are derived from real bid profiles and operational behaviours taken from generators in the relevant system, adjusted over time for any changing market conditions. Such conditions might include water availability, changes in regulatory measures or fuel availability.Frontier EconomicsFrontier Economics have a suite of energy modelling tools used for climate and other energy policy.Whirlygig calculates the least-cost mix of existing plant and new plant options to meet load over time by optimising total generation costs in the market and is most often used to inform decisions on the future long run marginal cost in an electricity market.Spark models generator bidding behaviour by combining a least-cost electricity dispatch engine with game theory.Strike determines an efficient mix of energy purchasing instruments from a suite of options (spot, physical and financial) for a range of risk levels using portfolio theory..Whirlygas examines the longer-term effects of changes in competitive gas markets such as the efficient (least-cost) operation and investment in a domestic or international gas market over a long-term investment horizon (e.g. 5-50 years). Integrated assessment models (IAMs) in useThere is a wide variety of international models (including IAMs) that have been used for greenhouse policy analysis. Details of these are available as part of the AR5 Scenario Database, prepared as part of the IPCCs Fifth Assessment report. The IPCC identified two broad types of IAM:Policy optimisations models (POMs) — which focus on a complete cost benefit analysis of climate change mitigation and optimal policy. These models include climate damage functions, and also include a rudimentary representation of the climate systems, so that the link between emissions and temperature increase and damages is closed.Policy effectiveness models (PEMs) — which focus on the cost effectiveness of achieving a particular target with a particular set of policies.Given that most of the models discussed above are in fact PEMs, the focus on IAMs here is specifically on POMs. One disadvantage is PEMs is that outcomes always appear as a cost; the benefit of the target (reduced future climate costs) is rarely discussed. An IAM has the potential to include benefits of the policy along with the abatement costs associated with it.Table 2.1 summarises some recent IAMs IF 1 = 1 " IF STYLEREF "Heading 6" \l \n 0 = 0 " IF STYLEREF "Heading 1" \l \n 2 = 0 "" " STYLEREF "Heading 1" \l \n 2." 2." "B." 2. SEQ Item \* ARABIC \s IF STYLEREF "Heading 6" \l \n 0 = 0 " IF STYLEREF "Heading 1" \l \n 2 = 0 "" "1" 1" "6" 1 1" "" 2.1Features of selected IAMsModelRegionsDamage functionTechnologyDICE (DICE- 2013R)GlobalLinear-quadraticExogenous; backstop technologyRICE (RICE- 2010)12Quadratic, region-specificExogenous; backstop technologyFEEM-RICE10Quadratic, region-specificEndogenous, learning by doing and researchENTICEGlobalLinear-quadraticEndogenous FUND16Complex, different damage functions for each regionAutonomous improvementPAGE098Power function with uncertain exponentEndogenous (learning)WITCH12Quadratic, regionalEndogenous, learning by doing and researchMERGEFlexibleQuadratic; considers catastrophesExogenousICAM17ComplexEndogenousSource: Farmer et al ‘A Third Wave in the Economics of Climate Change’ Environmental and Resource Economics (2015) 62:329-357. Table 2. Summary of model typesChart 2.2 provides a summary of the different model types discussed in this chapter. IF 1 = 1 " IF STYLEREF "Heading 6" \l \n 0 = 0 " IF STYLEREF "Heading 1" \l \n 2 = 0 "" " STYLEREF "Heading 1" \l \n 2." 2." "B." 2. SEQ Item \* ARABIC \s IF STYLEREF "Heading 6" \l \n 0 = 0 " IF STYLEREF "Heading 1" \l \n 2 = 0 "" "1" 1" "6" 1 2" "" 2.2Summary of model typesTypeCountry coverageSector coverageTreatment of climateCore useMulticountry CGE model. Examples include GTEM, G-Cubed, GTAP and BAEGEMMultiple countries, typically 20 or more with trade flows modelled. In some cases, investment flows are also modelled.Generally multiple sectors, in the order of 20 to 50 depending on the model configurationWhere imposed, climate change is outside the model and can be simulated through productivity and other shocksUsed to assess global costs of abatement and global income, trade and investment consequences of carbon policies.While typically used for multicountry comparisons, they can also be used to analyse outcomes for a single country or a constrained group of countries.Single country CGE models. Examples include MMRF/VURMOne country modelled in detail, trade with rest of world covered as an aggregate (individual destinations or sources not captured).Typically, of the order of 100 or more (in line with standard input-output table industry coverage).Where imposed, climate change is outside the model and can be simulated through productivity and other shocksUsed to assess cost of abatement and GDP and sectoral effects of efforts to reduce greenhouse gas emissions.Sector specific, typically energy sector, models. Examples include a range of proprietary models.Generally single countryGenerally, a single sector, but with very detailed coverage of that sector. For example, an electricity dispatch model will include details of all generators and generation types.Where imposed, climate change is outside the model and can be simulated through productivity and other shocksExamine detailed implications of climate policy for a single sector.Integrated Assessment Models (see table 2.1)Multicountry, but generally in regional groupings. Generally, very limited sectoral disaggregation, or the whole economy is represented as a single sectorClimate module generally included within the modelTypically used to assess aggregate global temperature or abatement targets where the cost of abatement can be related to the cost of climate change.Source: CIEOverview of key model outcomesModelling strategiesUnderstanding the effects of abatement policy on a particular country, and on particular sectors within that country, requires a conceptual modelling framework that can deal with economic interactions that take place:Between countries, through trade and capital flowsBetween sectors within a single country, through intersectoral linkages (buy and selling from one sector to another) and through competing demands for labour and capital inputsBetween different energy sources within an energy marketBetween different technologies in non-energy sectors (agriculture and transport, for example)As a consequence of macro-economic interactions in determining exchange rates, interest rates and the overall price level within an economy.Some modelling studies use a single, usually global, model to capture all of these effects. For example, the studies by Fisher and McKibbin noted in Table 1.1 each use a single global model to assess mitigation policies. A large number of the studies in Table 1.1 use of sequence of models to capture different aspects of the interactions noted above.While particular modelling strategies differ, the key point is that conceptually, all of these many interactions are simultaneous, with feedbacks between all levels of effect.Key model mechanismsGlobal price path and global model outcomesA typical procedure in many modelling studies is to first use a global model to establish a global carbon price consistent with achieving particular environmental targets. Often, targets are expressed in terms of greenhouse gas concentration or temperature changes. Concentration and temperature change are usually converted to actual emission targets using some form of global climate model.The global economic model then takes the emissions targets as a constraint to calculate a price path for global abatement. That is, the model is used to find the carbon price in each year that, when applied in each individual country, leads to the targeted amount of global abatement. The global model may also include constraints on individual country targets, although most commonly these are imposed subsequently.The global price path that emerges depends on the entire economic structure of the global model, most importantly the costs associated with substituting between carbon and non-carbon forms of production. If substitution is modelled as difficult (through the choice of particular model parameters) then the carbon price required to achieve a given level of abatement will be relatively high (and vice versa). At the same time, the economic relationships between countries will also influence the global price path.The price path for global abatement thus depends on individual country costs of abatement and any specific country targets within the overall global target. This price path, when simulated in the global model has a range of implications for individual country production, consumption and trade as well as energy consumption and use.The optimal timing of abatementThe economic analysis in most of the studies contains a number of implicit ideas about the optimal timing of abatement: that is, what this the optimal pathway to achieve a particular endpoint, or total budget, target?Not surprisingly, the applied model results follow closely the key findings of the more theoretical literature (see for example Goulder & Mathai 2000, and Pearce & McKibbin 2007) which has three broad findings about optimal abatement pathways.First, the marginal cost of abatement needs to be the same at all points in time, so that abatement is optimally spread. This in turn requires that the implicit cost of abatement — what would be the carbon price in an explicit pricing policy — rises at the discount rate. With a higher discount rate, more abatement is delayed to the future. With a lower discount rate, abatement is brought forward.Second, consistent with this first point, the amount of optimal abatement at each point in time depends upon the evolution of the cost of abatement which is itself partly a product of technology. Technological outcomes will be a combination of two distinct types of technological change: If the cost of abatement (technology) is exogenous, and falling, this tends to suggest delaying abatement over time, all other things equal.If the cost of abatement (technology) depends on ‘learning by doing’ so that the cost of abatement in the future depends on the amount of abatement in the past, then while the optimal path is difficult to predict, this does tend to suggest more abatement early on.Third, the cost of abatement in the future may be a function of specific policy choices today, such that the optimal abatement path depends specifically on policy choices.Underlying most modelling studies is a common strategy to specify that the global price path should increase each year at the discount rate. Given the structure of the models, and technological assumption, this then implies the optimal abatement path.Application to AustraliaInformation from the global model is then applied to the Australian country model. In the first instance, the global price path (when adjusted for the exchange rate) is imposed within the Australian economy under the assumption that Australia is effectively a price taker on international markets.In addition, the global model will imply a pattern of terms of trade and trade and capital flow changes that need to be directly simulated in the country model. Bottom up sectoral resultsAt the same time, individual sector models — particularly models of the electricity sector — are used to simulate the effect of carbon prices on the electricity and energy sectors. This results in changes in the generation mix as well as changes in overall energy prices. These changes are then taken from the ‘bottom up’ and imposed upon the country model.There is an iterative process between these two levels of model, as the Australian economywide model is used to impose an overall level of demand on the sector model. This changes once the carbon price is imposed, which changes again outcomes in the electricity sector model, and so on.Mechanisms underlying the cost of abatementThe ultimate cost of abatement (which can be measured in a number of ways) is the net result of a complex set of interactions within the modelling frameworks. An important feature of the Australian studies is that the cost of abatement to Australia depends on both interactions within the Australian economy as well as on abatement actions in other countries and hence trade and capital flows between countries. Within the Australian economy, the cost of abatement arises through the resources foregone in order to invest in new generation capacity along with the overall cost increase in the economy as a result of more expensive electricity and other inputs. One of the key mechanisms within the Australian CGE models is a reduction in export income through cost increases within the economy. The export intensive industries then reduce demand for goods and services, and effect which flows through to other sectors of the economy.Increases in export costs, and changes in exports, are generally associated with a depreciation of the real exchange rate, which has second round effects on exports, particularly export commodities that may not be subject to a carbon tax.Increases in costs also affect the rate of return to capital which reduces investment incentives throughout the economy (and may also change capital flow from overseas).Increased costs also reduce household real income, particularly through a reduction in the real wage, which has further flow on implications to the domestic economy. Real wages fall because of reduced demand for labour associated with reduced capital expenditure and reduced exports.The importance of international action in determining abatement costs in Australia arises for two main reasons:First, as will be illustrated further below and in the chapter on sensitivity analysis, most of the studies assume that Australia is able to take advantage of international abatement as part of meeting its own targets. The availability of international abatement lowers overall cost of abatement to Australia — as it is a general finding that the cost of abatement to Australia is higher than that for other countries. This reduction in the initial cost of abatement then modifies further flow on effects.Second, international abatement actions lead to reduced demand for key Australian exports, particularly energy exports such as coal. This imposes a cost to Australia through what is effectively a large reduction in Australia’s terms of trade. The extent of this effect depends on the extent of abatement elsewhere and the nature of the technologies that are taken up as that abatement takes place. This a reduction in exports in addition to the effect of increase costs in the economy.Within the Australia economy, the domestic cost of abatement depends largely on the cost of substituting to lower emitting forms of energy, particularly in the electricity sector. This in turn depends on the evolution of low carbon electricity technologies and on how the cost of switching emerges over time. This domestic cost of abatement interacts with the international cost of abatement to determine the ultimate point to which domestic abatement takes place.Important implicit assumptionsA general assumption of most of the modelling studies is that of a uniform global carbon price transmitted around the world. Within this broad assumption, there are a number of other underlying assumptions including:First, that an international market of some form (perhaps an offset market) is feasible and is established. Second, within this international market, prices reflect the marginal cost of abatement of participants. Third, the global cap (or the individual country targets) is met.If assumptions about country participation are relaxed or changed, then clearly this will have implications both for the global carbon price, and total global abatement.The Productivity Commission recently noted that most countries are not implementing carbon policies in the most cost effective way. This means that as policy is currently emerging, it is unlikely that the true cost of abatement will be revealed in international markets. Abatement and the implied carbon priceA simple starting point in considering what existing modelling studies say about the cost of abatement is to look at the relationship between abatement and the ‘carbon price’ implied by the various studies.Chart 3.1 maps this relationship, showing the amount of abatement relative to business as usual (BAU) on the horizontal axis, and the Australian carbon price associated with that abatement on the vertical axis. The top panel of chart 3.1 covers simulations including both domestic abatement as well as abatement through purchase in international units. The lower panel of the chart shows the relationship for domestic abatement only.This chart is constructed by taking published carbon price and abatement estimates from each of the modelling studies. In this chart (and those that follow) abatement refers to reduction in emissions relative to a particular BAU year. While the individual datapoints refer to different years, chart 3.1 abstracts from this in order to provide a common comparison between models.Different models have different time frames, different sequencing of abatement and different mechanisms operating. This makes it difficult to provide a common point of comparison between models.The information in chart 3.1 is thus a ‘reduced form’ of a large number of model mechanisms and scenario settings that lead to the link between a carbon price and abatement. In effect, the chart shows the implicit, net, abatement cost curve for Australia given these various scenario settings and model mechanisms. Note that while the carbon price required to induce a particular amount of abatement provides some indication of the cost of abatement, it does not cover all costs, and so is not a full measure of the cost of abatement. Other measures are considered in more detail below. Nevertheless, this marginal cost of abatement as presented in chart 3.1 has a number of important implications.The results indicate that there is quite a large spread of results across different models and scenarios. This provides an initial indication of the importance of using a variety of modelling studies and scenarios, rather than relying on one or two, as there is quite a large range of possibilities.All the studies show a strong upward slope to the marginal cost of abatement. Each of the panels in chart 3.1 also shows two fitted summary curves for the data (an exponential curve and a power curve). This provides a useful summary of the extent of slope of the marginal cost of abatement. In the case of the power curve, for example, the coefficient on the ‘x’ or abatement variable can be interpreted as an elasticity. Thus, the top panel indicates that a 1 per cent increase in abatement (relative to BAU) requires a 0.87 per cent increase in the carbon paring the slopes between the top and bottom panels of the chart indicates that the availability of some form of international abatement significantly lowers the marginal cost of abatement for Australia. This is a theme which emerges in a number of ways from all of the modelling studies.Many of the studies do not capture abatement beyond about 70 or 80 per cent relative to BAU, and the main clustering of results is up to around 50 per cent (relative to BAU). This means that care must be taken when considering very high levels of abatement as this may be outside the range of existing models.The position of the abatement cost curve implied by each study depends very much on the model mechanisms and the particular scenario considered. The results here do not suggest that more recent studies (which would be expected to include a lower cost of renewables, for example) have a lower marginal cost of abatement.This can be seen in more detail by looking at chart 3.2 which shows the abatement cost curve separately for each of the Treasury studies considered in this report. The variations within each study are much more significant than any overall differences between the studies. This is despite a common perception that renewable costs fell considerably in the time between the studies. The reasons for this are examined in more detail in the ‘Evolution of renewable costs’ section in chapter 5.Chart 3.2 (the lower panel) also shows the relationship between abatement and price for the BAEconomics and G-Cubed results. The former study clearly has much higher carbon prices associated with each level of abatement than any of the Treasury studies, while the G-Cubed results have a much lower carbon price associated with each level of abatement. These very stark differences arise through the very different structures of the models. It is important to note, however, that when the cost of abatement is measured in terms of foregone GDP (see the discussion below on macroeconomic impacts) the differences that remain are not as large.The overall relationship mapped out here is very similar to that set out in the IPCCs AR5 Working Group III report (illustrated in more detail below). The upward sloping marginal cost of abatement curve, along with a wide variety of outcomes depending on model and scenario is a common finding across the entire literature. IF 1 = 1 " IF STYLEREF "Heading 6" \l \n 0 = 0 " IF STYLEREF "Heading 1" \l \n 3 = 0 "" " STYLEREF "Heading 1" \l \n 3." 3." " STYLEREF "Heading 6" \l \n A." 3. SEQ Item \* ARABIC \s IF STYLEREF "Heading 6" \l \n 0 = 0 " IF STYLEREF "Heading 1" \l \n 3 = 0 "" "1" 1" "6" 1 1" "" 3.1Carbon price and abatement outcomes for Australia Real $2010Data source: Modified and updated from Pearce, D 2012 ‘Empirical uncertainties in climate policy implementation’ The Australian Economic Review, Vo. 45, No.1. Updated data from Commonwealth of Australia 2013, Climate Change Mitigation Scenarios and from Jiang et al 2013 Modelling the trade implications of climate mitigation policy RIRDC Publication No. 12/104, July. IF 1 = 1 " IF STYLEREF "Heading 6" \l \n 0 = 0 " IF STYLEREF "Heading 1" \l \n 3 = 0 "" " STYLEREF "Heading 1" \l \n 3." 3." " STYLEREF "Heading 6" \l \n A." 3. SEQ Item \* ARABIC \s IF STYLEREF "Heading 6" \l \n 0 = 0 " IF STYLEREF "Heading 1" \l \n 3 = 0 "" "1" 1" "6" 1 2" "" 3.2Carbon price and abatement outcomes for Treasury, BAEconomics and G-CubedNotes on these scenarios: There are a number of different scenarios compared for each set of Treasury results. In each case, as well as the different scenarios from the underlying study, separate results are presented for the abatement target with, and without, international permits. The higher cost variants in each case involve domestic abatement only with no use of international units. For ALFP, scenarios used are CPRS -5, CPRS -15, Garnaut -10 and Garnaut -25. For SGLP scenarios used are the ‘Core policy scenario’ and ‘High price scenario’. For CCMS, scenarios used are the ‘Central policy scenario’, ‘Low price scenario’ and ‘High price scenario’.Data source: Studies as noted in table 1.1.Macroeconomic impactsAnother way to summarise the model information is to look at changes in national income (GDP or GNI) associated with different levels of abatement for different studies and different scenarios. While GDP is a common measure reported by many studies, it is not ideal as a welfare measure as it does not account for income accruing to Australians. Where possible, the analysis below reports GNI or GNP as these capture income accruing to Australians and account for the need to purchase international permits.Studies typically measure the reduction in GNI/GNP or GDP relative to business as usual as a result of abatement. It is important to note that in each study, despite the reduction in GNI/GDP relative to BAU, the economy continues to grow overall, just at a slower rate than otherwise.Chart 3.3 provides an initial summary for the three sets of Treasury studies, along with an illustrative comparison of a number of studies from the 1990s.Broadly, this comparison shows that:For up to around 25 per cent abatement, all sets of studies are very similar, with this amount of abatement reducing GNP by between 1 and 2 per cent (relative to BAU),For higher abatement than this, the studies from the 1990s appear to have lower cost (per unit of abatement relative to BAU).From about 30 per cent to 60 per cent abatement, all the Treasury results have similar outcomes, but they start to diverge from about 70 per cent abatement.Overall, 40 to 50 per cent abatement is associated with about a 5 to 6 per cent reduction GNP (relative to BAU). Higher abatement has costs of about 7 to 8 per cent of GNP. IF 1 = 1 " IF STYLEREF "Heading 6" \l \n 0 = 0 " IF STYLEREF "Heading 1" \l \n 3 = 0 "" " STYLEREF "Heading 1" \l \n 3." 3." " STYLEREF "Heading 6" \l \n A." 3. SEQ Item \* ARABIC \s IF STYLEREF "Heading 6" \l \n 0 = 0 " IF STYLEREF "Heading 1" \l \n 3 = 0 "" "1" 1" "6" 1 3" "" 3.3Reduction in GNP relative to abatementData source: Treasury publications; McKibbin and Pearce 1996, ‘Global Carbon Taxes: An Australian Perspective’ in Bouma et al 1996 Greenhouse: Coping with climate change, CSIRO; McDougall and Dixon 1996, ‘Analysing the economywide effects of an energy tax: results for Australia from the Orani-E model’ in Bouma et al 1996 Greenhouse: Coping with climate change, CSIRO; Tulpule et al 1999, ‘The Kyoto Protocol: An Economic Analysis Using GTEM’ The Energy Journal Special Issue 1999, IAEE; McKibbin et al 1999, ‘Emissions Trading, Capital Flows and the Kyoto Protocol’ The Energy Journal Special Issue 1999, IAEE.As noted previously, these results abstract from time. However, in most of the studies the mid-levels of abatement refer to 2030, while the higher levels generally refer to 2050. It is also important to note that the studies compared all make a range of assumptions about BAU, implementation of the carbon price and so on. Chart 3.4 extends the comparison by including results from the G-Cubed model (some published as part of the original Treasury study, and some since published independently) as well as recently reported results from BAEconomics. This comparison illustrates a number of key points.The G-Cubed results are similar to other Treasury results for abatement of up to around 40 per cent, after which there is divergence (with G-Cubed implying lower costs).Lower costs for the G-Cubed results commissioned by DFAT. This outcome is discussed further below but is related to the fact that this simulation is for Australian abatement only. This is important in the G-Cubed context as it finds that most abatement costs are related to the flow on effects of international abatement.The BAEconomics results are similar to the other studies for abatement up to around 20 per cent, but higher than the others for abatement around 30 per cent and above. IF 1 = 1 " IF STYLEREF "Heading 6" \l \n 0 = 0 " IF STYLEREF "Heading 1" \l \n 3 = 0 "" " STYLEREF "Heading 1" \l \n 3." 3." " STYLEREF "Heading 6" \l \n A." 3. SEQ Item \* ARABIC \s IF STYLEREF "Heading 6" \l \n 0 = 0 " IF STYLEREF "Heading 1" \l \n 3 = 0 "" "1" 1" "6" 1 4" "" 3.4GNP and abatement including G-Cubed and BAEconomics resultsData source: Publications as listed.Next, chart 3.5 adds to the comparison the recent detailed scenario work undertaken by CSIRO as part of the Australian National Outlook (ANO) project. The ANO contains a large number of scenarios, but two in particular are presented here. First, ‘ANO standard’, which is a comparison between ‘do nothing’ and abatement with neither enhanced productivity changes nor enhanced markets for carbon. Second, ‘ANO high efficiency’, assumes is a comparison between no abatement and global action with higher productivity and more carbon markets.‘ANO standard’ shows very steeply increasing costs as abatement reaches just above 40 per cent. In contrast, ‘ANO high efficiency’ has similar costs of abatement up to around 40 per cent, then significantly lower after that, with the costs of abatement in fact decreasing slightly up to abatement of around 80 per cent. This dramatic difference is a consequence of the high productivity economy, with established carbon markets, assumed in the high efficiency scenario comparison. IF 1 = 1 " IF STYLEREF "Heading 6" \l \n 0 = 0 " IF STYLEREF "Heading 1" \l \n 3 = 0 "" " STYLEREF "Heading 1" \l \n 3." 3." " STYLEREF "Heading 6" \l \n A." 3. SEQ Item \* ARABIC \s IF STYLEREF "Heading 6" \l \n 0 = 0 " IF STYLEREF "Heading 1" \l \n 3 = 0 "" "1" 1" "6" 1 5" "" 3.5GNP and abatement including two ANO scenariosNote: ‘ANO standard’ refers to a comparison between ‘Materials intensive’ and M3CR. ‘ANO high efficiency’ refers to a comparison between H3X1 and ‘Stretch’. Data source: Studies as indicatedComparison with international outcomesChart 3.6 reproduces carbon price, GNP and abatement comparisons undertaken as part of the IPCCs Fifth Assessment Report. These international studies show a similar pattern of results to those summarised above for Australia, in particular the high dispersion of results along with steadily increases costs as abatement increases. IF 1 = 1 " IF STYLEREF "Heading 6" \l \n 0 = 0 " IF STYLEREF "Heading 1" \l \n 3 = 0 "" " STYLEREF "Heading 1" \l \n 3." 3." " STYLEREF "Heading 6" \l \n A." 3. SEQ Item \* ARABIC \s IF STYLEREF "Heading 6" \l \n 0 = 0 " IF STYLEREF "Heading 1" \l \n 3 = 0 "" "1" 1" "6" 1 6" "" 3.6Comparison with AR5 WGIII Data source: IPCC, Assessment report 5, Working group III, figure 6.23. (). While it is hard to make definitive comparisons, examination of chart 3.5 indicates that overall the international costs of abatement appear slightly lower than those for Australia. This is consistent with evidence from the Australian modelling studies as outlined further below.Bottom-up estimates of marginal abatement cost (MAC) curvesIn addition to the economywide modelling analysis noted above, the past 10 years has seen the publication of a number of abatement cost curves generated through bottom-up analyses of particular emissions saving technologies applied to particular activities within the economy. Typically these include detailed analyses of energy saving technologies, better production techniques along with particular changes in agriculture and land use.An early example of this approach is provided by McKinsey’s 2009 analysis An Australian Cost Curve for Greenhouse Gas Reduction. Chart 3.7 illustrates the basic cost curve presented by McKinsey. A key feature is that there is a considerable amount of abatement that can take place at ‘negative cost’, that is, for a net gain. IF 1 = 1 " IF STYLEREF "Heading 6" \l \n 0 = 0 " IF STYLEREF "Heading 1" \l \n 3 = 0 "" " STYLEREF "Heading 1" \l \n 3." 3." " STYLEREF "Heading 6" \l \n A." 3. SEQ Item \* ARABIC \s IF STYLEREF "Heading 6" \l \n 0 = 0 " IF STYLEREF "Heading 1" \l \n 3 = 0 "" "1" 1" "6" 1 7" "" 3.7McKinsey Australian Abatement Cost CurveSource: McKinsey 2009 An Australian Cost Curve for Greenhouse Gas Reduction negative cost feature has continued through more recent applications of the broad McKinsey approach including work by ClimateWorks (chart 3.8) and Energetics (chart 3.9). IF 1 = 1 " IF STYLEREF "Heading 6" \l \n 0 = 0 " IF STYLEREF "Heading 1" \l \n 3 = 0 "" " STYLEREF "Heading 1" \l \n 3." 3." " STYLEREF "Heading 6" \l \n A." 3. SEQ Item \* ARABIC \s IF STYLEREF "Heading 6" \l \n 0 = 0 " IF STYLEREF "Heading 1" \l \n 3 = 0 "" "1" 1" "6" 1 8" "" 3.8ClimateWorks Australia Abatement Cost CurveSource: ClimateWorks 2010 Low Carbon Growth Plan for Australia. IF 1 = 1 " IF STYLEREF "Heading 6" \l \n 0 = 0 " IF STYLEREF "Heading 1" \l \n 3 = 0 "" " STYLEREF "Heading 1" \l \n 3." 3." " STYLEREF "Heading 6" \l \n A." 3. SEQ Item \* ARABIC \s IF STYLEREF "Heading 6" \l \n 0 = 0 " IF STYLEREF "Heading 1" \l \n 3 = 0 "" "1" 1" "6" 1 9" "" 3.9Energetics Australia Abatement Cost CurveSource: Energetics 2016 Modelling and Analysis of Australia’s Abatement Opportunities GDP costsThe work by McKinsey and ClimateWorks related abatement relative to business as usual to GDP costs, so a broad comparison with the modelling work summarised above is possible. McKinsey noted that a 45 per cent reduction in emissions relative to BAU would come at a cost of 0.2 per cent of GDP, while a 72 per cent reduction in emissions is associated with a cost of 0.35 per cent of GDP. ClimateWorks found that a 37.5 per cent reduction in emissions relative to BAU would be associated with a 0.1 per cent reduction in GDP. Comparing this with any of the results in charts 3.3, 3.4 and 3.6 shows that these outcomes are below the lowest end of the modelling results.Carbon price and abatementChart 3.10 further illustrates the difference between the CGE models and the MAC curves by comparing plotting carbon prices against abatement for ClimateWorks and Energetics and comparing this with the Treasury CGE model results presented above.Again, the striking negative cost for low levels of abatement is clearly evident. It is also interesting to note that both ClimateWorks and Energetics have sharply increasing cost curves as abatement increases. It is important to note, however, that the sharp increase in costs may be an artefact of how the MAC curves are constructed — generally they refer to a particular target year, and they are constructed to find the cost of a particular level of abatement. This means that future MAC estimates generated for different levels of abatement may move to the right, with the sharp cost increase being pushed forward. IF 1 = 1 " IF STYLEREF "Heading 6" \l \n 0 = 0 " IF STYLEREF "Heading 1" \l \n 3 = 0 "" " STYLEREF "Heading 1" \l \n 3." 3." " STYLEREF "Heading 6" \l \n A." 3. SEQ Item \* ARABIC \s IF STYLEREF "Heading 6" \l \n 0 = 0 " IF STYLEREF "Heading 1" \l \n 3 = 0 "" "1" 1" "6" 1 10" "" 3.10Carbon price and abatement outcomes: CGE models versus bottom-up MAC curves Data source: As noted in table 1.1Negative costs a core differenceThe reason for the low GDP and carbon costs in these studies relates to the most striking feature of the MAC curves: the large negative component whereby some abatement is estimated to take place at negative cost, that is for a net benefit. This suggests that there are economic gains currently available that have not been taken up. This differs markedly from the approach taken in CGE models which assumes that all privately profitable abatement opportunities have already been taken up so that any incremental abatement must come at incremental cost.One of the challenges with the MAC approach is in explaining why these profitable opportunities have not, in fact, been taken up to date. Some of this is explained by the difference between private and social profitability. However, this still leaves a gap to be explained. There is a very large body of research seeking to explain this, summarised particularly by the IPCC.In essence, the reasons are to do with other market failures in addition to the core carbon externality market failure. These could be informational, incentive based and so on. One possible means of reconciling the CGE and MAC approaches would be to explicitly simulate these other market failures within the CGE model. Doing this would allow the incorporation of economywide effects which are not included in the MAC curves. For example, implementing abatement measures will divert resources from other activities, even if they are privately profitable. This will tend to arise costs in other industries, so the economy wide effect will be more costly than the single sector effect. Whether this would offset the negative cost is unclear, but it will at least reduce its magnitude. They may also be second round emission effects to take into account (increases in emissions elsewhere in the economy for the manufacture etc of items to reduce emissions in particular industries). Again, the overall magnitude of this is not clear, but the net effect would be to increase the economywide cost of abatement per tonne.It is also worth noting that there are very mixed research findings on the question of the net emissions effect of energy efficiency measures. Some authors find a considerable ‘rebound’ effect whereby initial efficiency increases actually results in subsequent increases in emissions. Recent analysis suggests that this rebound effect could be very large (even 100 percent). To some degree, economy wide models include a rebound effect and this could explain some of the difference. (A rebound effect will tend to increase the cost of a unit of abatement).The importance of international abatementThe results from the various studies indicate very clearly the importance of international abatement as part of Australia’s ability to meet particular abatement targets.Chart 3.11 illustrates the share of international abatement in some of the studies, with up to 70 per cent of abatement coming from international sources in some cases. STYLEREF 1 \s 3. SEQ Item \* ARABIC \s 1 11Proportion of abatement undertaken through international purchasesData source: SGLP chart 5.2. ALPF charts 6.2 and 6.4.Australia versus world abatement cost curvesChart 3.12 further illustrates one way of looking at the Treasury results. It plots world and Australian abatement (defined as reduction in emissions relative to BAU) against the (marginal) abatement cost for four scenarios (two domestic and two international) covered in the modelling. These curves show net abatement costs after all economic adjustments in the model scenarios have taken place. The ‘world’ results refer to all countries covered in the Treasury modelling (and so also includes Australia).Chart 3.12 illustrates that for these sets of simulations (medium versus ambitious abatement scenarios) the Australian abatement cost curve is clearly higher than the world abatement cost curve. The shape of the curves traced out in chart 3.7, as well as their positions, clearly depend on the particulars of the model simulations. The results are strongly indicative however, that Australian abatement costs are high relative to an average of world abatement costs. IF 1 = 1 " IF STYLEREF "Heading 6" \l \n 0 = 0 " IF STYLEREF "Heading 1" \l \n 3 = 0 "" " STYLEREF "Heading 1" \l \n 3." 3." " STYLEREF "Heading 6" \l \n A." 3. SEQ Item \* ARABIC \s IF STYLEREF "Heading 6" \l \n 0 = 0 " IF STYLEREF "Heading 1" \l \n 3 = 0 "" "1" 1" "6" 1 12" "" 3.12Comparative cost of abatement: Australia versus worldNote: Linear cost lines are: fitted from model data; indicative only; and have been extended beyond the data points for illustration. As the data points illustrate, cost curves are unlikely to be linear over the full range of abatement.Data source: CIE derivation from CCMS charts 2.4, 2.6, 3.1 and 3.6.Implications of Australian versus international abatement outcomesChart 3.13 further illustrates this outcome (higher Australian cost of abatement over the relevant range) by looking at the relationship between domestic abatement, international purchases and total abatement that arises in the results from the different Treasury reports.For example, in the 2008 report, the global abatement cost in 2020 was US$33/t, domestic Australian abatement was 189 Mt, total abatement was 249 Mt, and the difference — international purchases — was 60 Mt. According to the 2008 report, these purchases were from a range of developing countries.In the 2011 report, for a similar global abatement cost in 2020, domestic Australian abatement was 58 Mt, total abatement was 152 Mt, and the difference — international purchases — was 94 Mt. In each case at the indicative world cost, it was more cost effective for Australia to purchase international units rather than undertake domestic abatement. This clearly implies that the Australian cost of abatement, in the relevant abatement range, is higher than the combined cost of abatement for other countries.The same result continues to hold in the third Treasury analysis. In this case, in both 2020 and 2030, there is a clear range across which the Australian cost of abatement is higher than the international cost of abatement. The most stark result is in 2030 where total abatement is 392 Mt and international purchases are 236 Mt (60 per cent of total abatement). IF 1 = 1 " IF STYLEREF "Heading 6" \l \n 0 = 0 " IF STYLEREF "Heading 1" \l \n 3 = 0 "" " STYLEREF "Heading 1" \l \n 3." 3." " STYLEREF "Heading 6" \l \n A." 3. SEQ Item \* ARABIC \s IF STYLEREF "Heading 6" \l \n 0 = 0 " IF STYLEREF "Heading 1" \l \n 3 = 0 "" "1" 1" "6" 1 13" "" 3.13Domestic versus purchased abatement in Treasury modellingData source: CIE derivation from SGLP chart 5.2. ALPF charts 6.2 and 6.4, CCMS charts 2.9 and 3.6.The use of international abatement has the effect of lowering the cost of particular Australia targets compared with the cost of achieving the targets through domestic action alone.The discussion in international abatement cost sensitivities in Chapter 5 considers the implications of changes in the availability of international permits for the Australia’s cost of abatement.The cost of international action to AustraliaA feature of the G-Cubed modelling results (both the studies undertaken by DFAT and the more recent study by Liu et al) is the very significant impact of other country abatement on Australian GNP.For example, in the Liu et al study, 1.9 per cent of the 2.6 per cent reduction in GDP (in 2030, as a result of Paris Agreement commitments) still occurs even if Australia does not participate. That is, around 70 per cent of the cost to Australia of a particular global abatement scenario where Australia participates is a consequence of the flow on effects on Australia of changed trade patterns as a result of other country abatement.In contrast, analysis as part of ALPF indicated that the cost of international action was only between 11 and 18 per cent of the total cost of the abatement scenario (ALPF Box 6.2).This difference between models appears to relate to the much higher investment response in the G-Cubed model which for a give change in terms of trade means a greater reduction in investment and therefore GDP. This high investment response is a function of key features of G-Cubed including: forward looking behaviour for a proportion of investors; explicit treatment of international capital flows; and modelling of nominal aspects of the macroeconomy.While this response is not as strong in all models, it illustrates an important mechanism of costs for the Australian economy.Structural change and adjustmentThe economywide modelling studies, because of their sectoral detail, are able to provide information on structural adjustment that takes place as a consequence of abatement and abatement costs along with the effects of abatement overseas. Chart 3.14 illustrates a typical pattern of change in the broad sectoral composition of the Australian economy — this is a broad pattern seen in a number of different studies. Abatement leads to a reduction in construction, mining and services and to an expansion in some manufacturing and renewables activities. IF 1 = 1 " IF STYLEREF "Heading 6" \l \n 0 = 0 " IF STYLEREF "Heading 1" \l \n 3 = 0 "" " STYLEREF "Heading 1" \l \n 3." 3." " STYLEREF "Heading 6" \l \n A." 3. SEQ Item \* ARABIC \s IF STYLEREF "Heading 6" \l \n 0 = 0 " IF STYLEREF "Heading 1" \l \n 3 = 0 "" "1" 1" "6" 1 14" "" 3.14Structural change illustratedData source: CIE estimates based on Treasury simulations.The reasons underlying these changes vary by sector. Key features are as follows:The mining industry contracts as a result of a reduction in coal production, due to reduced demand both in Australia and overseas. Oil production also declines as does non-ferrous metal ore mining. The later result is the effect of reduced aluminium production (which is itself energy intensive).At the same time, there is a slight increase in mining due to an increase in iron ore production. This is explained by both a real exchange rate effect as well as a change in export demand.Energy intensive manufacturing industries also decline, along with manufacturing associated with mineral products. Some manufacturing, including food production and textiles clothing and footwear expands, again as a results of a real exchange rate effect. The electricity sector contracts due to the demand effects of an increase in prices, however the renewables component of the electricity sector expands as a direct result of the carbon policy.The construction sector contracts because it is indirectly very carbon intensive, particularly through the use of cement and concrete and other energy intensive materials. The construction sector also contracts as a result of the general reduction in activity (reduction in GDP) from the mitigation policy.The services sector contracts for a similar reason — the reduction in household incomes and general economic activity.Finally, the transport sector declines due to an increase in input costs (fuel) along with a general reduction in economic activity.Note that the modelling results are most accurate when they are concerned with measuring changes in existing activities (activities that are in the model database). They are less reliable, or require careful consideration, when abatement is associated with the growth in completely new activities or activities that are not represented in the model database.Chart 3.15 illustrates another aspect of structural change by looking at the distribution of changes in industry output (across roughly fifty individual industries, in percentage terms relative to the baseline) for each set of Treasury simulations. A small number of industries have a very large change in output (either positive or negative), while the majority of industries have much smaller changes. IF 1 = 1 " IF STYLEREF "Heading 6" \l \n 0 = 0 " IF STYLEREF "Heading 1" \l \n 3 = 0 "" " STYLEREF "Heading 1" \l \n 3." 3." " STYLEREF "Heading 6" \l \n A." 3. SEQ Item \* ARABIC \s IF STYLEREF "Heading 6" \l \n 0 = 0 " IF STYLEREF "Heading 1" \l \n 3 = 0 "" "1" 1" "6" 1 15" "" 3.15Structural changeData source: Treasury reportsTypical results with more industry detailWhile not published in as much detail in many studies, the results reported in SGLP provide a good representation of the more detailed sectoral findings, particularly in terms of the ranking of output changes (relative to baseline) by sector. Changes in output for declining and growing industries are presented in charts 3.16 and 3.17 respectively. IF 1 = 1 " IF STYLEREF "Heading 6" \l \n 0 = 0 " IF STYLEREF "Heading 1" \l \n 3 = 0 "" " STYLEREF "Heading 1" \l \n 3." 3." " STYLEREF "Heading 6" \l \n A." 3. SEQ Item \* ARABIC \s IF STYLEREF "Heading 6" \l \n 0 = 0 " IF STYLEREF "Heading 1" \l \n 3 = 0 "" "1" 1" "6" 1 16" "" 3.16Declining Australian industries in 2050 relative to baselineNote: Aggregate abatement in this scenario is 68 per cent relative to BAUData source: SGLP, High price scenario IF 1 = 1 " IF STYLEREF "Heading 6" \l \n 0 = 0 " IF STYLEREF "Heading 1" \l \n 3 = 0 "" " STYLEREF "Heading 1" \l \n 3." 3." " STYLEREF "Heading 6" \l \n A." 3. SEQ Item \* ARABIC \s IF STYLEREF "Heading 6" \l \n 0 = 0 " IF STYLEREF "Heading 1" \l \n 3 = 0 "" "1" 1" "6" 1 17" "" 3.17Expanding Australian industries in 2050 relative to baseline Note: Chart excludes gas fired electricity which expands by 165 per cent and other (renewable) electricity which expands by 443 per centData source: SGLP High price scenarioNote that these are all relative changes compared with the underlying baseline. The declining industries continue to grow over time in an absolute sense with the exception of oil, aluminium and coal fired electricity which decline in absolute terms relative to 2010. (Iron and steel production falls relative to baseline but increases slightly relative to 2010). Note that industries in these simulations decline despite the increase in use of renewable energy because of the overall net increase in the price of energy.The relative declines by industry largely reflect either policy intent (in the case of the large decline in coal fired electricity output) or the energy intensity of the relevant sector — seen most starkly, for example, in aluminium, alumina and iron and steel. As noted previously, the changes in some services are largely the result of reductions in income.In terms of changes in the share of employment, the SGLP results have the following key features:The share of agriculture in total employment doubles (from 2.4 to 4.3 per cent). This is a reversal of historical trends. It also depends crucially on particular assumptions concerning the coverage of agriculture under the abatement policies simulated.The share of services employment increases (from 80 to 83 per cent). This is the continuation of an ongoing trend.The share of manufacturing employment declines (6.5 per cent to 4.4 per cent). This is the acceleration of an existing trend. The share of employment in mining declines (from 1.7 to 1.1 per cent). This is a reversal of the trend in recent years.The share of employment in construction declines from 9.4 per cent to 7.1 per cent. This is a reversal of recent trends. While most of these results closely relate to energy or emissions intensity, the relatively large decline in construction is interesting as construction is not directly emissions intensive, but it is an intensive user of other emissions and energy intensive inputs. Put another way, construction is a large consumers of emissions (rather than a producer, which is the case for most other industries). This is further illustrated in chart 3.18 which summarises results from a recent ABS comparison of production and consumption-based emissions by broad sector. Consumption based emissions in the construction sector are 7 times larger than production emissions for that same sector. Thus, while construction is a minor emitter in a production basis, it is a major source of embodied emissions when measured on a consumption basis.The same is true for the commercial and services sector (which consists of wholesale and retail trade, telecommunications, and a range of services including financial and real estate) and the manufacturing sector.In contrast, consumption emissions are much lower than production emissions for the agriculture, mining, electricity and transport sectors. IF 1 = 1 " IF STYLEREF "Heading 6" \l \n 0 = 0 " IF STYLEREF "Heading 1" \l \n 3 = 0 "" " STYLEREF "Heading 1" \l \n 3." 3." " STYLEREF "Heading 6" \l \n A." 3. SEQ Item \* ARABIC \s IF STYLEREF "Heading 6" \l \n 0 = 0 " IF STYLEREF "Heading 1" \l \n 3 = 0 "" "1" 1" "6" 1 18" "" 3.18Ratio of consumption emissions to production emissions by broad sector Data source: CIE estimates based on Australian Bureau of Statistics Linking Greenhouse Gas Emissions with Economic Production and Consumption in Australian Environmental-Economic Accounts 2014, ABS Catalogue Number 4655.0, 3 April 2014.Other studiesThese broad patterns of structural change are common across a range of studies, although some of the specific details vary because of particular assumptions used in each study. For example:The Fisher (BAEconomics) study projects declines in most sectors, and in particular:A sharp reduction in coal and other mining output, along with an increase in iron ore production — similar to the analysis summarised above and for the same broad reasons.A reduction in all agricultural output and a reduction in the food processing industry. This is different to the summary above because of the coverage of agriculture within the policies simulated.A reduction in most manufacturing activities — similar to the analysis summarised above and again related to the emissions intensity of these activities.A reduction in electricity and all forms of transport — as expected from the analysis above.A small reduction in services — largely an income effect as noted above.A reduction in construction, again because of the relative consumption emissions intensity of construction.The McKibbin 2015 analysis, when looking at Australian abatement projects declines in all industry groups, in particular:The largest percentage reduction in coal and electric utilities, followed by gas extraction — because of their emissions intensity.Similar reductions for transport, services, other mining and non-durable manufacturing.The smallest reductions for agriculture and durable manufacturing.The McKibbin 2015 analysis looking at international abatement projects:The largest changes in coal mining outputRelatively small, or zero, changes in most other sectors.Other forms of structural changeAs well as changing the composition of the economy by changing the relative output of existing industries, it is likely that abatement policy will lead to the development of new structural relationships between industries that are not captured in existing databases.For example, the development of electric vehicles or large solar value chains may result in a new set of economic relationships. To date, it is not clear that this form of structural change is captured in existing models.Adjustment costsStructural adjustment is in some ways a consequence of the overall cost of abatement. It does not necessarily imply an increase in costs unless there are explicit adjustment costs associated with the shift of resources from one sector to another. Adjustment costs in this sense are defined as the real resource costs associated with shifting activity between sectors, or in some cases within a sector.The GTEM and MMRF models do not explicitly account for adjustment costs that take place as a consequence of the structural change outlined above. The potential importance of adjustment costs can be seen by comparing outcomes between models with and without adjustment costs.Chart 3.19 compares the cost of abatement for the three different models used in the Treasury analysis. These costs are the reduction in GNP per capita for a single year (2020) relative to the reference scenario, and are expressed as a proportion of the reduction in emissions that year (relative to the reference scenario). The MMRF model, for example, implies a 0.054 per cent reduction in GNP per capita for each per cent reduction in emissions (each in 2020). It is interesting to note that the standardised cost for G-Cubed are just under two times higher than for the other two models. However, G-Cubed’s costs are only higher for the 2020 comparison, not for 2050. As G-Cubed is a model with adjustment costs, this implies that at least some of these higher costs are due to the adjustment costs associated with the mitigation objectives. The published G-Cubed results don’t allow the isolation of adjustment cost effects.Of course, there are other differences between G-Cubed and the other two models. However, the large difference, along with the different modelling of adjustment costs, is suggestive of the importance of adjustment costs in designing mitigation policies. STYLEREF 1 \s 3. SEQ Item \* ARABIC \s 1 19Comparative cost of abatement in 2020 for three models Data source: CIE estimates based on Treasury Australia’s Low Pollution Future, table 6.4.Practical implication for policy and modellingAn important practical implication of this structural adjustment discussion is to note that these results depend very much on the effect of an increase in the price of energy. This in turn depends on the future cost of energy under abatement options. As this is not known with any certainty, there is likely to be a number of scenarios for structural adjustment as mitigation proceeds.Relative merits of different policy instrumentsThe different scenarios reported in the Treasury/Garnaut modelling (published in ALPF) contain different concentration targets, different degrees of market flexibility and different coverage under emissions trading. Chart 3.20 summarises the overall cost of abatement (between 2010 and 2050) for each unit of cumulative abatement. This is one way of summarising in a single measure the different time paths of the various scenarios.Chart 3.15 shows for example that the cost of CPRS -5 is $42 per tonne of abatement. Not surprisingly, the scenarios with lower concentration targets (that is, more abatement) are more costly. What is interesting is that the Garnaut scenarios, which generally have broader coverage and less restrictions to trade, have lower costs of abatement than the CPRS scenarios. Indeed, the Garnaut -25 scenario, with a target of 450ppm, is less costly per unit that the CPRS -15 scenario with a target of 510ppm. STYLEREF 1 \s 3. SEQ Item \* ARABIC \s 1 20‘Levelised cost’ of abatement under different scenarios?Present value of GDP loss per unit of cumulative abatement 2010 to 2050Data source: CIE estimates based on Treasury results.This confirms to an extent the idea that flexible markets provide for less costly abatement than do policies with various restrictions to trade. This outcome is a general finding of a number of different modelling studies, including in the international literature. Indeed, the finding that broader coverage in the presence of flexible markets lead to maximum cost effectiveness is now, quite literally, a ‘textbook’ position and change in the energy marketChart 3.21 illustrates compositional change in the electricity market as abatement increases. The outcomes show a steady decline in black and brown coal as abatement increases, along with a steady increase in geothermal and a very sharp increase in solar. IF 1 = 1 " IF STYLEREF "Heading 6" \l \n 0 = 0 " IF STYLEREF "Heading 1" \l \n 3 = 0 "" " STYLEREF "Heading 1" \l \n 3." 3." " STYLEREF "Heading 6" \l \n A." 3. SEQ Item \* ARABIC \s IF STYLEREF "Heading 6" \l \n 0 = 0 " IF STYLEREF "Heading 1" \l \n 3 = 0 "" "1" 1" "6" 1 21" "" 3.21Compositional change in the electricity market Data source: ACIL ALLENThe exact nature of compositional change in the electricity market depends on the particular model used and, importantly, the assumed future price path for each of the renewable technologies. For example, chart 3.22 shows the evolution of black coal shares for three different energy models (SKM, ROAM and ACIL ALLEN). The timing of the decline of black coal varies significantly between these three. IF 1 = 1 " IF STYLEREF "Heading 6" \l \n 0 = 0 " IF STYLEREF "Heading 1" \l \n 3 = 0 "" " STYLEREF "Heading 1" \l \n 3." 3." " STYLEREF "Heading 6" \l \n A." 3. SEQ Item \* ARABIC \s IF STYLEREF "Heading 6" \l \n 0 = 0 " IF STYLEREF "Heading 1" \l \n 3 = 0 "" "1" 1" "6" 1 22" "" 3.22Black coal shares and abatement Data source: ALPF, CCMS, ACIL ALLENFinally, chart 3.23 shows a different pattern of renewable growth between the three models, with substantive differences between geothermal and solar output.This pattern of differences is related to underlying model assumptions, particularly future technology cost profiles. IF 1 = 1 " IF STYLEREF "Heading 6" \l \n 0 = 0 " IF STYLEREF "Heading 1" \l \n 3 = 0 "" " STYLEREF "Heading 1" \l \n 3." 3." " STYLEREF "Heading 6" \l \n A." 3. SEQ Item \* ARABIC \s IF STYLEREF "Heading 6" \l \n 0 = 0 " IF STYLEREF "Heading 1" \l \n 3 = 0 "" "1" 1" "6" 1 23" "" 3.23Different pattern of renewables Data source: ALPF, CCMS, ACIL ALLEN.Energy price changes Changes in electricity prices as a consequence of abatement are determined using electricity specific models.Chart 3.24 illustrates a typical profile of price changes (wholesale, retail and industrial) as abatement increases. Prices initially increase rapidly as abatement increases, but after abatement of around 30 per cent, the price increase stabilises somewhat. As new investment is put in place at higher levels of abatement, the stabilised price effect is determined by the cost of a mix of low emissions new entrants. This mix varies by state but is a mix of solar, gas fired, geothermal and black coal with CCS. IF 1 = 1 " IF STYLEREF "Heading 6" \l \n 0 = 0 " IF STYLEREF "Heading 1" \l \n 3 = 0 "" " STYLEREF "Heading 1" \l \n 3." 3." " STYLEREF "Heading 6" \l \n A." 3. SEQ Item \* ARABIC \s IF STYLEREF "Heading 6" \l \n 0 = 0 " IF STYLEREF "Heading 1" \l \n 3 = 0 "" "1" 1" "6" 1 24" "" 3.24Electricity prices and abatement Data source: ACIL-ALLENSources of model input informationConfidence in the cost of abatement estimates depends in part on the models providing:A reasonable representation of the current structure of the global and Australian economiesA believable pathway for the evolution of these economiesA credible pathway for the development of technology, particular in the energy sector.Data for underlying baseline constructionEach model framework has its own set of data requirements and sources. In general this data is sourced from publicly available information and in the case of larger models often contributed to by a large number of researchers.The GTEM database is not publicly available, however it is derived from the database for the GTAP model. The underlying GTAP base data comes from an international consortium of contributors. Effectively, a number of organisations contribute individual country databases (input-output tables or social accounting matrices, SAM) which are then combined and balanced using a number of methods by the overall GTAP organisation. Regular data releases are made and overall the construction of the database is transparent and subject to ongoing review and feedback. The latest GTAP database is version 9.0 which has 2011 as the latest reference year.The MMRF database is based largely around the Australian input-output tables with a variety of modifications and additions. The latest Australian input-output tables are for 2015-16, although MMRF is based on an earlier set of tables.Individual electricity model databases come from a variety of sources but are essentially designed to cover underlying generation in Australia. Information on existing and committed generators comes from a variety of sources including AEMO and individual generator sources. The starting point for new renewables or other new generation are estimates of current construction costs. Estimates are available from a number of sources, including the IPCC and international groups such as Lazard. Cost information is also available from the CSIRO, in particular CSIROs regular work on electricity generation technology cost projections.Data for simulationsSimulations consist of a baseline and policy simulations. There are a large number of assumptions that are used for both the baseline and for responses in the policy simulations.Future productivity, economic growth and populationAssumptions made for the construction of the international and Australian economic baseline include:Australian population and labour force participation projectionsGlobal population and labour force participation projectionsAustralian productivity projections Global productivity projectionsTerms of tradeEnergy commodity pricesFuel cost for electricity generationIntermediate input efficiencyHousehold taste shiftsEnergy efficiency: economywide, sector specificElectricity technology assumptionsThe choice of values for the majority of these sets of assumptions are either based on historical trends or come from statistical agencies (such as the UN in the case of population projections). In some cases, projections are made consistent with other government publications such as the Budget Papers. Electricity generation cost evolutionThe future cost path of electricity generation technologies, particularly renewables, is clearly a crucial assumption in the energy sector specific models.For Australia, this cost information has been obtained from a number of sources including:The 2012 and 2013 Australian Energy Technology Assessments, published by the Bureau of Energy and Resource EconomicsThe 2015 Australian Power Generation Technology Report published by the CO2CRCThe 2017 Electricity generation technology cost projections published by CSIRO.Key parameter choicesAll models have a wide set of behavioural parameters. These include:Primary factor substitution elasticitiesConsumer demand parameters (demand response to income and price changes)International trade elasticities (substitution between different country sources as a result of price changes)Investment parameters (relating changes in investment to changes in rate of return)These are generally associated with the original development of the model. Values of these parameters are generally not explicitly reported.Dealing with uncertaintyAll of these sets of assumptions and parameters are subject to considerably uncertainty. In terms of economic growth and related assumptions, forecasting 20 or 30 years (or more) into the future is extremely difficult and the underlying assumptions used are subject to considerable variation. Different modellers often make different assumptions and there is generally no simple basis to judge between them.Underlying model parameters are often sourced from statistical or econometric analysis which is, of course, subject to a number of forms of uncertainty. While models use a single point estimate for parameters within simulations, strictly speaking parameters can only every be estimated as a range, or with a particular confidence interval.Most of the studies covered here use a single or limited set of exogenous assumptions, and generally do not vary underlying parameter estimates. As note further below, there are techniques available to further enhance the modelling by exploring a wider range using systematic sensitivity analysis.Dependence of results on underlying assumptionsPrice pathway sensitivitiesAs noted above, to estimate the global carbon price (given assumptions about targets and country participation), a typical model procedure is to impose an annual price increase (set at the real interest rate, for example) and to then find the starting carbon price (and by implication price path) that leads to the required global abatement.Implicit in the solution to this problem is the abatement cost of all participants in global abatement. Higher cost of abatement requires an initial higher price, and vice versa. Importantly, it is the entire future path for the cost of abatement that determines the starting price — and therefore the price path, and therefore abatement in each year. For example, if future renewable costs are not expected to decline, then the global carbon price required to achieve a particular target will be relatively high as a larger price will be required to induce a switch to renewable sources of energy. In contrast, if renewable prices are expected to decline substantively in the future, then the global carbon price required to achieve abatement will be lower.In this way, the optimal price path for emissions depends crucially on the projected cost path for low carbon energy sources.The estimated price path feeds into all the subsequent results from the modelling, so it is important to understand how sensitive the price path is to changes in underlying model parameters and assumptions including:the ‘global’ cost of abatement;the composition of participation in global abatement;underlying BAU growth of emissions.None of these factors is known with certainty. Without access to the GTEM model it is difficult to estimate how sensitive the price path is to these assumptions, but some insight can be gained from a simple ‘back of the envelope’ reproduction of Treasury’s price path, and then using this to understand the sensitivity of prices to changes in the aggregate costs of abatement at various points in time.Approximate sensitivities of global price to changes in the cost of abatement are illustrated in chart 5.1. This chart shows, for example, that if the cost of abatement is varied in all years (the last column of the chart) the ‘elasticity’ of the global carbon price with respect to the cost of abatement is around 2. That is, a 1 per cent increase in the cost of abatement in each year would lead to a 2 per cent increase in the global carbon price in each year. STYLEREF 1 \s 5. SEQ Item \* ARABIC \s 1 1Sensitivity of carbon price to changes in cost of abatement a a Results expressed as an ‘elasticity’, that is, percentage change in price for a 1 per cent change in cost of abatement.Data source: CIE estimates based on results from SGLP.A key result that emerges from this is that the future cost of abatement has a much greater effect on the price path than the near term cost of abatement. This is important as the future cost of abatement is considerably more uncertain than the near term cost of abatement.Further, the future cost of abatement is amenable to policy interventions, particularly R&D style interventions. In all, what this relatively simple sensitivity analysis suggests is that risks (in terms of an uncertain price path) are mostly driven by unknown costs in the future and that there is an unknown (from the modelling reported) relationship between R&D actions today and the cost of abatement in the future. International abatement cost sensitivitiesSensitivity analysis: counting domestic abatement costs onlyIt is interesting to consider how much higher than international costs Australian costs are. There is no single answer to this question as it depends on the particulars of the scenario modelled. However, combining the results from a number of different simulations provides an indication of a composite cost of abatement curve for Australia. This is illustrated in chart 5.2. Note that this composite curve contains a large number of factors within it, but it nevertheless provides a good summary of the results implied by the most recent Treasury modelling. This chart (in particular, the line of fit) shows the cost (in $/t) of different levels of abatement. Reading from this chart gives an indication of the marginal cost for any particular abatement target. The fitted line shown in chart 5.2 indicates that:(Cost of abatement) = 0.41*(quantity of abatement).Comparing this with the abatement gaps noted in chart 3.8 (along with other gaps reported for other scenarios in CCMS), it is possible to infer how much higher the Australian cost of abatement is compared with the international cost at a particular point in the abatement cost curve. This difference ranges from 40 per cent to 750 per cent. (The very large difference is in the low price scenario). For example, in the case of the central policy scenario in 2030, total abatement is 392 Mt (bottom right panel of chart 3.8). For this to be undertaken domestically, the marginal cost for the last unit of abatement would be around $160/t (392*0.41). This is nearly 3 times the international cost at that point in the relevant Treasury simulation. Using this same comparison, the total cost of abatement using Australian costs would be 70 per cent higher than the total cost of abatement with a combination of Australian and global costs.While there are different ways of constructing this calculation, they all give broadly the same result. IF 1 = 1 " IF STYLEREF "Heading 6" \l \n 0 = 0 " IF STYLEREF "Heading 1" \l \n 5 = 0 "" " STYLEREF "Heading 1" \l \n 5." 5." " STYLEREF "Heading 6" \l \n A." 5. SEQ Item \* ARABIC \s IF STYLEREF "Heading 6" \l \n 0 = 0 " IF STYLEREF "Heading 1" \l \n 5 = 0 "" "1" 1" "6" 1 2" "" 5.2Australian marginal cost of abatement curve implied by 2013 Treasury analysisNote: Plotted points show domestic abatement outcomes and prices for central policy scenario, low price scenario and high price scenario for each of 2020 and 2030. The slope coefficient for the fitted line is 0.41 with a standard error of 0.02. Data source: Derived from CCMS charts 3.1 and 3.6.Which countries?Treasury analysis in SGLP suggested that international abatement would come largely from India and Asia and a range of other countries by 2050 (see chart 5.3). STYLEREF 1 \s 5. SEQ Item \* ARABIC \s 1 3Sale of abatement a a Regions in the GTEM model are defined in SGLP table A1. Other Asia here refers to Other South and East Asia and includes Brunei, Malaysia, Philippines, Thailand, Cambodia, Maldives, Korea, Timor-Leste, Laos, Myanmar, Singapore and Vietnam.Data source: SGLP table 3.8It is interesting to consider how sensitive Australian abatement cost results are to changes in cost of abatement in different countries (as well as to the changing composition of policies in different countries) and to any restrictions in abatement trade between countries. That is, given that the overall results for Australia depend on buying international permits, what are the consequence of these permits being less available, or higher cost?Without access to the original model, it is difficult to undertake this analysis. However, postulating a simple linear model of abatement and trade in abatement provides some useful insights.A couple of points emerge from this simple model.First, the world carbon price (and by implication the outcomes for Australia through the need to purchase abatement) is most sensitive to the marginal cost of abatement of the lowest cost abating country. Thus, in terms of sensitivity analysis, the countries we need to be concerned with understanding are those expected to sell abatement (by implication, these are the lowest cost abating countries).An implication of this is that under the hypothesis of international trade in abatement, Australian outcomes may be more sensitive to uncertain costs in China, India and South East Asia than they are to the Australian cost of abatement.Put another way, while considerable effort has gone into the cost of abatement in Australia’s electricity sector, this simple illustration suggests that outcomes for Australia may in fact be more sensitive to future costs in China, India and South East Asia. This implies that some importance should be attached to understanding future costs of abatement in those countries.Assessing risk and uncertainty in a world where Australia intends to purchase international permits therefore requires some understanding of the cost of abatement of those countries that are presumed to sell abatement on the international market.The question that naturally arises is the empirical base of the cost of abatement for net selling countries. This is a function of a number of aspects of the GTEM model including a variety of parameter choices. It is unlikely that any of these is known with certainty — particularly given that these are countries where data reliability is an issue.The G-Cubed sensitivityTo further understand the importance of assumptions about international abatement, in 2011 the CIE undertook and analysis, using the G-Cubed model, of the implications of restrictions to the availability of international abatement. This involved the comparison of a baseline simulation which included all international abatement with a simulation in which some countries were excluded from the sale of international permits.To construct the base simulation, we replicated as closely as possible the principles embedded in the Treasury modelling for SGLP, especially as far international action is concerned. In particular, the base simulation (and the model baseline underlying it) built on previous analysis of the Copenhagen Commitments. It essentially involved simulating the effect of global action on climate change by imposing the carbon price (implemented through emissions trading) necessary to achieve the abatement targets.In undertaking the underlying base simulation, we found the global carbon price path using the same ‘Hotelling rule’ adopted by Treasury (that is, the global price path is chosen in order to achieve the abatement target, assuming that the price increases by 4 per cent real a year). In this base simulation, by 2020, international trade in abatement was around 30 per cent of the global abatement task (consistent with Treasury findings). In contrast with the Treasury results, the major net sellers of permits in the G-Cubed model were India, Eastern Europe and the Former Soviet Union (EEB) and other LDCs. In the G-Cubed results China was not a net seller of abatement (where as it provided around 30 per cent of abatement sales in 2020 in the Treasury results). This is a consequence of different baseline projections.Chart 5.4 shows results for carbon prices under the base and sensitivity simulations. Carbon prices are expressed as an index, with the base simulation price in 2016 set at 100. Under the base simulation, the carbon price increases by 4 per cent a year in line with the Hotelling rule assumptions adopted by Treasury and also implemented in the G-Cubed model.With the exclusion of LDCs, the carbon price increases by 18 per cent in each year — the increase in the carbon price relative to the base simulation is the same in each year because of the assumption of constant growth in carbon prices over time. The exclusion of EEB and India in addition to the LDCs leads to an increase in the carbon price (relative to the core simulation) of 47 per cent. This confirms the idea that removing low cost sellers of permits increases the carbon price and that the pathway of permit prices is clearly sensitive to country coverage. IF 1 = 1 " IF STYLEREF "Heading 6" \l \n 0 = 0 " IF STYLEREF "Heading 1" \l \n 5 = 0 "" " STYLEREF "Heading 1" \l \n 5." 5." " STYLEREF "Heading 6" \l \n A." 5. SEQ Item \* ARABIC \s IF STYLEREF "Heading 6" \l \n 0 = 0 " IF STYLEREF "Heading 1" \l \n 5 = 0 "" "1" 1" "6" 1 4" "" 5.4Effects on abatement prices removing sellers of abatement Note: LDC refers to other less develop countries and EEB refers to Eastern Europe and the Former Soviet Union.Data source: CIE 2011 The increase in abatement costs led to higher costs for Australia, with the GDP loss from abatement is 36 per cent higher following the exclusion of some countries from providing abatement to Australia.This is also broadly consistent with the sensitivity analysis presented in McKibbin’s Report 2 to DFAT where allowing international permits lowers the overall cost by around 40 per cent.Technology sensitivitiesThe modelling comparisons reported by the IPCC have confirmed that the key factor determining modelling outcomes is the assumed availability and flexibility of low carbon technologies to substitute for fossil fuel energy.In addition, the availability (or otherwise) of CCS is the factor associated with the largest effect on the overall cost of abatement. A recent report notes that — globally — assuming a limited supply of bioenergy or CCS leads to mitigations costs (in present value terms) between 64 and 138 per cent higher than in baseline scenarios with unlimited amounts of bioenergy or CCS. In contrast, limited solar, wind or nuclear energy results in a corresponding 6 to 7 per cent increase.A number of the studies covered report some degree of technology sensitivity analysis and are broadly consistent with the international research. ALPF presented some important results, many of which are consistent with subsequent studies (these are summarised in table 5.5). IF 1 = 1 " IF STYLEREF "Heading 6" \l \n 0 = 0 " IF STYLEREF "Heading 1" \l \n 5 = 0 "" " STYLEREF "Heading 1" \l \n 5." 5." "B." 5. SEQ Item \* ARABIC \s IF STYLEREF "Heading 6" \l \n 0 = 0 " IF STYLEREF "Heading 1" \l \n 5 = 0 "" "1" 1" "6" 1 5" "" 5.5Technology sensitivities. Changes in deviation of GNP from baseline202020502100%%%Base scenario (Garnaut -10)-1.3-4.8-7.1Improved CCS technology-1.2-4.4-4.4Extra energy-efficiency improvements-1.0-3.9-6.5Higher learning rates-1.2-4.1-5.7Agricultural backstop-1.2-4.5-5.5Enhanced technology scenario – fully costed-1.0-3.6-4.3Enhanced technology scenario – partly costed-1.0-3.6-2.7Costed MAC (marginal abatement cost) curvesa-1.5-5.7-12.2No CCS technology-1.5-5.7-8.8Zero emission CCS technology-1.3-4.4-4.3a This refers to a change in the modelling treatment of the marginal cost of abatement for emissions produced from industrial activities. In the base treatment MACs are treated as input neutral (the benefits of lower emissions and input costs offset each other). In the sensitivity treatment, full cost impacts are accounted for.Source: ALPF table 6.9.In 2050, the most important single factors are the availability CCS technology and the assumption of actual costs when applied to the marginal abatement cost (MAC) curves for all sectors in the economy. (While the enhanced technology scenario has a larger impact, it represents a combination of other scenarios).The next most important factors in 2050 are those associated with either improved rates of energy efficiency, or with faster declines in the cost of renewables.By 2100, the most important single factor is the costing of MAC curves, followed by the CCS technology factors.The global importance of CCS is further reinforced in the international simulations undertaken as part of ALPF. Reproduced in table 5.6, two extreme assumptions (unavailability of CCS, left panel, versus efficient CCS, right panel) indicated dramatically different outcomes for the share of different generations sources in global electricity generation. IF 1 = 1 " IF STYLEREF "Heading 6" \l \n 0 = 0 " IF STYLEREF "Heading 1" \l \n 5 = 0 "" " STYLEREF "Heading 1" \l \n 5." 5." " STYLEREF "Heading 6" \l \n A." 5. SEQ Item \* ARABIC \s IF STYLEREF "Heading 6" \l \n 0 = 0 " IF STYLEREF "Heading 1" \l \n 5 = 0 "" "1" 1" "6" 1 6" "" 5.6Global technology shares in electricity generation, the role of CCSNo carbon capture and storageZero emissions carbon capture and storage Chart source: ALPF, chart 5.26G-Cubed sensitivitiesThe G-Cubed analysis prepared for DFAT contains some sensitivity analysis around technologies. Overall, the G-Cubed results show that a 27 per cent reduction in technology costs leads to around a 30 per cent reduction in economic costs of abatement.Electricity model simulationsThe electricity model simulations undertaken by ACIL ALLEN include a sensitivity analysis of the effects of not having either geothermal energy or CCS available in the new energy mix. This results in an increase in emissions of around 30 per cent relative to where they would be if geothermal and CCS were available. To achieve the same abatement, these results imply that an additional 10 per centage points of abatement are needed to achieve the same outcome. While ACIL ALLEN do not report the incremental costs of this, based on results presented earlier in this report, this could increase costs by up to 1 percentage point of GNP.Evolution of renewable costsRenewable cost profiles in recent modellingSince the beginning of the modelling studies around 2008, the cost of renewables has declined substantially, particularly in the case of solar. From a modelling outcome perspective, what is important is the projected decline in renewables over the model projection period. This particular pathway is crucial as it determines the cost of switching from emitting technologies to lower emitting ones. While the efficiency of coal and gas is also expected to continue to improve, in the absence of cheap CCS, the set up of the modelling ensures that there must be substitution away from fossil fuels (to achieve the specified targets) towards renewables. The cost profile of each renewable will determine which are chosen in the generation mix.Charts 5.7, 5.8 and 5.9 show the evolution of wind, solar and geothermal costs for each of the Treasury modelling studies. For with and solar (charts 5.7 and 5.8) this is also compared with the latest CSIRO Technology Road Map. Table 5.10 compares the rates of cost decline for these various technologies. IF 1 = 1 " IF STYLEREF "Heading 6" \l \n 0 = 0 " IF STYLEREF "Heading 1" \l \n 5 = 0 "" " STYLEREF "Heading 1" \l \n 5." 5." " STYLEREF "Heading 6" \l \n A." 5. SEQ Item \* ARABIC \s IF STYLEREF "Heading 6" \l \n 0 = 0 " IF STYLEREF "Heading 1" \l \n 5 = 0 "" "1" 1" "6" 1 7" "" 5.7Projected cost of wind: modelling studies and CSIRO projections Data source: Studies as cited IF 1 = 1 " IF STYLEREF "Heading 6" \l \n 0 = 0 " IF STYLEREF "Heading 1" \l \n 5 = 0 "" " STYLEREF "Heading 1" \l \n 5." 5." " STYLEREF "Heading 6" \l \n A." 5. SEQ Item \* ARABIC \s IF STYLEREF "Heading 6" \l \n 0 = 0 " IF STYLEREF "Heading 1" \l \n 5 = 0 "" "1" 1" "6" 1 8" "" 5.8Projected cost of solar: modelling studies and CSIRO projections Data source: Studies as cited IF 1 = 1 " IF STYLEREF "Heading 6" \l \n 0 = 0 " IF STYLEREF "Heading 1" \l \n 5 = 0 "" " STYLEREF "Heading 1" \l \n 5." 5." " STYLEREF "Heading 6" \l \n A." 5. SEQ Item \* ARABIC \s IF STYLEREF "Heading 6" \l \n 0 = 0 " IF STYLEREF "Heading 1" \l \n 5 = 0 "" "1" 1" "6" 1 9" "" 5.9Projected cost of geothermal: modelling studies Data source: Studies as cited. IF 1 = 1 " IF STYLEREF "Heading 6" \l \n 0 = 0 " IF STYLEREF "Heading 1" \l \n 5 = 0 "" " STYLEREF "Heading 1" \l \n 5." 5." "B." 5. SEQ Item \* ARABIC \s IF STYLEREF "Heading 6" \l \n 0 = 0 " IF STYLEREF "Heading 1" \l \n 5 = 0 "" "1" 1" "6" 1 10" "" 5.10Rate of cost decline to 2050 (per cent per year)ALPFSGLPCCMSCSIRO 2017CSIRO/AEMO 2018%%%%%Wind0.50.61.00.60.3Solar1.01.73.13.22.9Geothermal1.01.0-0.50.20.2Source: Studies as cited. CSIRO 2017 refers to Hayward and Graham Electricity generation technology cost projections 2017-2050. December 2017. CSIRO 2018 refers to Graham et al GenCost 2018, December 2018.Several points emerge when examining these comparisons.ALPF had geothermal as a low cost renewable source, with a steadily declining cost profile. Geothermal was in fact cheaper than solar for these simulations. The cost profile for geothermal was for more rapid reduction than CSIRO currently expects (table 5.10).ALPF underestimated the rate of solar cost decline, and slightly underestimated the rate of wind cost decline compared with CSIRO’s 2017 estimates, but slightly overestimated it compared with CSIRO’s 2018 estimates (table 5.10).SGLP saw a substantive change in the expected level of cost of geothermal (although the rate of decline was the same as for SGLP, and still an overestimate compared with CSIRO).SGLP also saw and increase in the rate of cost decline for wind, to a level consistent with current CSIRO 2017 projections, but higher than CSIRO 2018 projections.SGLP also saw in increase in the rate of cost decline for solar, but still substantially less than current MS saw a substantive increase in the expected rate of cost decline for wind, to a level higher than projected by CSIRO in either 2017 or MS substantively revised the expected cost of solar, to a level consistent with CSIRO.Finally, CCMS completely reversed expectations regarding geothermal. While starting with the same level of costs as SGLP, CCMS expected increases in costs over time. The upshot of all of this that while the early studies substantively underestimated the cost of solar decline, they had at the time in place an alternative low-cost renewable (geothermal). Across the simulations, geothermal and solar essentially switched places. Simulations today would have lower cost renewables but not by as large a margin as might be expected given recent observed changes in costs.Other recent renewable cost developmentsWhile there has been recognition for a while that renewables are becoming the lowest cost energy source based on new construction costs, another important recent development in renewable costs is that according to some estimates, the new build cost of some renewables is now cheaper than the marginal operating cost of coal and some other fossil fuels. This is important as it substantially reduces the opportunity cost of retiring coal-based generation before the full end of the economic life of the asset. If this development turns out to be well founded, then it is likely that new simulations would show a lower electricity price increase, and a lower overall cost of abatement to the economy.Integration costsHowever, these new cost developments may be offset to some extent by increasing concerns about the cost of integrating renewables into the existing electricity system. These costs arise because most renewable sources are intermittent (often referred to as variable renewable energy, VRE) and need to be balanced with ‘firm’ sources of generation.According to the IPCC (AR5 WGIII, Chapter 7) integration costs include:balancing costs, reflecting the need to maintain a balance between demand and supply;capacity adequacy costs required to ensure system operation at peak times; andtransmission and distribution costs, the incremental costs of distributing the additional renewable energy.The IPCC note that studies of high VRE penetration:…suggest that integrating significant RE generation technology is technically feasible, though economic and institutional barriers may hinder uptake. Integrating high penetrations of RE resources, particularly those that are intrinsically time variable, alongside operationally inflexible generation is expected to result in higher system-balancing costs. Compared to other mitigation options variable renewable generation will contribute less to capacity adequacy, and, if remote from loads, will also increase transmission costs. The determination of least-cost portfolios of those options that facilitate the integration of fluctuating power sources is a field of active and ongoing research (p. 534).Depending on the country and the particular scenario, these additional costs may be a significant proportion of the new construction costs for the relevant technology. As the IPCC note, this is a substantive area of research. A recent study undertaking a detailed literature review, while noting the high variability of results, made the following broad conclusions.Wind and solar integration costs are high if these technologies are deployed at large scale: in thermal systems, wind integration costs are about 25-35 €/MWh at 30-40% penetration, assuming a base price of 70 €/MWh. Integration costs are 35-50% of generation costs. As integration costs can be large in size, ignoring them in cost-benefit analyses or system optimization can strongly bias results. The size of integration costs depends on the power system and VRE penetration: integration costs can be negative at low (<10%) penetration, they generally increase with penetration, and are typically smaller in hydro than in thermal systems. System adaptations can significantly reduce integration costs.In thermal systems with high VRE shares, the utilization effect amounts to more than half of all integration costs. Maybe this is the most important finding of this study: the largest integration cost component is the reduction of utilization of the capital embodied in the power system. Most previous integration cost studies have not touched upon this effect. VRE-rich power systems require flexible thermal plants, but even more so they require plants that are low in capital costs. The same study also noted an important distinction between short term and long term integration costs:Integration costs depend on the properties of the legacy system: short-term integration costs are increased by a large stock of inflexible and capital-intensive base-load power plants, a scarce grid connection to regions with high renewable potentials and an inflexible electricity demand profile that hardly matches VRE supply. In contrast, over the long term, the power system can fully and optimally adapt to increased VRE volumes. These potential changes comprise operational routines and procedures, market design, increased flexibility of existing assets, a shift in the capacity mix, transmission grid extensions, a change in load patterns, demand-side management and technological innovations.This is also an area of ongoing research in Australia, with recent recognition of the need to account for variable integration costs when comparing the costs of different energy sources.The need for various forms of innovation to deal with integration costs has recently been explicitly considered by the International Renewable Energy Agency (IRENA). IRENA point out a range of substantial technical organisational developments that will be needed to successfully integrate variable renewable energy. Chart 5.11 summarises the very large range of options canvassed by IRENA. IF 1 = 1 " IF STYLEREF "Heading 6" \l \n 0 = 0 " IF STYLEREF "Heading 1" \l \n 5 = 0 "" " STYLEREF "Heading 1" \l \n 5." 5." " STYLEREF "Heading 6" \l \n A." 5. SEQ Item \* ARABIC \s IF STYLEREF "Heading 6" \l \n 0 = 0 " IF STYLEREF "Heading 1" \l \n 5 = 0 "" "1" 1" "6" 1 11" "" 5.11IRENA suggestions of innovations to deal with VRE integration issuesData source: IRENAEnergy modelling incorporating integration costsRecent energy system modelling commissioned by ANLEC R&D illustrates a range of scenario and planning issues associated with the high penetration of renewables.Boston et al 2018 argue that it is not possible to go beyond 65 per cent reduction in emissions (in the electricity market) with renewables alone without incurring large system costs. In order to achieve further decarbonisation, Boston et al argue for the incorporation of carbon capture and storage (CCS) in the energy system. Their key results are is illustrated in chart 5.12. The chart compares emissions reductions against incremental costs of abatement. IF 1 = 1 " IF STYLEREF "Heading 6" \l \n 0 = 0 " IF STYLEREF "Heading 1" \l \n 5 = 0 "" " STYLEREF "Heading 1" \l \n 5." 5." " STYLEREF "Heading 6" \l \n A." 5. SEQ Item \* ARABIC \s IF STYLEREF "Heading 6" \l \n 0 = 0 " IF STYLEREF "Heading 1" \l \n 5 = 0 "" "1" 1" "6" 1 12" "" 5.12Incremental cost of abatement: renewables versus coal-CCS Electricity market.Note: This is the right panel of the original chart and assumes renewable capital costs consistent with the CSIRO 2017 estimates. Cost of gas in the original is not reproduced here.Data source: Boston et al 2018, figure 7, right panel. As expected, abatement costs increase with abatement, but Boston et al find that these costs increase dramatically for renewables at around 65 to 70 per cent abatement. In contrast, incorporating CCS in the generation mix allows costs to be held constant up to around 85 per cent abatement. Importantly, the renewable and CCS cost curves cross: while CCS seems more expensive at low levels of abatement, it becomes cheaper at higher levels.Relevance to economic modellingThe issue of integration costs is relevant to modelling because it has implications for the type of energy system Australia (and ROW) develops, the cost of that system, and importantly the flow on economic effects of potential innovation to deal with integration issues.It is not clear the extent to which these system integration costs have been explicitly included in previous studies, so any new modelling would need to carefully account for this. Efforts to deal with integration issues will involve the use of economic resources and so will have implications for a range of economic outcomes.Further, the discussion of integration costs, and the responses proposed to it, illustrate that there is a wide range of technological futures, and that as yet there is no strict agreement about what will emerge.Structural change sensitivitiesStructural change in the economy, as a consequence of abatement policy, is sensitive to all the sorts of assumptions outline above.Because structural change is a consequence of:energy cost changestechnology (in Australia and overseas)policy changes overseas, as well as in Australia,then changes in assumptions relating to any of these will have implications for structural change. As already noted, model outcome are generally very sensitive to each of these factors, and it can be expected that many aspects of structural change would be similarly sensitive. The studies covered here, however, have not reported this sensitivity in any detail.Potential use of Integrated Assessment ModelsIAMs bring a potential advantage in that they (partly) ‘close’ the system by incorporating the cost of climate change along with the cost of mitigation.Critiques of IAMsThese potential benefits, however, come at a large potential cost. While they are widely used, IAMs have also been criticised for a number of reasons. Three key issues in IAMs include:The nature of the damage functions implemented within the model.The estimate of climate sensitivity used with the model.The discount rate chosen (when comparing benefits and costs over a long period of time).A number of recent studies by Robert Pindyck provide a comprehensive critique of the use of IAMs. Rather starkly, Pindyck recently argued that:These models have crucial flaws that make them close to useless as tools for policy analysis: certain inputs (e.g., the discount rate) are arbitrary, but have huge effects on the SCC estimates the models produce; the models’ descriptions of the impact of climate change are completely ad hoc, with no theoretical or empirical foundation; and the models can tell us nothing about the most important driver of the SCC, the possibility of a catastrophic climate outcome. (Climate Change Policy: What Do the Models Tell Us?).Damage functions for AustraliaSome of the challenges using IAMs can be seen by examining the implications of damage functions for Australia embedded within some of the models.Overall, the available damage functions appear to imply very low costs of climate change to Australia. Three examples can be used to illustrate this:Ronson and Sartori 2016 ‘Estimation of Climate Change Damage Functions for 140 Regions in the GTAP 9 Database’. (Journal of Global Economic Analysis, Volume 1, No. 2).Kompas et al 2018 ‘The Effects of Climate Change on GDP by Country and the Global Economic Gains from Complying with the Paris Climate Accord’ (Earth’s Future, 6).Ricke et al 2018 ‘Country-level social cost of carbon’, Nature Climate Change Vol. 8, pages895–900.Ronson and Sartori construct climate damage functions for countries within the GTAP model drawing on a range of estimates in the climate impact literature. Their results illustrate, for example, that the long-term economic effect of a 3 degree temperature increase on Australia is a reduction in GDP (compared with what it would otherwise have been) of 0.92 per cent.Kompas et al take the Ronson and Sartori results as a starting point, and simulate them within a very detail, forward looking CGE model. This model is able to account for the various flow on effects of climate impacts between countries. They find, for example, that the long run climate impact on Australia, for a 3 degree temperature increase, is a reduction in GDP (compared with what it would otherwise have been) of 1.08 per cent. Ricke et all use a very wide range of model comparisons to estimate the social cost of carbon (SCC, that is, the cost of 1 tonne addition CO2 emissions) for the whole globe, as well as by country. Ricke et al produce a very large range of results, but all of their results have the same broad patterns. For illustration one of their key published scenarios estimates global SCC at US$418 per tonne. Within this, the SCC for Australia is estimated at US$5.14 per tonne.A feature of each of these estimates of costs of climate change for Australia (whether expressed as a reduction in GDP or as a SCC) is that compared with the cost of abatement (as presented in the various studies above) these climate change costs are very low. An IAM (particularly a POM) using any of these cost estimates for Australia alone would likely recommend lower abatement for Australia than currently planned under the Paris agreement.This implies that before developing and using an IAM for Australia, considerable effort may need to be put into examining the climate damage functions for Australia. This would in effect be a continuation of the detailed analysis first started as part of the Garnaut Climate Change Review in 2008. There would be two important aspects of this development work: first, to incorporate non-market effects into the costs of climate change and second, to account for risk or the probability of extreme events or ‘tipping points’ in the climate system (as distinct from steady or even consequences from temperature increases). It is interesting to note that a recent analysis using an IAM found that accounting for each of these substantially changed the cost benefit assessment of climate policies.Illustrating the global nature of the problemThese results provide a very effective illustration of the global nature of the climate issue, and the very substantive externalities involved. While the impacts on Australia appear quite low, the impacts on a range of developing countries are, in fact, very large. The Kompas et al results show that the GDP impact on a range of Asian countries is up to a 15 per cent reduction (relative to otherwise).Similarly, while the SCC for Australia is US$5 per tonne, the SCC for near neighbours Indonesia and Malaysia is around US$10 per tonne, and the SCC for India is $85 per tonne.The difference between the Australian SCC and that for the rest of the world illustrates the very substantive externality in greenhouse gas emissions.IAMs are not, in fact, fully closedIt is important to note that while IAMs account for externalities associated with greenhouse gas emissions and climate effects, they do not account for other induced environmental effects of policy. For example, there are expected to be environmental consequences at various points in the renewable chain ranging from lithium and other mining for batteries to the ultimate disposal costs of solar panels.Lessons for future modellingLessons from modelling to dateEconomic costs increase with level of abatementEconomic costs increase with the level of abatement, and increase over time and increase over time as abatement increases over time. Modelling undertaken to date has given a good sense of the order of magnitude of the economic costs involved and the sorts of structural changes that are likely to result from energy and other cost increases. This is consistent with IPCC findings relating to international studies:A robust result across studies is that aggregate global costs of mitigation tend to increase over time and with stringency of the concentration goal. (IPCC AR5 WGIII, 6.3.6.2).There are a range of estimates of economic costs, and they are all sensitive to assumptions about technologyModelling undertaken to date shows that economic costs vary between models as a consequence of different model assumptions and parameters settings. Importantly, all model results are very sensitive to assumptions about the cost and availability of low carbon energy alternatives. Results seem particularly sensitive to assumptions about CCS.This is also consistent with IPCC findings relating to international studies:…the literature broadly confirms that mitigation costs are heavily influenced by the availability, cost and performance of mitigation technologies. In addition, these studies indicate that the influence of technology on costs increases with the stringency of the concentration goal. (IPCC AR5 WGIII, 6.3.6.3).As yet there is no single agreed narrative about the low carbon technological future. There are a range of different possibilities with different potential outcomes. There is, however, a common view that abatement will require considerable change in the structure of the energy sector, particularly electricity.The rest of the world mattersAbatement actions in the rest of the world — and the policies and technologies that drive them — are very important for Australian economic outcomes. There are a number of reasons for this.First, the overall economic cost of abatement estimated in a number of studies is closely related to the ability to purchase international abatement. Without this, costs of abatement are considerably higher. (This is also consistent with IPCC findings about policy integration around the world).Second, abatement actions in the rest of the world directly affect Australia through the demand for raw materials of various kinds.The effect can be negative, in the case of reduced demand for coal and other fossil fuelsThis effect can also be positive through demand for copper, lithium and other minerals associated with renewable energy or with electrification.Third, abatement actions, particularly adoption of renewable energy, around the world directly affect the cost structure of renewable energy as it appears to Australia.Fourth, specific policy actions — in particular their timing and coverage — affect the results of particular Australian abatement policies. This is especially true in the well known case of emissions intensive and trade exposed industries.The pattern of future structural change depends on all these findings and sensitivitiesThe extent of future structural change in the Australian economy in response to abatement policy in Australia and elsewhere in the world depends on all of the factors that drive the economic outcomes of mitigation policy. These factors are both technological and policy related and there is no certainty about the outcomes that will emerge. The value of economic modelsWorking through the future implications of climate mitigation — in terms of overall economic costs or the threats and opportunities related to structural change within the economy — essentially amounts to understanding the implications of a wide variety of technological futures. As noted, there is no single technological future that can be relied on to provide a single narrative for future outcomes.Models are particularly useful for exploring the implications of different assumptions about the future, along with different assumptions about future policy actions. Models have been called a ‘prosthesis for the intellect, supporting the discovery of implications of a priori knowledge’ — and in this sense they can be used to assess the detailed implications of scenarios for technological and policy futures.Modelling undertaken to date has demonstrated the ability of key model frameworks to integrate a very large amount of technological and economic information within a logical and coherent structure.Recommended ‘style’ of modellingIn considering the potential role of future modelling, it is worth noting a distinction between two very different ‘styles’ or ‘purposes’ of modelling. In some branches of the broad modelling and climate change literature, a distinction is made between ‘consolidative modelling’ and ‘exploratory modelling’.Consolidative modelling is an approach that brings together known facts into a single package which is used as a surrogate for the real world and then used to predict particular outcomes. This is the approach taken by most climate economic modelling.In contrast, in exploratory modelling, models are not used to generate predictions or to elicit ‘answers’ to explicit initial questions, but are used to generate new information helpful in deriving informed decisions. Under exploratory modelling, simulations are used to explore a very wide range of possible outcomes and to understand the broad properties of these outcomes. Exploratory modelling is much more suited to situations in which there is fundamental uncertainty about key parameters or key exogenous assumptions. This is particularly the case, for example, regarding future technology In the CGE context, exploratory modelling is closely related to ‘systematic sensitivity analysis’ which is the very detailed examination of the consequences of a full exploration of parameter values and exogenous assumptions. Rather than examining results for a single set of parameter choices, or single versions of exogenous assumptions, systematic sensitivity analysis involves exploring the implications of a very wide set of parameters and policy choices. Areas for further analysisWhile the studies undertaken to date are comprehensive and have all involved considerable analytical and development work, there remains considerable scope for further exploration of the implications of future technology and policy scenarios.More and better-defined sensitivity analysisWhile there are some large variations, results from studies to date, particularly when expressed as a percentage reduction in income related to particular abatement levels, show a clearly defined range of outcomes.What is not so clear is the full extent to which particular assumption lead to, or would change, particular outcomes.Some of the convergence between studies may be the result of common assumptions. That is, the results of too narrow an underlying set of assumptions.Further, some of the sensitivity analysis is not explicitly constructed in terms of cost of abatement or of future structural change, and generally has not been undertaken ‘systematically’ in the sense used above.Further systematic sensitivity analysis around key input and parameter assumptions would provide better understanding of the full potential range of results.Realistic modelling of country policies to understand the global cost of abatement to AustraliaThe existing studies clearly indicate that international actions, particularly relating to abatement, have important implications for the cost of abatement in Australia.One of the limitations of the existing modelling is that it uses an idealised representation of global abatement policies. This is particularly implemented through an assumption of uniform global prices.Real policy and abatement action is likely to be considerably more fragmented and unlikely to be uniform across all sectors within a given country.Given that Australian outcomes will be determined by international actions through a variety of channels, further exploration of realistic policy scenarios is likely to provide important insights into the evolution of costs for Australia.Richer technology scenarios including ‘costs of integration’ and various technologies to deal with large renewable penetration.This is related to the sensitivity analysis point above, but is focused specifically on exploring a wider set of energy technology scenarios to help understand potential adjustment paths.This analysis could consider a range of ‘integration’ issues including understanding the variety of technologies needed to maintain a stable and effective electricity grid.This would also include analysis of more complex technology scenarios where energy across different sectors (electricity and transport, for example) is integrated. An example is the potential integration of electric vehicles into the electricity grid (as a source of storage).It ma also include implications of more remote developments in solar technology that move beyond current silicon based PV techniques.Understanding more about pathways and ‘lock in’Some of the technology scenarios noted above will require particular ‘pathways’ for development; a particular sequence of investments.This would explore whether particular sequences result in lower cost overall abatement if they are implemented correctly.The other side of this is the potential cost of ‘lock in’ to ultimately inappropriate technologies, or unintentionally ‘locking out’ other technologies.Australia’s terms of trade under different technology futuresAs noted in this report, one of the channels of abatement cost to Australia is through a reduction in traditional energy based export (such as coal) as other countries go about abating.However, new energy technologies will require alternative sets of mineral inputs (lithium and copper, for example) and so it is possible that there will be increased exports from these products.This analysis would explore the extent to which this is already implicit in the economic modelling or whether there is an additional economic effect to be captured.Understanding higher order structural changeRelated to this is analysis of new structural relationships within the Australia economy that may emerge through the development and implementation of new energy technologies Specific modelling suggestionsMore analysis with global modelsA large number of the issues noted above could be fruitfully examined with careful analysis using one, or a number, of the available global economic models. Global models are key because of the need to capture a number of levels of interaction between Australia and the rest of the world, as outlined above. Further, it would be more valuable if the global models were able to capture capital flows as carbon policy particularly has significant implications for returns to capital and hence the availability of capital within the Australian economy.The nature of the Paris Agreement — involving bottom up and highly variable targets and policies proposed by countries themselves — means that variation across countries needs to be captured, and the implications for Australia well understood.More energy sector scenarios Some of the rapid changes in expectations about renewable costs illustrated above — along with the various technical developments required for integration — indicate that careful scenario analysis is required to fully explore the potential implications of higher renewable uptake.These scenarios need to go beyond the simple renewable uptake story simulated to date and capture other aspects of structural developments associated with renewable uptake.Because of the close linkages between the energy sector and the rest of the economy, it would be preferable if these scenarios were analysed using a modelling tool that incorporated CGE and energy markets within the same model. While the past strategy of iterating between a detailed dispatch energy model and a CGE country model has been fruitful, this approach has limitations and runs the risk of missing key interactions that should be kept endogenous.More end user abatement scenariosWhile energy sector developments have been a major focus of scenarios to date, it is also important to continue to explore abatement potential in other energy end user sectors and industrial processes. Again, this is ideally explored in an integrated CGE modelThe Centre for International Economics.au-27940-530606000 ................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download