Ancillary benefits of climate policy in a small open economy: The case of Sweden

Energy Policy 39 (2011) 4985–4998

Contents lists available at ScienceDirect

Energy Policy journal homepage: www.elsevier.com/locate/enpol

Ancillary benefits of climate policy in a small open economy: The case of Sweden$ b ¨ Anna Krook Riekkola a,c,n, Erik O. Ahlgren a, Patrik Soderholm a

Energy Systems Technology, Division of Energy Technology, Department of Energy and Environment, Chalmers University of Technology, SE-412 96 G¨ oteborg, Sweden Economics Unit, Lule˚ a University of Technology, SE-971 87 Lule˚ a, Sweden c Institute for Energy, Joint Research Centre, European Commission, P.O. Box 2, NL-1755ZG Petten, The Netherlands b

a r t i c l e i n f o

abstract

Article history: Received 10 February 2011 Accepted 8 June 2011 Available online 29 June 2011

It is increasingly recognised that GHG reduction policies can have important ancillary benefits in the form of positive local and regional environmental impacts. The purpose of this paper is to estimate the domestic ancillary pollution benefits of climate policy in Sweden, and investigate how these are affected by different climate policy designs. The latter differ primarily in terms of how the country chooses to meet a specific target and where the necessary emission reductions take place. The analysis relies on simulations within the energy system optimisation model TIMES-Sweden, and focuses on four non-GHG pollutants: Nitrogen Oxides (NOX), Non Methane Volatile Organic Compounds (NMVOC), inhalable particles (PM2.5), and Sulphur dioxide (SO2). The simulations permit detailed assessments of the respective technology and fuel choices that underlie any net changes in the estimated ancillary effects. The results indicate that the ancillary benefits constitute a far from insignificant share of total system costs, and this share appears to be highest in the scenarios that entail the largest emission reductions domestically. This result reflects the fact that carbon dioxide emission reductions abroad also implies a lost opportunity of achieving substantial domestic welfare gain from the reductions of regional and local environmental pollutants. & 2011 Elsevier Ltd. All rights reserved.

Keywords: Ancillary benefits Climate policy Swedish energy system

1. Introduction The balance of evidence suggests that anthropogenic emissions of greenhouse gases – out of which carbon dioxide is the most significant – are having a distinct negative impact on the global climate (e.g., IPCC, 2007). Since the Framework Convention on Climate Change was concluded in 1992, nations have been negotiating commitments to stabilise and then reduce emissions of greenhouse gases, which will otherwise continue to build up in the atmosphere. The debate on climate change policy, particularly with respect to the Kyoto Protocol in 1997, has been heavily focused on the economic costs and feasibility of the proposed mitigation plans. Despite concerns about the costs of Kyoto implementation – expressed by politicians, analysts, and industry

$ Financial support from the Swedish Energy Agency (AES program and International Climate Policy program) as well as the European Commission funded FP6 project NEEDS (New Energy Externalities for Development for Sustainability) is gratefully acknowledged, as are the helpful comments from Rodica Sandu-Loisel and two anonymous reviewers. Any remaining errors, however, reside solely with the authors. n Corresponding author at: Economics Unit, Lulea˚ University of Technology, ˚ Sweden. SE-971 87 Lulea, E-mail address: [email protected] (A. Krook Riekkola).

0301-4215/$ - see front matter & 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.enpol.2011.06.015

representatives in industrial countries – the Protocol was ratified by a large number of states and therefore came into force in February 2005. Some nations, such as the USA and Australia, based their decisions to withdraw from the Kyoto process in part on the high perceived costs for their respective economies. Also in the countries that have ratified the Protocol continued concerns exist, not the least about the future costs of the additional policy measures needed to stabilise greenhouse gas concentrations. This became evident during the 2009 Copenhagen (COP15) meeting at which no new global commitment of continued reductions of greenhouse gas (GHG) emissions could be reached. One of the most important strategies to reduce GHG emissions is to move away from the use of fossil fuels. In substituting carbon-free fuels for fossil fuels other harmful emissions are likely to be reduced along with the reduction of carbon dioxide, e.g., in replacing coal with renewable energy sources the emissions of regional air pollutants such as nitrogen oxides (NOX) and sulphur dioxide (SO2) are reduced as well. The resulting reductions in damages to health, crops and materials represent real economic benefits, i.e., reduced costs that typically are referred to as the ancillary benefits from climate mitigation (e.g., Ekins, 1996; Hourcade et al., 2001; Burtraw et al., 2003). Clearly these sideeffects can also be negative (e.g., increases in the emissions of particles when diesel replaces gasoline in the transport sector),

4986

A. Krook Riekkola et al. / Energy Policy 39 (2011) 4985–4998

but many previous studies show that compared to a baseline scenario the net economic cost of climate policy could be reduced substantially (e.g., Repetto and Austin, 1997; Boyd et al., 1985; Van Vuuren et al., 2006). In addition, by addressing the impacts of ancillary benefits and costs the optimal abatement strategy may change in terms of the reduction level, the timing of policy measures as well as the allocation of mitigation efforts across the different sectors of the economy (OECD, 2002; Kuosmanen et al., 2009). The objective of this paper is to estimate the ancillary pollution benefits of climate policy in a small open economy, and compare the outcomes of different climate policy designs. We analyse policy designs that differ in terms of how the country chooses to meet a specific target in the year 2020 including where the necessary emission reductions take place. Sweden is used as a case country, and methodologically we employ the so-called TIMES-Sweden model, a dynamic technology-rich energy system optimisation model. It represents a partial equilibrium model of the entire Swedish energy system, including stationary sources as well as the transport sectors. In addition to the supply and energy conversion sectors, five different demand sectors are described: agriculture, commercial, residential, industry and transportation. TIMES-Sweden permits the analysis of several non-GHG pollutants, including Nitrogen Oxides (NOX), Non Methane Volatile Organic Compounds (NMVOC), inhalable particles with a diameter less than 2.5 mm (PM2.5), and Sulphur dioxide (SO2). For each climate policy scenario the model is here used to address three different ancillary benefit measures: (a) total reduced damage cost; (b) reduced damage cost per reduced tonnes of CO2; and (c) reduced damage costs as a share of the total increase in system costs following the imposed climate policy. For any country the choice between domestic GHG-reduction efforts on the one hand and financing similar efforts abroad on the other is important. In accordance with the Kyoto Protocol and the EU Burden Sharing Agreement, Sweden is committed to an Assigned Amount Unit (AAU) for the compliance period 2008–2012 corresponding to an increase by 4% compared to the 1990 emission level. Still, in an attempt to precede stricter future requirements Sweden decided in 2002 on a national emission target stating that during this five-year period the country’s greenhouse gas emission level must not exceed five times 96% of the 1990 level. An important implication of this policy target has been that if a firm that participates in the European Union Emissions Trading Scheme (EU ETS) buys permits, a corresponding emission reduction has to be made in the non-trading sector (e.g., through adjustments in the CO2 tax) (Carle´n, 2004; ¨ Soderholm and Pettersson, 2008). In this way emission reduction burdens are transferred from the trading to the non-trading sector. Other than Germany and Great Britain, Sweden is the only EU country that has decided to focus on a national emissions target, addressing thus emissions made on domestic soil. In a Government Bill (2008/09:162) a new target is outlined, namely to decrease domestic emissions in the non-trading sector by 40% by the year 2020 (compared to the 1990 level). In meeting these stricter reduction requirements the Swedish government aims at locating one third of the obligated carbon dioxide reduction in other countries, thus increasing the reliance on international flexible mechanisms in the nation’s compliance strategy. Previous studies indicate that accounting for the local and regional ancillary benefits arising from climate policy that are achieved jointly with the reduction of carbon dioxide can be significant in the Swedish case, and may thus partly strengthen the case for the present adoption of a domestic emissions target. ¨ stblom and Samakovlis (2004, 2007) employ the For instance, O static general equilibrium model EMEC to evaluate the economic impacts of Swedish climate policy in the presence of benefits to

health and labour productivity following reductions in nitrogen dioxide (NO2) emissions. Their results indicate that the costs of climate policy could be substantially reduced, and the benefits of international emissions trading for the Swedish economy become less pronounced once the ancillary benefits of nitrogen dioxide reductions are taken into consideration. Similar results for Sweden are presented in Nilsson and Huthala (2000), where the EMEC model is used to address the ancillary impacts of both nitrogen oxides and sulphur dioxide. Bye et al. (2002) provide a review of the cost of climate policies and the associated ancillary benefits in the Nordic countries, the UK and Ireland. The approach in this paper differs from many earlier studies in a number of ways. First, a number of previous energy system studies use the estimated monetary damages costs for different non-GHG pollutants, and investigate, for instance, the consequences on CO2 emissions of internalising these external costs (e.g., Das et al., 2007; Klaassen and Riahi, 2007; Krook Riekkola and Ahlgren, 2003). While these studies thus address the climaterelated ancillary benefits of other environmental policies, we instead focus on the corresponding side-effects of different climate policy designs. These earlier studies also address only the external costs of the electricity (and heat) sectors. Furthermore, another set of previous studies employ general equilibrium models to analyse the ancillary impacts of GHG mitigation (Davis ¨ stblom and Samakovlis, 2007), while we instead et al., 2000; O adopt a technology rich bottom-up representation of the entire energy system. Van Vuuren et al. (2006) employ a similar bottomup approach but with a focus on Europe (divided into East, Central and West), and with no results for neither individual countries nor different climate policy designs. The TIMES-Sweden model can explicitly address important discrete technology shifts and their consequences in the presence of stringent climate policy. Moreover, the model covers the entire chain from energy supply to useful energy per demand segment, something which facilitates the identification of the sectors and technologies where the ancillary benefits are most prevalent. ¨ stblom and Samakovlis (2004, 2007), Moreover, in line with O the present paper also addresses the ancillary benefits arising from different climate policy designs for a small open economy, but unlike the previous studies we highlight in more detail the technology choice trade-offs involved in abating emissions domestically or relying more extensively on CO2 emission reductions abroad. Specifically, we investigate how the estimated ancillary benefits are influenced by changes in the stringency of the policy target, the CO2 permit price within EU ETS and the energy sector’s rate-of-return requirement. We also present detailed results for the electricity and transport sectors, and highlight important differences in fuel mixes across the different policy scenarios. Finally, the range of non-GHG pollutants considered in the analysis is also wider, not the least given the inclusion of particles (PM2.5) and NMVOC, substances that are typically ignored in most previous studies. Section 2 presents the TIMES modelling framework, and clarifies how the assessment of ancillary benefits can be incorporated into the model simulations. In Section 3 we discuss the different policy scenarios, while Section 4 presents the model simulation results. Finally, some concluding remarks are provided in Section 5.

2. An energy-economic modelling framework 2.1. The TIMES modelling framework In order to estimate the economic costs and the future emission levels for each policy scenario the TIMES-Sweden model

A. Krook Riekkola et al. / Energy Policy 39 (2011) 4985–4998

is used. TIMES-Sweden is built on the platform of TIMES (an acronym for The Integrated MARKAL-EFOM System), and represents a linear programming cost-optimising model developed within ETSAP (www.etsap.org). TIMES minimises the total discounted system cost over the modelled time-horizon to meet a given demand of useful energy in line with the following general objective function: TSC ¼

Y X

ASCy  ð1þ dÞð2000yÞ

ð1Þ

y ¼ 2000

where TSC is the total (discounted) system cost (in Euro) during the entire modelling period, y represents the modelling year, Y is the last year of modelling (i.e., the modelling horizon plus any additional years during which costs must be borne), ASCy is the annual undiscounted system costs in year y, and d is the discount rate employed. TIMES offers a technology-rich description of any local, national or multi-national energy system, and allows for an exploration of possible energy scenarios. The dynamic modelling framework permits a flexible number of time-slices over the modelling horizon (Loulou et al., 2005). Unlike many other bottom-up energy system models, TIMES assumes that energy demand is responsive to price changes. The model is flexible in describing different policy instruments, i.e., taxes, subsidies, tradable emission permits, and these policy instruments can be applied on fuels, emissions and technologies (activity and capacity). The cost of an imposed climate policy includes the increased system cost compared to a baseline (no-policy) scenario. Since all investment costs have been annualised (given a certain economic lifetime and discount rate) it is straightforward to present the change in system cost for a given year. 2.2. The TIMES-Sweden model The TIMES-Sweden model was initially developed as one of the European national models in the New Energy Externalities Developments for Sustainability (NEEDS) project (see www.needs-pro ject.org), and all national models were further improved and used within the RES2020 project (www.res2020.eu). All national models optimised together in turn constitute the Pan European TIMES (PET) model. TIMES-Sweden describes a large number of existing and potential energy technologies, and includes a detailed description of

Policy Instruments

4987

the stationary energy systems, the transport sector and several demand sectors in the country. In order to represent the load curve, the electricity and heat sectors can be divided into twelve timeslices (four seasons, day/night, as well as daily peak-hours). All twelve time-slices are used to describe electricity demand, while only four seasons are used to describe the corresponding demand for heat. A simplified description of the processes and commodities in TIMES-Sweden is provided in Fig. 1. Primary energy includes both domestic supplies and import of energy commodities to the country. Furthermore, the supply sector includes processes refining some of the primary energy sources into secondary energy carriers. The electricity and heat sectors include units generating electricity and district heating services. In addition to different industrial segments, the industry sector also includes internally generated electricity and district heating production. The industrial segments are described in detail with different steps of processes, and in TIMES-Sweden the two largest segments are the iron and steel industry and the pulp and paper industry. The agricultural sector is represented by only a few process alternatives available, although all with several combinations of inputs. The commercial and residential sectors are described in detail with different demand (and technology options) for space heating, water heating, cooling, lightning, coking, etc. Space heating can be served by different technologies relying on wood, oil, gas, electricity, heat pumps or district heating. The useful demand is calculated outside the model, and it is derived based on assumptions about, for instance, GDP, population, changes in the residential housing stock, etc. However, final use of energy per demand segment is calculated endogenously in the model, it is sensitive to price changes and presented by energy commodity. National price adjustments in TIMES-Sweden are based on statistics from the Swedish Energy Agency (2006). In this study we use a uniform social discount rate of 4%, equal to the one recommended by the Swedish Institute for Transport and Communications Analysis for use in the case of publicly funded infrastructure projects (SIKA, 2002). In the sensitivity analyses, though, we also investigate the consequences of employing different discount rate assumptions. In-data are kept in line with the assumptions specified in RES2020 (2009), and with several national adjustments. The main source for the base-year energy balances is the Eurostat data-base provided by the Statistical Office of the European Communities.

Emissions

Electricity Heat

IMPORT

EXPORT International Markets

Agriculture Commercial

S P R I M A R Y

E N E R G

U P P L

Electricity & Heat

Residential Transport

Y Industry

Y

Fig. 1. An overview of TIMES-Sweden.

U

D

S

E

E

M

F

A

U

N

L

D

4988

A. Krook Riekkola et al. / Energy Policy 39 (2011) 4985–4998

Table 1 Three different ancillary benefit measuresa. Ancillary benefit measure

Definition

Description

Reduced total damage cost Reduced damage cost per reduced CO2 Reduced damage cost per increased system costs

DADC DADC/DCO2 DADC/(  DASC)

Annual reduced damage costs (Euro) Reduced annual damage costs per reduced tonne of CO2 emissions (Euro/tonne reduced CO2) Reduced damage costs per increased annual total system costs (percentage share)

a

All damage costs refer to changes in the emissions from non-GHGs, and do thus not incorporate the economic effects of greenhouse gas emissions.

Specifically, base year energy flows and existing technologies are based on Eurostat data for the year 2000, and the model is calibrated for the year 2005. Other important data sources for the RES2020 database are outlined in Garqiulo (2008). The strong reliance on official statistical sources as well as technologyspecific information derived from previous research projects (e.g., the BRED study and the REFUEL project) should ensure the reliability of the RES2020 data.1 Moreover, details of the present Swedish energy system have been collected from various Statistics Sweden reports (e.g., Statistics Sweden, 2007; Nordel, 2008). The database for new technologies, NewTechs-RES2020, contains more than 700 technologies, described in terms of efficiency, availability factors, investment costs, annual fixed, variable operation and maintenance costs, emissions factors, output, etc. In TIMES-Sweden this database has been adjusted to account for country-specific conditions, e.g., for the electric power sector complementing data have been drawn from Elforsk (2007). TIMES-Sweden does not incorporate endogenous technological change (learning-by-doing), i.e., technological progress in, for instance, the energy conversion sector is assumed to be independent of domestic diffusion rates and policies. The immature technologies are instead characterised by exogenous efficiency improvements and thus lower costs over time. In the model technological improvements can either be attributed to the entire stock of a specific technology, or be specified to affect only new investments. In TIMES-Sweden the following emissions are described in detail: Carbon Dioxide (CO2), Nitrogen Oxides (NOX), Non Methane Volatile Organic Compounds (NMVOC), Particulate 2.5 (PM2.5) and Sulphur Dioxide (SO2). Emissions can be defined by fuel consumption (e.g., CO2 emissions from burning coal) and by the activity of a process (e.g., SO2 emissions when generating electricity from CHP plants). In the RES2020 project, the emission factors were based on the average emission coefficients calculated from the RAINS model, which includes regional data; the emission factors are in turn assumed to decrease over time. In TIMESSweden the emission factors have been validated and updated ¨ based on national data from, for instance, Bostrom et al. (2004) and the Swedish Environmental Protection Agency (2009). Overall the Swedish emission factors are generally lower than those typically assumed for other European countries. 2.3. Incorporating the ancillary benefits of climate policy in TIMESSweden In analysing the ancillary benefits of climate policy the definition of the baseline scenario is critical for the results (Morgenstern, 2000). For our purposes it is particularly important to include policies that affect the emissions of regional and local air pollutants, and our baseline scenario therefore includes, for 1 Assumptions of future technology costs are of course afflicted with uncertainties and this concerns in particular assumptions regarding technology development (efficiency and cost developments) and factors related to site-specific conditions. Most notably, the future costs of scaling up existing technology or introducing still fairly new technology may be difficult to assess.

instance, the existing taxes on fossil fuels in Sweden. The assumed baseline scenario is discussed in more detail in Section 3.1. In TIMES-Sweden the annual emissions for each pollutant are ultimately an outcome of the model optimisation. By multiplying these with the unit damage cost estimates from the literature, a monetary valuation of the yearly change in damage from a given climate policy is attained. We have

DADCy ¼

P X

DQpy DCp

ð2Þ

p

where DADCy is the change in total damage costs (in Euro) from the baseline scenario, DQpy is the net change (in tonnes) in the emissions of pollutant p (p ¼1,y,P) in time period y, and DCp represents the corresponding damage cost per tonne of pollutant p (Euro per tonne). In this paper we follow Hourcade et al. (2001), and present the yearly ancillary effects from each imposed climate policy in terms of the changes in: (a) reduced total damage costs; (b) reduced damage cost per reduced tonne of CO2; and (c) reduced damage costs as a share of the increase in annual system costs following the climate policy (all compared to the baseline). These measures are summarised in Table 1. The total amount of emissions generated from energy conversion can depend on technology, choice of fuel and be due to behavioural changes. Our focus lies on the ancillary benefits from fuel switching and technology choice (including the adoption of more efficient technologies that use less fuel). Overall behavioural patterns are assumed to change only as the result of the different economic incentives provided by the varying climate policy designs. For instance, the transport sector is defined by a fixed yearly demand (in million person kilometres) per vehicle category (e.g., short-distance person traffic) even though the sector has a large potential to reduce emissions through behavioural changes (e.g., by moving away from studded tires in larger cities and by switching from private to public transportation). The included non-GHG emissions all have local and/or regional environmental impacts (see Table 2). In Sweden the main sources of NOX and SO2 include shipping companies, process industries (e.g., pulp and paper, iron and steel, chemical industry, etc.), and other combustion. NOx derives from all kind of combustion, while SO2 only derives from combustion with fuels containing sulphur, i.e., coal and oil. Both these pollutants cause acidification (on forests and buildings) and give rise to negative health effects. In addition, NOX emissions also stem from the transportation, agriculture and construction sectors, and cause eutrophication of the oceans. Nitrogen oxides are also involved in the formation of ground-level ozone. The main sources of inhalable particles with a diameter less than 2.5 mm (PM2.5) are industry processes, traffic-exhaust and small-scale wood combustion. Particles cause negative health effects through, for instance, increased risks of respiratory diseases. Inhalable particles with a diameter between 2.5 and 10 mm also cause negative health effects; they stem mainly from road dust, but are not addressed in the TIMES-Sweden model. NMVOC origins from traffic-exhaust and small-scale wood combustion, and gives rise to ground-level ozone emissions and

A. Krook Riekkola et al. / Energy Policy 39 (2011) 4985–4998

4989

Table 2 Main sources and impact of selected non-GHG pollutants in Sweden. Sources: Statistics Sweden (2008) and Swedish Environmental Protection Agency (2008). Pollutant

Symbol

Main sources of the pollutant

Main impacts

Nitrogen oxides

NOX

Sulphur dioxide

SO2

Health effects, acidification, eutrophication, ground-level ozone Health effects, acidification

Inhalable particles, diameter o2.5 mm. Non methane volatile organic compounds

PM2.5

Shipping companies, transport, process industry (pulp and paper, iron and steel, and chemical industry), agriculture, construction, and combustion. Shipping companies, process industry (pulp and paper, iron and steel, and chemical industry), and combustion of coal and oil. Industry and combustion processes (traffic-exhaust and small-scale wood combustion). Small-scale wood combustion, traffic (evaporation from gasoline), use of solvents.

NMVOC

Table 3 External costs for non-GHG pollutants in Sweden (Euro2000 per tonne). Source: Holland and Watkiss (2007). Sources

NOX

SO2

PM2.5

NMVOC

ExternE core CAFE/WHO-low CAFE/WHO-high

760 2200 5900

1500 2800 8100

11,000 12,000 34,000

230 330 980

negative health effects (including cancer). In the model, only emissions that origin from energy conversion are included. In order to estimate the damage cost per pollutant we have used country-specific external cost estimates per tonne of emission of each pollutant generated within the MethodEx project (www.methodex.org), which in turn is based on the methods developed and used within ExternE and CAFE/WHO, respectively. The external cost estimates include health impacts and the effects from ozone on crops, but not the impacts on the ecosystem or on materials. The so-called BeTa-MethodEx reference outputs for Sweden are presented in Table 3, and they display a range of monetary estimates for the involved emissions. All three sets of estimates include the total damage costs from core health effects and crops, while the results presented in CAFE/WHO-high also include so-called sensitivity health effects. In the empirical analysis we acknowledge the uncertainties that exist with respect to these environmental damage costs, and subsequently present model simulations that rely on all of the three studies/approaches presented in Table 3.

3. Scenario definitions In the baseline scenario the existing Swedish policy instruments are maintained over the modelling period (the year 2020), while the different climate policy scenarios all imply a stricter emission target but with different policy designs. The main inspiration for the design of the policy scenarios stems from the Swedish government’s goal to achieve a 40% reduction in GHG emissions in the non-trading sector compared to 1990 years level by the year 2020. Two thirds of these GHG reductions should take place domestically, while one third can be achieved through investments in other nations within the European Union or through the Clean Development Mechanism (CDM). However, our policy scenarios involve a 40% reduction for the entire country (trading sector þnon-trading sector), but with varying opportunities for making use of the flexible mechanisms. Thus, this paper does not aim at providing an explicit evaluation of the existing Swedish climate policy; it investigates instead the ancillary benefits in various climate policy scenarios that differ significantly in the importance placed to domestic reduction commitments versus the use of permit trading.

Health effects Ground-level ozone, health effects

An important reason for departing from the current Swedish climate policy target is also that our simulations indicate that this is close to being attained already in the baseline scenario. Moreover, in the presence of a future commitment on global emissions reductions, any country would have to address in more detail the interaction between the trading sector and the non-trading sector. In our analysis the domestic reductions will be achieved by the existing European Emissions Trading System (EU ETS) for the electricity and heat sectors and specified industry segments (i.e., the trading sector), and through increased carbon, fuel and energy taxes for the non-trading sectors (Government Offices of Sweden, 2009). The reductions in the trading sector will thus depend on the price of tradable emission permits. 3.1. The baseline scenario In accordance with the Kyoto Protocol and the EU Burden Sharing Agreement, Sweden is committed to an Assigned Amount Unit (AAU) for the compliance period 2008–2012 corresponding to an increase by 4% compared to the 1990 emission level. Still, in 2002 the Swedish government decided on a national emission target stating that during this five-year period the country’s greenhouse gas emission level must not exceed five times 96% of the 1990 level. In order to meet this goal, Sweden has introduced a number of policy instruments. One of the most important policy instruments in Swedish climate policy is the CO2 tax, which was introduced already in 1991. The trading sector’s emissions are in turn determined by the carbon permit prices in EU ETS. In the baseline scenario we assume that the existing CO2 taxes are remained unchanged throughout the modelling period. Moreover, the tradable emission permit (TEP) price is assumed to be constant at 11 Euro2000 per tonne of CO2 based on the average prices of Futures Contract at EEX (ECX, 2009). In addition to the above-mentioned climate policies there are also a number of additional environmental and fuel taxes (Swedish Tax Agency, 2008a, 2008b) and technology support schemes that influence the emissions of non-GHG pollutants in the baseline. These existing policy instruments are addressed in the analysis (see Appendix A for an outline of existing energy-related taxes). During an initial transition period, the trading sector pays both a TEP price and a CO2 tax. In the model these rebates are consistent with the existing regulations for the trading sectors; fuels used for electric power generation and industrial heat have a 100% tax rebate from the energy and CO2 taxes, while heat production in combined heat and power units has a 85% tax rebate from the CO2 taxes. Still, in the model simulations all CO2 taxes in the trading sector are assumed to be phased out in the year 2010. The use of fossil fuels for motor-driven purposes is subject to energy and CO2 taxes. Biofuels are however not taxed, ˚ but following Kageson (2007) we assume that from 2010 and onwards also these renewable fuels will face an energy tax corresponding to those currently facing the different fossil fuels.

4990

A. Krook Riekkola et al. / Energy Policy 39 (2011) 4985–4998

Table 4 Modelled CO2 restrictions in each policy scenario. Baseline

Country-cap

Sector-cap

EU

Trading sector Non-trading sector

No No

Not applicable Not applicable

No No

Total

No

Emitted CO2 r national target

No Emitted CO2 in the non-trading sector rnational target - emitted CO2 in the trading sector Emitted CO2 in the trading sectorþ emitted CO2 in the non-trading sector r national target

In 2003 Sweden implemented a green certificate market for renewable electricity; in 2016 the production of green electricity should have increased by a total of 17 TWh compared to the 2002 level (Swedish Energy Agency, 2007). This target has been implemented in the baseline model simulations.2 The energy sources that entitle Swedish electric power producers to issue green certificates are wind power, solar energy, wave energy, geothermal energy, new hydropower, existing small-scale hydropower, and biomass (including peat). In the model, the Swedish green certificate system is modelled endogenously to meet the quotas outlined in Swedish Energy Agency (2007). Each MWh of electricity from renewable energy sources gives one Tradable Renewable Certificate (TRC), and the amount of TRC should be equal or greater than a given share of the total electricity consumption for non-production purposes. 3.2. Climate policy scenarios In this section we define three main climate policy scenarios, which all involve a stricter climate policy compared to the baseline scenario. For all three policy scenarios the cap in EU ETS is tightened, resulting in increasing prices of TEP. The assumed prices are 22 EUR2000 per tonne of CO2 in the period 2010–2014 and 44 EUR2000 per tonne of CO2 from the year 2015 ¨ stblom and Samakovlis and onwards. In addition, following O (2004) three scenarios, each describing a given climate policy target at the country level, are employed:

 The Country-Cap scenario represents a scenario in which the





entire 40% reduction target must be achieved through domestic reductions. Thus, in this scenario we have a national cap on emissions, which in addition to the existing cost of emitting CO2 (either through permit trading or a tax) creates a uniform shadow price on CO2 emissions. The marginal cost of abatement is equalized across all sources. In this way we obtain a cost-effective reduction of emissions within the country, but it is not possible to utilise the benefits of emissions trading. The EU scenario is a scenario in which the domestic policies remain the same as in the baseline scenario, and in which all sectors of the Swedish freely can trade permits at a prespecified EU ETS price level (see above). The marginal abatement cost is thus equalized across all sources at the level of this exogenous price. In other words, in this scenario the climate policy target is achieved by including the permits traded, and total emissions may exceed the national target by permits bought as these emission reductions are accomplished in other countries. In the Sector-Cap scenario the trading sector can engage in permit trading within EU ETS but (unlike the EU scenario) the 40% target must be achieved exclusive of traded permits. Thus, an important implication of this national emission target is

2 In 2010 this target was revised, and the present Swedish target outlines an increase in green electricity by 25 TWh by the year 2020.

Emitted CO2 in both sectors— net purchase of TEP r national target

Table 5 National targets for each policy scenario, 2010 and 2020a.

2010 2020

– –

Country-cap

Sector-CapA

Sector-CapB

EU

49.2 (96%) 30.8 (60%)

49.2 (96%) 37.6 (73%)

49.2 (96%) 30.8 (60%)

– –

a National emissions presented in million tonnes of CO2 and in the brackets the ratios (in percentage) of the resulting CO2 emissions in the baseline scenario (equalling 48.5 million tonnes of CO2 in the year 2000) and the emission level in the year 1990 are outlined.

that if a firm in the trading sector chooses to buy permits, a corresponding reduction has to be made in the non-trading sector (e.g., through adjustments in the carbon tax). In this way the EU ETS participants can ‘‘involuntarily’’ transfer emissions reductions to the non-trading sector. Given the present Swedish policy it is useful to also consider the case where only a fraction of the national target must be achieved domestically. The scenario is thus divided into two subscenarios: the Sector-CapA scenario allows for one third of the reductions abroad (through purchases of permits), while in Sector-CapB the entire reduction target must be achieved domestically. Table 4 summarises the modelled restrictions on CO2 emissions in the respective scenarios, while Table 5 also outlines the respective national targets (in million tonnes) for the years 2010 and 2020. 3.3. Sensitivity analyses TIMES-Sweden relies on a representation of the base year energy system as well as on assumptions about the future. It is therefore important to perform some sensitivity analysis, and in this paper we present simulation results, which rely on varying assumptions concerning the CO2 target, the TEP price in EU ETS and the discount rate. In the case of alternative CO2 targets, we focus solely on the Country-Cap scenario, and illustrate to what extent the estimated ancillary benefits differ if this target is assumed to be both higher (53%) and lower (20%) than the current 40% target (all compared to the 1990 emission level). The future prices of TEP are clearly uncertain; only in the year 2008 the price of futures contracts (December 2010) varied from 15 to 32 Euro per tonne of CO2 (ECX, 2009). In the sensitivity analysis we assume that from the year 2015 the price of TEP equals 22 Euro2000 and 99 Euro2000 per tonne of CO2, respectively, in the EU scenario (instead of 44 Euro2000 per tonne). In this case we thus focus on the outcomes of the EU scenario. Moreover, the choice of discount rate when comparing different climate mitigation policies have been discussed intensively in the literature (e.g., Azar and Sterner, 1996; Stern, 2006). The current Swedish practise for large public infrastructure investments is to use a social discount rate of 4% (SIKA, 2002). The Swedish Institute for Transport and Communications Analysis recommends the use of 2% and 7% discount rates,

A. Krook Riekkola et al. / Energy Policy 39 (2011) 4985–4998

respectively, in sensitivity analyses, and this is also the approach followed in this paper. In our analysis the discount rate is varied in the Country-Cap scenario, and for consistency the baseline scenario has been treated in the same manner.

4. Model simulation results and discussion In this section we present and discuss the results of the model simulations. We pay particular attention to the assessment of the ancillary benefits in each of the three main climate policy scenarios, and analyse the extent to which these results appear sensitive to varying assumptions about TEP prices, overall CO2 targets, and discount rates. It should be noted that since the baseline scenario includes existing policy instruments, the estimated ancillary benefits are only those that arise as a result of the stricter policy targets. Before proceeding, however, it is useful to briefly investigate the national CO2 emissions reductions in each of the policy scenarios investigated. 4.1. Carbon dioxide emissions in the respective scenarios Our simulation results for the year 2020 show that there is a net decrease in emissions already in the baseline scenario. This reduction is largely due to a phase out of oil for heating purposes as a result of increasing oil prices. In addition, there is also a reduction of CO2 in the transport sector due to a shift to more efficient vehicles as well as to biofuels, induced by the existing fuel taxation, increasing oil prices and technological progress. The EU target of 10% renewable energy sources in the road transportation sector is in fact achieved already in our baseline scenario. After the year 2020, however, the increases in CO2 emissions in the industrial sector as well as in the electricity and heat sectors exceed the emission decreases in the other sectors, thus calling for additional and/or strengthened policy instruments to maintain a decreasing overall trend. Table 6 presents the emissions of CO2 in the respective policy scenarios, all compared to the baseline scenario. In order to meet the Swedish climate target of 40% emissions reduction by 2020 solely through domestic measures (i.e., the Country-Cap and Sector-CapB scenarios, respectively), the reduction from the baseline path needs to be 24.5%. The Sector-CapB scenario has equal overall domestic reductions to that in the Country-Cap scenario, but with different policies for the trading and non-trading sectors, respectively. As will be illustrated below, this difference across the two policy scenarios has some interesting implications for the estimated ancillary benefits. The price of permits provides the main incentive for carbon reduction in the trading sector, and any additional domestic reductions needed to reach the 40% national target must be met by the strengthening of existing domestic policy instruments.

Table 6 Swedish CO2 emissions in the baseline scenario and percentage reductions from the baseline in the respective policy scenarios (2020). Baseline (million tonnes)

Countrycap

EU

SectorCapA

SectorCapB

Percentage change from Baseline scenario (%) Trading sectors Non-trading sectors Total

15,908 24,826

 37.4  16.2

40,734

 24.5

 26.1  26.0 5.0 2.3  7.1

 8.7

 7.2  35.6  24.5

4991

In the model simulations we impose the residual target on the non-trading sector and a new shadow price on CO2 emissions is attained. The results in Table 6 show that that in the Sector-CapB scenario the national target can only be met through deep reductions (  35.6%) in the non-trading sector. The corresponding reductions in the Country-Cap scenario are much more modest. The EU and the Sector-CapA scenarios both imply that Sweden can depart from a national emissions target, and make use of the benefits of permit trading. In both scenarios total domestic emissions are significantly higher compared to the other two scenarios, this since the lack of domestic reductions is compensated by permits bought in the international market. Both these scenarios thus imply a significant loosening of the aggregate national target, something which in turn explains why the nontrading sector’s emissions increase (Table 6). This also indicates that compared to other Member States, Sweden is a country with a relatively high marginal cost of CO2 abatement (especially in the non-trading sector). In the EU scenario, the extent to which Sweden relies on reductions in other countries is determined entirely endogenously in the model simulations, while the Sector-CapA scenario only permits one third of total emissions reductions to take place abroad. The decrease in domestic CO2 emissions is only slightly lower in the EU scenario, thus indicating that the two-third principle represents a binding constraint, but it does not impose a significant burden on the Swedish economy (see also Section 4.2). Specifically, it appears that the flexibility to count one-third of internationally purchased permits toward the national target is enough flexibility to almost completely reverse the Sector-CapB outcome. The Sector-CapA scenario thus comes close to the EU scenario with no ‘‘supplementarity’’ constraint. 4.2. An assessment of the resulting ancillary benefits Fig. 2 shows – for each policy scenario – the estimates of the changes (from the baseline) in: (a) total annual system costs excluding any ancillary benefits (DASC); (b) the annual ancillary benefits given three different estimates of the external costs (DADC); and (c) the total annual system costs including the ancillary benefits (DASC þ DADC). The results are presented for the year 2020 and indicate that total system costs (excluding any ancillary benefits) are typically lower in the two scenarios that permit Sweden to make use of emissions trading to comply with the climate policy target. It can also be noted that imposing a uniform additional price on CO2 in Sweden (Country-Cap) would represent a more cost-effective policy than the one represented by the Sector-CapB scenario in which the trading and the non-trading sectors face different additional CO2 prices. This result is well in line with other studies using computable general equilibrium models of the Swedish ¨ stblom and Samakovlis, 2004)3. The results also economy (e.g., O illustrate that the Sector-CapA scenario imposes marginally higher system costs than the EU scenario, thus reinforcing the above conclusion that the proposed minimum cap on domestic reduction in this scenario imposes only a very minor constraint on overall policy compliance. Moreover, the Sector-CapB scenario implies substantially higher costs than the other three scenarios, essentially since in this case the energy system is forced to 3 It should be noted, though, that the ways in which economic costs (or reduced economic welfare) are defined differ between energy system models and general equilibrium models, respectively. For instance, the former models typically address the importance of direct technology-specific costs, while ignoring, for instance, the impacts that a policy targeted towards one market can have on economic decisions in other markets. The ways in which non-market costs – and the internalisation of these through policy – are treated may also differ.

4992

A. Krook Riekkola et al. / Energy Policy 39 (2011) 4985–4998

MEuro 1300 1100 900 700

"CountryCap" "EU" "SectorCapA" "SectorCapB"

500 300 100 -100 -300 Change in Annual System Cost excl Ancillary Effects

ExternE

CAFE WHO Low

CAFE WHO High

Change in Ancillary Effects

ExternE

CAFE WHO Low

CAFE WHO High

Change in Annual System Cost incl Ancillary Effects

Fig. 2. Estimated change in system cost and ancillary effects compared to the baseline scenario (million Euro2000).

Table 7 Ancillary benefits (Euro) per ton CO2 reduced and as a share of total costs. Country-Cap

EU

Sector-CapA

Sector-CapB

Reduced damage cost (DADC)/tonnes of CO2 reduced (Euro2000/tonne CO2) ExternE 3.9 5.1 6.0 5.0 Cafe/WHO-low 7.5 8.9 11.6 10.9 Cafe/WHO-high 20.8 24.7 32.1 29.8 Reduced damage cost (DADC)/increased system cost (DASC) (percentage share) ExternE 6.1 3.1 4.4 3.8 Cafe/WHO-low 12.0 5.4 8.5 8.3 Cafe/WHO-high 32.0 15.0 24.0 23.0

undertake substantial domestic reductions in the non-trading sector in which the marginal cost of CO2 abatement is high compared to the corresponding cost in the trading sector. Economically the ancillary benefits constitute a far from insignificant share of total system costs (see also below), and these appear to be most prevalent in the scenarios that entail the largest emission reductions domestically. The latter results are far from unexpected since an increase in emission reductions abroad also implies a lost opportunity of achieving substantial welfare gains from the reductions of a number of regional and local environmental pollutants. Still, the estimated size of the ancillary benefits is not large enough to fundamentally alter the ranking of climate policy design in terms of system cost impacts. One can note, though, that in the two cases where the CAFE/WHO external cost estimates are used, the Sector-CapA scenario implies lower overall system costs than the EU scenario, thus in part speaking in favour of the current Swedish government’s policy to restrict emission reductions abroad. Table 7 presents more details on the estimated total ancillary benefits, and displays these in terms of reduced (non-carbon) damage costs per tonne (domestically) reduced CO2 and as a share of increased total system costs. The relatively low domestic CO2 reductions in the EU and the Sector-CapA scenarios imply a high reduced damage cost per tonne reduced CO2. The ancillary benefits of Swedish climate policy, expressed as a share of the total cost of the policy, are overall the highest in the Country-Cap scenario. Again, the advantages of a climate policy relying heavily on international permit trading become less

pronounced when taking into account the local and regional environmental impacts of reduced CO2 emissions. Although the total policy costs is lower with permit trading the difference in cost between, say, the Country-Cap scenario and the EU scenario, is significantly reduced when the ancillary benefits are subtracted from total system costs. The economic benefits of domestic reduction become even more significant with higher permit prices (see also Section 4.3), and higher economic valuations of the damages from regional and local pollution impacts. On a per tonne basis it is also interesting to note that the Sector-CapB scenario involves larger ancillary benefits than the Country-Cap scenario. An important reason for this result is likely to be that the former scenario puts more emphasis on limiting CO2 emissions in the non-trading sector that includes sources with presumably less pollution control and/or higher exposure than the large stationary sources within the trading sector. Still, it is useful to analyse in more detail to what extent the different pollutants contribute to the total ancillary benefits reported in Fig. 2 as well as the role of different technologies and fuels. Fig. 3 presents the total ancillary benefits by pollutant. It can be noted that for all four types of pollutants, there is a positive benefit in terms of reduced environmental costs. Thus, in none of the cases we witness a net emissions increase as a result of the different climate policies. However, the relative contribution of the different pollutants to the total ancillary benefits differs, and this is in part contingent on the choice of source for the external cost estimates. For instance, the ExternE-based estimates indicate an important role for the reduction of particles, while the CAFE-estimates instead display a more significant contribution from reduced nitrogen emissions. Still, overall all pollutants – with the notable exception of NMVOC – contribute significantly to reduced system costs following the implementation of more stringent climate policies. Finally, the TIMES-Sweden model permits a technology-rich description of the energy system, and for this reason it is useful to exploit this feature of the model to discuss some of the technology choices underlying the results presented above. Fig. 4 shows the primary energy supply by fuel and technology source in 2020 for each policy scenario, while Figs. 5 and 6 display the corresponding fuel mixes for the electricity and transport sectors, respectively. In Figs. 4–5 the generation from nuclear and hydropower has been excluded, primarily since these do not change over the relevant period. The installed capacity of large-scale

A. Krook Riekkola et al. / Energy Policy 39 (2011) 4985–4998

4993

MEuro 250.0 200.0 VOC SO2 PM2.5 NOX

150.0 100.0 50.0

Extern

ECAFE WHO - Low

SectorCapB

SectorCapA

EU

CountryCap

SectorCapB

SectorCapA

EU

CountryCap

SectorCapB

SectorCapA

EU

CountryCap

0.0

CAFE WHO - High

Fig. 3. Ancillary benefits by pollutant in all policy scenarios for the year 2020 (million Euro2000).

Fig. 4. Primary energy supply in Sweden by energy source (PJ in 2020). Excluding hydro, nuclear and net electricity imports, which are assumed to be constant across all scenarios. Overall these options supply 1045 PJ in the year 2020.

hydropower is constant over the modelling period, assuming that the refits of old units will be executed independent of the scenarios investigated. In the case of nuclear the existing and planned capacity (in 2008) is also constant with options to invest in new capacity, i.e., assuming that any refit of existing nuclear plants will be executed. During the period and with the present assumptions of investment cost, new nuclear is however not an economical option (and the model simulations suggest no new nuclear power in Sweden during the period). Figs. 4–6 display a number of interesting results. First it may be noted that the scenarios involving substantial domestic CO2 emission reductions (Country-Cap and Sector-CapB) show a comparatively low use of oil, while the shares of wind power and biomass instead are high compared to the other scenarios. In the Country-Cap scenario, which represents a cost-effective (domestic) policy with a uniform price on CO2, the main reductions occur in the industry sector and in the electricity and heat sectors (e.g., the substitution of wind power for fossil fuels). The Country-Cap scenario in fact represents the scenario with the most rapid expansion of wind power (Fig. 5), and indicates a wind

power generation level of 20 TWh by the year 2020 (with considerably lower figures in the other policy scenarios).4 Additional CO2 reductions are also achieved through an increased use of biomass and electricity for transportation purposes. The Sector-CapB scenario involves the most significant increase in biofuels in the transport sector, primarily due to a substitution away from diesel use (Fig. 6). This substitution effect is not as prevalent in the Country-Cap scenario since this involves instead more significant emission reductions in the trading (nontransport) sector. The penetration of electric vehicles is marginal in all scenarios due to high costs. The resulting ancillary benefits in the Country-Cap and SectorCapB scenarios can be explained by reductions of NOX emissions

4 These results are quite consistent with the current rapid expansion of wind power in Sweden. It may be noted that our baseline scenario indicates a wind power generation level of less than 3 TWh in the year 2020, but this level was exceeded already in 2010. The Country-Cap result of 20 TWh in 2020 is just in line with the Swedish government’s so-called planning goal for onshore wind power in 2020.

4994

A. Krook Riekkola et al. / Energy Policy 39 (2011) 4985–4998

Fig. 5. Electricity generation by source in Sweden (TWh in 2020). Excluding hydro, nuclear and net electricity imports, which are assumed to be constant across all scenarios. Overall these options supply 132 TWh (in net energy terms) in the year 2020.

400 350 Biofuels for Blending

300

Biofuels

250 (PJ)

Natural Gas

200

Kerosene/Jet Fuels Oil

150

Electricity

100

Diesel

50 Gasoline

0 Baseline

Country-Cap

EU

Sector-CapA Sector-CapB

Fig. 6. Fuel use in the Swedish transportation sector (PJ in 2020).

from the transport sector (due to complex freight traffic fuel substitutions creating more favourable conditions for, for instance, the use of DME compared to methanol) as well as reductions of SO2 in the non-trading industry segments when substituting away from oil to biomass. The model simulations suggest a substitution of biomass for oil for process heating within the agricultural and industrial sectors. These impacts involve important reductions in non-GHG emissions (e.g., SO2). Moreover, as was noted above these scenarios are essentially forcing the energy system to undertake domestic emissions reductions in the non-trading sector, implying greater use of biofuels in the transport sector. This means in turn that a significant share of the estimated ancillary benefits stem from the substitution of biomass and imported biofuels for diesel and gasoline. Even though electricity use overall is higher in these scenarios, Fig. 5 shows that the use of biomass in the electricity sector is lower than in the baseline. This can be explained by the fact that under these policy designs the willingness-to-pay for biomass is higher in the transport sector and in the other non-trading sectors. In those scenarios where there instead exist fewer restrictions on the non-trading sector – i.e., the EU and the Sector-CapA scenarios – we find different results. In these scenarios, the primary use of biomass in the year 2020 is similar to the corresponding use reported for the baseline scenario (but lower compared to the

Country-Cap and the Sector-CapB scenarios). For this reason we also witness a lower reduction in diesel use in the transport sector (Fig. 6). However, instead more biomass is used in the electricity sector at the expense of less wind power (Fig. 5), and there is also a shift from biomass to electricity within the industry sector resulting in lower net reductions of (primarily) NOX but also NMVOC. Although the EU and the Sector-CapA scenarios are overall similar, important differences do exist. For instance, in the SectorCapA scenario biomass is to a greater extent used to produce process heat in the non-trading industries compared to the EU scenario, and the penetration of wind power is also higher in the former case. The substitution of wind power for fossil fuels in the electric power sector is an important source of ancillary benefits. Significant NOX and SO2-emissions (see Fig. 4) are avoided while wind power in itself does not give rise to any of the emissions that are considered here. Still, the overall penetration of wind power is significantly lower in these flexible policy scenarios since these involve a greater range of cost-effective alternatives via permit trading. 4.3. Sensitivity analyses The above model simulation results indicate that the magnitude of the ancillary benefits is most significant in the policy scenarios that involve substantial CO2 emission reductions

A. Krook Riekkola et al. / Energy Policy 39 (2011) 4985–4998

4995

MEuro 250.0 200.0 VOC SO2 PM2.5 NOX

150.0 100.0 50.0 0.0

ExternE

CAFE WHO - Low

High

National

Low

High

National

Low

High

National

Low

-50.0

CAFE WHO - High

Fig. 7. Ancillary benefits in 2020 by pollutant: country-cap scenario under various CO2-reduction targets (million Euro2000).

MEuro 100.0 80.0 VOC SO2 PM2.5 NOX

60.0 40.0 20.0 0.0

ExternE

CAFE WHO - Low

Euro 99

Euro 44

Euro 22

Euro 99

Euro 44

Euro 22

Euro 99

Euro 44

Euro 22

-20.0

CAFE WHO - High

Fig. 8. Ancillary benefits in 2020 by pollutant: EU-scenario under various TEP-prices in EU ETS (million Euro2000).

domestically. For this reason it is of interest to investigate in more detail the role of the ancillary effects in the presence of different ambition levels for the required domestic reductions. Fig. 7 shows the estimated ancillary benefits by pollutant for 2020 in the Country-Cap scenario, and compares these to the corresponding impacts in the case of a lower and a higher reduction target, respectively. The base (national) case assumes a 40% reduction in CO2 emissions compared to the 1990 level, while the ‘‘low’’ and ‘‘high’’ cases build on the assumption of 20% and 53% emission reductions, respectively. The simulation results indicate that the absolute sizes of the estimated ancillary benefits are sensitive to the assumed CO2 reduction targets. For instance, the ancillary benefits increase by a factor of 1.5 as we increase the emission reduction target from 40% to 53%. This result is fairly independent of the source used for the external cost estimates. Fig. 7 also shows that the relative contribution of SO2 to the estimated total ancillary benefits increases with stricter CO2 reduction targets. This can be attributed to the increased substitution of biomass for oil for process heating within the agricultural and industrial sectors. In Fig. 8 we illustrate some impacts on the estimated ancillary effects in the presence of higher TEP prices in EU ETS. In this sensitivity analysis we pay attention only to results from the EU

scenario. This is in part motivated by the fact that in our base case (44 Euro2000 per tonne of CO2), the ancillary benefits were comparatively low compared to the other policy scenarios. In the sensitivity analysis we assume a TEP price of 22 and 99 Euro2000 per tonne of CO2, respectively (from 2015 and onwards), and the model simulations permit thus an assessment of whether (and, if so, by how much) these changes induce a significant change also in the estimated ancillary benefits as the relative economic cost of domestic CO2 reduction options is altered. The results in Fig. 8 show that with increases in the TEP prices the absolute size of the ancillary benefits increases since the economic viability of domestic reductions is improved (relative the option of purchasing permits). This effect is most pronounced when comparing the low TEP price of 22 Euro2000 per tonne of CO2 with our base case of 44 Euro per tonne of CO2. In the former case the ancillary benefits are even negative, that is in this scenario climate policy induces net increases in non-GHG emissions (most notably NOX and SO2) that in turn cause an increase in total system costs. The latter effect is partly a result of a higher electricity price with a TEP price of 22 Euro2000 per tonne of CO2 compared to 11 Euro (in the Baseline scenario), resulting, for instance, in more district heating (from biomass boiler) and less use of heat pumps in the residential sector.

4996

A. Krook Riekkola et al. / Energy Policy 39 (2011) 4985–4998

MEuro

200.0 150.0

VOC SO2 PM2.5 NOX

100.0

ExternE

CAFE WHO - Low

DR7

DR4

DR2

DR7

DR4

DR2

DR7

DR4

0.0

DR2

50.0

CAFE WHO - High

Fig. 9. Ancillary benefits in 2020 by pollutant: country-cap scenario under various discount rate assumptions (million Euro2000).

Finally, assumptions about the discount rate can affect the energy system in various ways, the most obvious being perhaps that the lifetime cost of new energy technologies will change and alter the merit orders for these. For instance, some technologies (e.g., wind power, nuclear energy, etc.) typically loose competitive ground from the use of higher discount rates (e.g., Pettersson and ¨ Soderholm, 2009). This is a result of the fact that the capital costs involved in the development of these technologies form a sizeable part of the total lifetime costs, and this may in turn influence the resulting ancillary effects. Fig. 9 shows the resulting ancillary effects in the Country-Cap scenario when three different discount rates are used: the base case of 4% and then two alternative discount rates of 2% and 7%, respectively. The results indicate that the estimated ancillary benefits appear to be only marginally affected by the use of alternative discount rates in the model simulations. Moreover, discount rate changes do not always imply monotonic changes in the magnitudes of the estimated ancillary benefits. It may be noted, for instance, that the total ancillary benefits decrease as the discount rate is raised from 4% to 7%. The main reason for this decrease is a reduction in avoided NOX damages, in part due to a reduction in wind power generation. With its high share of investment costs out of total lifetime costs, this technology becomes significantly more expensive with higher rate-of-return requirements. The most significant emission reductions of particulates and VOC are found with a discount rate of 2%, when new investments in biomass based heating is partly switched to heat pumps (the latter associated with higher investments costs). The SO2 reductions from the baseline are largest when comparing scenarios with a discount rate of 7% when coal based power plants are switched to less capital-intensive natural gas (with no SO2 emissions) when implementing CO2 targets, while for lower discount rates coal is instead replaced by biomass (which also gives rise to some SO2 emissions). When comparing these results to the corresponding results in the Sector-CapA scenario the results are overall very similar to the ones presented in Fig. 9 – i.e., marginal effects of the use of alternative discount rates – and the differences in total estimated ancillary benefits across these two scenarios (i.e., higher in the Country-Cap scenario) remain unaltered.

5. Concluding remarks In this paper we build on the well-established literature suggesting that GHG reduction policies, which create incentives

to alter the use of fossil fuels, can have important local and regional environmental impacts quite distinct from the global and longer-term benefits directly associated with avoided climate change. The paper has addressed a number of health-related improvements (ancillary benefits) that could accompany the reduction in CO2 under different climate policy designs in Sweden. These designs differ primarily in terms of how the country chooses to meet a specific target and where the necessary emission reductions take place. The reliance on a technology-rich energy system optimisation model of the Swedish energy system (TIMES-Sweden) has permitted us to address the economic significance of these environmental side-effects as well as to provide a detailed assessment of the respective technology and fuel choices that underlie any net changes in the estimated ancillary benefits. The results indicate that significant ancillary benefits accompany Swedish climate policy and they constitute a far from insignificant share of the increase in total system costs, thus reducing the overall cost of climate policy. Moreover, this share appears to be particularly significant in the scenarios that entail the largest emission reductions domestically. The latter results reflect the fact that since an increase in emissions reductions abroad also implies a lost opportunity of achieving important welfare gains from the reductions of a number of regional and local environmental pollutants. This shows, thus, that the notion of full flexibility in compliance measures (including geographical location) may not necessarily represent the most cost-effective strategy for an individual country. Still, in our case the estimated size of the ancillary benefits is overall not large enough to fundamentally alter the ranking of climate policy design in terms of system cost impacts. Nevertheless, under the model assumptions made the results tend to provide some support for the current Swedish government’s policy to partly restrict emission reductions abroad. While these overall results have been highlighted also in previous work, an important contribution of the present paper has been to investigate in more detail the technology choices underlying the aggregate figures, including a sensitivity analysis of the impact of policy target levels, permit prices and discount rates on the estimated ancillary benefits. The choice of bottom-up (technology-rich) energy system model facilitates these assessments. For instance, from an energy system perspective the results suggesting higher ancillary benefits in the policy scenarios restricting permit trading can be explained by the role of climate policy in inducing relatively large reductions of NOX emissions from the transport sector as well as reductions of SO2 in the

A. Krook Riekkola et al. / Energy Policy 39 (2011) 4985–4998

non-trading industry segments when substituting away from oil products to biomass. The model simulations also illustrate that the estimated ancillary benefits of climate policy in Sweden appear to be a non-linear function of the reduced CO2-emissions, both in terms of the ancillary benefits per reduced tonne of CO2 and as a share of the total system cost. This is explained by differences in the technology choices following each of the policy scenarios. Overall our findings illustrate the usefulness of analysing the ancillary benefits of climate policy with a bottom-up energy system model. In this paper we have, for instance, highlighted the allocation of biomass in the presence of the weight given to domestic emission reduction versus the use of permit trading. We also find important differences across policy scenarios with respect to wind power and oil use, in part resulting in significantly higher ancillary benefits from NOX and SO2 reductions in the scenarios involving a stronger focus on domestic emission reductions. The sensitivity analyses show that the absolute size of the ancillary benefits are sensitive to the assumed CO2 reduction target (in the domestic reduction scenario) and to changes in the permit price (in the permit trading scenario), but appear not to be heavily influenced by changes in the discount rate. Clearly additional research efforts are needed to shed more light on the issue of to what extent the ancillary benefits of climate policies could potentially offset a portion of the costs of these policies. For instance, in most studies the baseline issues probably deserve more attention. In our analysis we pay attention mainly to the impact of environmental policy and fuel taxes. Still, other baseline issues may be equally important to consider in more detail, including other policy issues (e.g., health policy) and non-policy issues such as technological change, transportation trends and demographic developments. In the Swedish context it is worth noting that a significant proportion of the estimated ancillary benefits appear in the transport sector, and limited diffusion of alternative fuels in the transport sector appears already in the baseline scenario. Clearly, however, the future evolution of vehicles and fuels is highly uncertain, not the least due to the competition for the biomass (in part illustrated in this paper), and the presence of significant network externalities in the fuel supply infrastructure. It is also useful to highlight in more detail the importance of domestic ancillary benefits following the implementation of JI and CDM projects. For instance, van Vuuren et al. (2006) show that Western European countries may experience domestic non-climate ancillary effects from JI projects in Eastern Europe, not the least through the reduction of reduced trans-boundary pollution (e.g., sulphur dioxide). Additional research on understanding the presence of ancillary benefits may not the least be motivated from a policy transition perspective. In national climate mitigation strategies there is a need to identify (no regrets) policy measures that generate important external benefits, and that for this reason become politically legitimate. The resulting reductions in non-GHG damages to health, crops and materials represent real economic benefits, and other side-benefits of the promotion of carbon-free energy sources include, for instance, improved security of supply and regional employment impacts. So far energy modelling studies have tended to pay most attention to analysing the impact of well-defined and uniform carbon taxes on the energy system, while fewer studies factor in the role of policy and institutional change in achieving these energy futures in practices.

Appendix A See Appendix Table A1.

4997

Table A1 Energy and fuel taxes in the baseline scenario. Sources: Swedish Tax Agency (2008a, 2008b). Energy tax

CO2-tax

SO2-tax

Unit

9.33 9.33 9.33 8.78 7.62 6.28 10.70

Euro/GJ Euro/GJ Euro/GJ Euro/GJ Euro/GJ Euro/GJ Euro/GJ Euro/GJ Euro/GJ Euro/GJ Euro/GJ Euro/GJ Euro/GJ Euro/GJ Euro/GJ

Non-transportation Residual fuel oil Heavy fuel oil—class 1 Heavy fuel oil—class 2 Heavy fuel oil—class 3 Liquefied petroleum gas Natural gas Hard coal, coke Peat Municipal solid waste

2.47 4.13 4.95 5.06 0.38 0.72 1.39 0 0.03

0.63

0 0 0.33 0.32 0 0 0.61 0.56 0

Transportation Motor spirit/gasoline Diesel oil Kerosene—jet fuels Liquefied petroleum gas Natural gas Ethanol/methanol (bio)

10.42 4.13 4.13 0 0 0

8.25 9.33 9.33 3.93 3.86 0

0 0 0 0 0 0

Electricity consumption Non-industry Industry

8.70 0.15

Nuclear power Installed capacity

16.9

Euro/GJ Euro/GJ Euro/kW and Year

References Azar, C., Sterner, T., 1996. Discounting and distributional considerations in the context of global warming. Ecological Economics 19 (2), 169–184. ˚ Flodstrom, ¨ ¨ ¨ Stationar ¨ Bostrom, C.-A., E., Cooper, D., 2004. Emissionsfaktorer for ¨ ¨ Forbr anning (Emission Factors in Stationary Combustion). Report nr 3-2004, Swedish Environmental Emissions Data (SMED). Boyd, R., Krutilla, K., Viscusi, W.K., 1985. Energy taxation as a policy instrument to reduce CO2 emissions: a net benefit analysis. Journal of Environmental Economics and Management 29, 1–24. Burtraw, D., Krupnick, A., Palmer, K., Paul, A., Toman, M., Bloyd, C., 2003. Ancillary benefits of reducing air pollution in the US from moderate greenhouse gas mitigation policies in the electricity sector. Journal of Environmental Economics and Management 45 (3), 650–673. Bye, B., Kverndokk, S., Rosendahl, K.E., 2002. Mitigation costs, distributional effects and ancillary benefits of carbon policies in the Nordic countries, the UK and Ireland: a survey of model results. Mitigation and Adaption Strategies for Global Change 7, 339–366. Carle´n, B., 2004. EU’s Emissions Trading System in the Presence of National Emission Target. Working Paper 2004:16, Department of Economics, Stockholm University. Das, A., Rossetti di Valdalbero, D., Virdis, M.R., 2007. ACROPOLIS: an example of international collaboration in the field of energy modelling to support greenhouse gases mitigation policies. Energy Policy 35 (2), 763–771. Davis, D.L., Krupnick, A., McGlynn, G., 2000. Ancillary benefits and costs of greenhouse gas mitigation—an overview. Paper presented at the IPCC Expert Workshop on Assessing the Ancillary Benefits and Costs of Greenhouse Gas Mitigation Policies, Washington, DC, March 27–29. Ekins, P., 1996. The secondary benefits of CO2 abatement: how much emission reduction do they justify? Ecological Economics 16 (1), 3–24. ˚ Nya Anlaggningar ¨ Elforsk, 2007. El Fran 2007. Elforsk Report 07:50, Stockholm. European Climate Exchange (ECX), 2009. Prices and Volume: ECX EUA Futures Contract (22 April 2005–17 March 2009). Internet: /www.ecx.eu/EUA-FuturesS. Garqiulo, M., 2008. RES 2020. The modeling approach in RES2020 and the data used for the potential of renewable energy sources in EU27. Presentation at the RES2020 International Workshop, 10–11 November, Prague. Internet: /www.res2020.euS. ˚ Government Bill 2008/09:162. En sammanhallen klimat-och energipolitik— klimat, Ministry of the Environment, Government of Sweden, Stockholm. Government Offices of Sweden, 2009. A Sustainable Energy and Climate Policy for the Environment. Competitiveness and Long-term Stability, Swedish Prime Minister’s Office, Internet: /www.sweden.gov.se/sb/d/2031/a/120088S. Holland, M., Watkiss, P., 2007. Benefits Table Database: Estimates of the Marginal External Costs of Air Pollution in Europe. Version E1.02a. Hourcade, J.-C., Shukla, P.R., Cifuentes, L., Davis D., Emonds, J., Fisher, B., Fortin, E., Golub, A., Hohmeyer, O., Krupnick, A., Kverndokk, S., Loulou, R., Richels, R., Segenovic, H., Yamaji, K., 2001. Global, Regional, and National Costs and Ancillary benefits of Mitigation. Climate Change 2001: Mitigation. Intergovernmental Panel on Climate Change, Third Assessment Report (TAR). Cambridge University Press, Cambridge, UK, pp. 499–559.

4998

A. Krook Riekkola et al. / Energy Policy 39 (2011) 4985–4998

Intergovernmental Panel on Climate Change (IPCC), 2007. Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to the Fourth Assessment Report of the Inter-governmental Panel on Climate Change, Geneva. ˚ ¨ ˚ Kageson, P., 2007. Miljobilar och Biodrivmedel—Hur Paverkas Sverige av EU:s Direktiv? Vinnova Report VR 2007:10, Stockholm. Klaassen, G., Riahi, K., 2007. Internalizing externalities of electricity generation: an analysis with MESSAGE-MACRO. Energy Policy 35 (2), 815–827. ˚ EU:s Krook Riekkola, A., Ahlgren, E., 2003. Externaliteter—En Analys utifran ¨ Etapp 2, Profu i Goteborg ¨ ExternE-Resultat. Nordleden 2003. Slutrapport for AB, Gothenburg, Sweden. Kuosmanen, T., Bijsterbosch, N., Dellink, R., 2009. Environmental cost–benefit analysis of alternative timing strategies in greenhouse gas abatement: a data envelopment analysis approach. Ecological Economics 68 (6), 1633–1642. Loulou, R., Remne, U., Kanudia, A., Lehtila, A., Goldstein, G., 2005. Documentation for the TIMES Model—PART I, Internet: /www.etsap.org/documentation.aspS. Morgenstern, R.D., 2000. Baseline issues in the estimation of the ancillary benefits of greenhouse gas mitigation policies. Paper presented at the IPCC Expert Workshop on Assessing the Ancillary Benefits and Costs of Greenhouse Gas Mitigation Policies, Washington, DC. Nilsson, C., Huthala, A., 2000. Is CO2-Trading Always Beneficial? A CGE-model Analysis on Secondary Environmental Benefits, Working Paper. National Institute of Economic Research, Stockholm. Nordel, 2008. Annual Statistics 2007. Nordel’s Statistics Report, Internet: /www. entsoe.eu/resources/publications/nordic/annualstatistics/S. OECD, 2002. Ancillary Benefits and Costs of GHG Mitigation: Policy Conclusions, ENV/EPOC/GSP(2001)13/FINAL, Paris. ¨ stblom, G., Samakovlis, E., 2004. Cost of Climate Policy when Pollution Affects O Health and Labour Productivity: A General Equilibrium Analysis Applied to Sweden. Working Paper No. 93, Swedish National Institute of Economic Research. ¨ Ostblom, G., Samakovlis, E., 2007. Linking health and productivity impacts to climate policy costs: a general equilibrium analysis. Climate Policy 7 (5), 379–391. ¨ Pettersson, F., Soderholm, P., 2009. The diffusion of renewable electricity in the presence of climate policy and technology learning: the case of Sweden. Renewable and Sustainable Energy Reviews 13 (8), 2031–2040.

Repetto, R., Austin, D., 1997. The Costs of Climate Protection: A Guide for the Perplexed. World Resources Institute, Washington, DC. RES2020, 2009. EU27 Synthesis Report. The European Commission’s Research Program: Intelligent Energy. Project no: EIE/06/170/SI2.442662. Deliverable D.4.2. Internet: /www.res2020.euS. ¨ Soderholm, P., Pettersson, F., 2008. Climate policy and the social cost of power generation: impacts of the Swedish national emissions target. Energy Policy 36 (11), 4154–4158. Statistics Sweden, 2007. Electric Supply, District Heating and Supply of Natural and Gasworks Gas 2005. Statistical Report EN11SM0701, Stockholm. Statistics Sweden, 2008. Statistical Data of Air Borne Emissions by Sector. Internet: /www.scb.seS. Stern, N., 2006. The Economics of Climate Change, The Stern Review. Oxford University Press. Swedish Energy Agency, 2006. Energy in Sweden Facts and Figures 2006. Internet: /www.swedishenergyagency.seS. Swedish Energy Agency, 2007. The Electricity Certificate System 2007. ID-No ET2007:27. Internet: /www.swedishenergyagency.seS. ¨ ¨ Swedish Environmental Protection Agency, 2008. Forslag till ny Forordning om ¨ ¨ Utomhusluft (Proposal for New Regulation of Miljo-Kvalitetsnormerna for Environmental Standards for Outdoor Air Quality). Report 5884, Stockholm. Swedish Environmental Protection Agency, 2009. Thermal Values and Emission Factors Energy CLRTAP sub2009. Appendix 2 to National Inventory Report 2009 Sweden, Stockholm. Internet: /www.naturvardsverket.se/sv/Tillstande t-i-miljon/UtslappsdataS. Swedish Institute for Transport and Communications Analysis (SIKA), 2002. ¨ vergripande kalkylparametrar). Figures Underlying Economic Calculations (O Report 2002:7. Internet: /www.sika-institute.seS. Swedish Tax Agency, 2008a. Tax Rates in 2009 (Skattesatser 2009). Internet: /www.skatteverket.seS. Swedish Tax Agency, 2008b. Selective Purchase Taxes, SKV 505 Edition 21 ˚ (Punktskatter, SKV 505 utgava 21). Internet: /www.skatteverket.seS. Van Vuuren, D.P., Cofala, J., Eerens, H.E., Oostenrijk, R., Heyes, C., Klimont, Z., den Elzen, M.G.J., Amann, M., 2006. Exploring the ancillary benefits of the Kyoto protocol for air pollution in Europe. Energy Policy 34 (4), 444–460.