The Economic, Energy, and Emissions Impacts of Climate Policy in South Korea

Using an economy-wide model, we evaluate the impact of policies to meet South Korea’s Paris pledge to reduce greenhouse gas (GHG) emissions by 37% relative those under business as usual (BAU) in 2030. Simulated BAU emissions in 2030 are 840.8 million metric tons (Mt) of carbon dioxide equivalent (CO2e), indicating that economy-wide emissions should be constrained to 529.7 MtCO2e. Under South Korea’s Emissions Trading System (KETS) and fuel economy standards, a 2030 carbon price of $89/tCO2e is needed to meet this goal. Without considering benefits from avoided climate damages, these policies reduce 2030 GDP by $20.6 billion (1.0%) and consumer welfare by 7.9 billion (0.7%). Comparing this scenario to one where South Korea’s Paris pledge is met solely by an ETS, indicates that adding a fuel economy standard reduces GDP and welfare by, respectively, $4.2 billion and $1.1 billion. Declines in sectoral production are largest for fossil-based energy sectors and the chemical, rubber and plastic products, and iron and steel sectors. .


Introduction
how interpretations of a target can vary depending on a focus on a reduction from BAU or an absolute reduction from an historic year. Climate Action Tracker (2017) also points out that 2030 emissions will be 81% above 1990 levels, reflecting very rapid growth between 1990 and 2012.
Projections given current policies indicate that South Korea's emissions will remain above the 2020 target, and additional new policies will be needed to achieve the 2030 goal. The Government indicated that a 25.7% reduction will be achieved domestically and a further 11.3% reduction will be achieved by international market mechanisms (Climate Action Tracker, 2017). To meet these goals, South Korea launched three major policies: a Target Management Scheme (TMS), an Emission Trading Scheme (ETS), and 2020 Corporate Average Fuel Economy (CAFE) Regulations. The TMS is a precursor to the ETS, with lower penalties for non-compliance. CAFE regulations were introduced in 2014, with targets to be fully phased in by 2020.
In this paper, we develop and deploy a custom-made economy wide model to evaluate the impacts of key climate policies in South Korea in 2030. Several studies focus on the global implications of the Paris agreement (Fawcett et al. 2015, Aldy et al. 2016, Vandyck et al. 2016, and Jacoby et al. 2017. While these studies provide regional details, they typically do not report results for South Korea or assume cost-minimizing attainment of the NDC emissions goals, rather than specific policy proposals. To our knowledge, our study is the first detailed economy-wide analysis of South Korea's 2030 NDC emissions pledge. This paper has four further sections. Section 2 provides a further overview of key climate policies in South Korea. Section 3 describes the structure and data sources for our economy-wide model and the scenarios implemented in our analysis. Our results are presented and discussed in Section 4. Section 5 concludes.

Climate policy in South Korea
Key climate policy legislation in South Korea includes the Korean Emissions Trading Scheme (KETS) and fuel economy standards. The long-term goal of the KETS is to reduce 2030 emissions by 37% relative to business as usual (BAU). The foundations for the capand-trade system were set by The Framework Act on Low Carbon, Green Growth Sectors covered by the KETS include (1) electricity, (2) industry (e.g., mining, oil refining, food and beverages, cement, steel, non-ferrous metals, automobiles, shipbuilding, electronic equipment), (3) building (including telecommunication), (4) domestic aviation, and (5) public waste treatment. GHGs included in the KETS include emissions of carbon dioxide (CO2), methane (CH4), nitrous oxide (N2O), hydrofluorocarbons (HFCs), and sulfur hexafluoride (SF6) from energy, industrial processes, product use and waste.
Emissions from land use, land-use change and forestry (LULUCF) are not currently included in the KETS.
In Phase I, there will be 100% free allocation of allowances, with allocations based on firm activities in 2011-2013. In subsequent phases, a proportion of allowances will be auctioned, with at least 10% of allowances auctioned in Phase III. Energy-intensive, tradeexposed sectors will receive 100% of their allowances for free in all phases. Banking of emission permits is allowed without any restrictions. Borrowing permits is only allowed within each phase, with restrictions on the amount that can be borrowed. In all phases, up to 10% of emission rights can be sourced from outside the ETS in the form of offsets. In Phases I and II, only domestic offset credits can be used for ETS compliance. In Phase III, up to 50% of offsets can be sourced internationally (i.e., international offsets can contribute up 5% of the total number allowances submitted for compliance obligations).
In our analysis, we evaluate the KETS in 2030 under the stated objective of reducing business as usual emissions by 37% in this year. Our representation of the policy in 2030 is guided by legislation for Phase III of the KETS, or where specifics for Phase III are yet to be set, the latest year for which legislation has currently been set. We do not consider banking and borrowing of emission permits in our analysis as evaluating these mechanisms would require emissions caps for each year out to 2030. The representation of the KETS in our modeling framework is described in Section 2.
South Korea's Ministry of Environment introduced new fuel economy and GHG standards for passenger cars, buses with a maximum seating capacity of 15 or fewer persons, and trucks that weigh less than 3.5 tons (ICCT, 2015). The regulations will be phased-in from 2016 to 2020. They require a 30.7 percent reduction in the fleet average South Korea's targets are among the most stringent in the world (Figure 1). The European Union (EU) at 95 gCO2/km emission is nominally the most stringent, however, an important consideration is how the test standard for fuel economy. The EU standard produces significantly lower emissions than achieved under average actual driving conditions. South Korea uses the US test standards which are more closely calibrated to actual driving averages. Factoring in the test standard difference makes South Korea's standard more stringent than Europe's.
Climate Action Tracker (2017) concludes that current policies in South Korea will be insufficient to meet the NDC target, and hence further tightening of policies would be required to meet it. South Korea has not set fuel economy targets beyond 2020 at this point.

Modeling Framework
Our analysis develops a bespoke multisector applied general equilibrium model of  Table 1. The model represents 13 sectors related to energy extraction, production and distribution, including eight electricity generation technologies. Transportation is represented by separate commercial and household transportation (transportation in ownsupplied vehicles and household purchases of commercial transportation). The model also represents 13 manufacturing sectors and five non-transportation service sectors. In each sector, there is a representative firm that produces output by hiring primary factors and purchasing intermediate inputs from other firms. Production in each sector is represented by a multi-level nest of constant elasticity of substitution (CES) functions.
Nesting structures for sectoral groups are outlined in Figures 2 to 5. All sectors except fossil fuel extraction, electricity production and agriculture are built on the production structure described in Figure 2. A key feature of the production nest is substitution between aggregate energy and a capital-labor composite, which allows price-induced improvements in energy efficiency. Other opportunities to abate emissions are provided by the ability to substitute between electricity and (in aggregate) non-electricity energy, and among non-electricity energy inputs (coal, gas, and refined oil). The top-level nest combines nonenergy intermediate inputs with the energy-value added composite using a Leontief aggregation. Mining activities, including fossil fuel extraction sectors, are produced by a CES aggregate of a sector-specific resource (e.g. coal resources for the coal sector) and a composite of capital, labor and intermediate inputs ( Figure 3).
In fossil-based electricity sectors (Figure 4a), there is substitution between fuel inputs and a capital-labor aggregate to capture price-induced improvements in energy conversion efficiency. There is also the potential for fuel-switching within each fossil-fuel electricity sector, but substitution among fuels is limited by the small (or zero) share of other fossil fuels used in each fossil electricity sector. 2 A key characteristic of non-fossil electricity sectors is the aggregation of a technology specific factor and (aggregated) other inputs in the top level of each production nest ( Figure   4b). For nuclear electricity and hydroelectricity, which are largely determined by regulations, the top-level elasticity is set equal to zero. This feature allows output for these sectors to be assigned exogenously using estimates from external sources. For other nonfossil electricity sectors (wind, solar, and other electricity), top-level elasticity values capture constraints due to intermittency and resource availability, while at the same time allowing production of these technologies to respond to price changes. To produce supplied electricity (which is purchased by firms and consumers), fossil electricity types and nonfossil electricity outputs are combined using separate CES functions, and the two aggregates are combined using a further CES function ( Figure 4c). In this nesting structure, non-fossil electricity sources are perfect substitutes for each other, and aggregate fossil fuel electricity is a perfect substitute for non-fossil electricity ( = = ∞).
Aggregate electricity is combined with transmission and distribution in a Leontief nest.
A representative agent derives income from selling factor services and allocates expenditure across private consumption, government consumption, and saving/investment.
The nesting structure for final consumption is outlined in Figure 5. Important features of the specification include substitution among goods with different GHG intensities and a detailed representation of household transportation. The household transportation specification allows substitution between purchased transportation (supplied by the commercial transport sector) and own-supplied transportation. Building on Karplus et al. (2013), own-supplied transportation distinguishes old and new vehicles, and breaks down new vehicles into several components based on the services they provide. Importantly, substitution between powertrain capital and refined oil inputs used in new vehicles reflects the scope for consumers to purchase more fuel efficient cars, at a higher cost, in response to increases in the price of fuel and/or policies. A mandated maximum amount of fuel per kilometre of travel services (to represent a fuel economy standard) is specified using a powertrain certificate system. In this system, powertrain capital used for the production of new cars is allocated one certificate per unit of output, and both powertrain capital and refined oil require certificates to be used in new car production. Under this system, the share of fuel inputs in transport services from new cars is (1 − ) , where is the share of the fuel-powertrain aggregate in transport services from new cars. As the fuel-powertrain aggregate is combined with other inputs using a Leontief function, is fixed and a fuel economy standard can be imposed by setting the value for . If a fuel economy standard is not simulated, the powertrain certificate system is turned off in the model.
A government sector collects taxes and provides subsidies, and purchases good and services. Net fiscal deficits and, where applicable, revenue from the sale of emission permits are passed to consumers as (implicit) lump sum transfers. Although the model is static, investment is included as a proxy for future consumption and is a fixed proportion of expenditure by each regional household. Elasticity values in production and consumption that, in tandem with input cost shares, govern substitution possibilities are guided by those used in the MIT Economic Projection and Policy Analysis (EPPA) model (Paltsev et al., 2005. International trade in goods and services follows the 'Armington approach' that assumes goods are differentiated by country of origin (Armington, 1969). Specifically, for each commodity, domestic production is differentiated from imports using a CES function.
Values for elasticities of substitution in the trade specification are sourced from Hertel et al. (2007). Also for each commodity, production is allocated across the domestic market and exports using a constant elasticity of transformation function.
Turning to closure, factor prices are endogenous and there is full employment; capital and labor are mobile across sectors (and technology/sector specific resources are immobile); and the current account deficit is a fixed proportion of GDP.
The model is calibrated using the Global Trade Analysis Project (GTAP) Power Database (Peters, 2016). This database augments version 9 of the GTAP Database (Aguiar et al., 2016) and includes economic data and CO2 emissions from the combustion of fossil fuels for 140 regions and 68 sectors. We extract the data for South Korea and aggregate the sectors to those listed in Table 1 by extending tools provided by Lanz and Rutherford (2016). We also augment GTAP-Power with data on non-CO2 emissions from Irfanoglu The model is formulated and solved as a mixed complementarity problem using the Mathematical Programming Subsystem for General Equilibrium (MPSGE) described by Rutherford (1995) and the Generalized Algebraic Modeling System (GAMS) mathematical modeling language (Rosenthal, 2012) with the PATH solver (Dirkse and Ferris, 1995).  (5) improvements in total factor productivity. In the ETS scenario, we implement an emissions trading system across covered sectors that reduces economy-wide emissions in South Korea by 37% relative to those in the BAU scenario (i.e., meets South Korea's 2030 NDC pledge). The ETS includes 30 of the 35 sectors in the model (see Table 1), and covers all GHG emissions with trading across gases and sectors. 3 As the ETS targets a reduction in economy-wide emissions but does not cover all sectors, the emissions cap on covered sectors is chosen endogenously in the model to target a desired level of economy-wide emissions.

Scenarios
As noted above, up to 10% of emissions rights can be sourced from outside the ETS.
We model the supply of domestic offsets using a (secondary) cap-and-trade program for all sectors not included in the ETS (except fossil fuels purchased by households) with a BAU emissions cap. Under this system, emissions reductions by non-KETS sectors below the BAU level create domestic offsets that can, up to a certain limit, be used for compliance with the KETS. We assume that international offsets are available at a fixed price of $5/tCO2e.
We model the use of total offsets and international offsets with two (pseudo) certificate schemes to ensure that the limits on offset use are enforced. Specifically, to use an (domestic or international) offset to meet ETS obligations for one ton of emissions, an entity must turn in an offset credit for one ton of emissions and one 'offset certificate'. To use an international offset, in addition to providing an offset credit for one ton of emissions and an 'offset certificate', an entity must hand over one 'international certificate'. On the supply side, certificates are 'produced' in fixed proportions with ETS permits: for each permit for one ton of emissions there are 'offset certificates' and 'international certificates'. By setting = 0.1 and θ = 0.5, we impose the upper limits on the use of total offsets and international offsets set out in the KETS (offsets can be used to meet up 10% of obligations to surrender emissions rights, and international offsets can contribute up 50% of the total amount of offsets). As the quantity of ETS emissions is endogenous, maximum limits on the supply of offsets (in tons) are also endogenous in the model.
In the ETS-Vehicles scenario, we impose a fuel economy standard for new vehicles in addition to the policies in the ETS scenario. Current legislation mandates fuel consumption per kilometer for cars of 31.1% by 2020 relative to a 2013 baseline. We assume the fuel economy standard continues to be tightened beyond the 2020 mandate by extending to 2030 the same annual rate of improvement between 2013 and 2020. This results in an estimated reduction in fuel consumption per kilometer of 59.5% relative to 2013.

Results
A summary of results for each scenario is presented in Table 3, with additional results reported in Table 4 (GHG emissions by gas),   As noted above, several emissions sources not covered by the ETS are included in a secondary cap-and-trade program to approximate a domestic offset market (with a BAU emissions cap and a limit on the quantity of permits that can be sold to ETS sectors). This mechanism decreases emissions from sources not covered by the ETS by 5.7% relative to BAU. The price of domestic offsets in the ETS scenario is equal to the price of international offsets ($5/tCO2e). This is because the use of international offsets (13.1 MtCO2e or 44.8% of the total quantity of offsets) is less than the maximum amount allowed (50% of the quantity of offsets). That is, at a fixed price of $5/tCO2, 55.2% of the total allowed offsets is sourced domestically. Note: * Primary energy from nuclear is based on the amount of heat generated in reactors assuming a 33% conversion efficiency. For wind, solar and hydro, the primary energy equivalent is the physical energy content of electricity generated.
Although private transportation is not included in either the ETS or the domestic offset cap-and-trade program, there is an improvement in the fuel economy of new cars in the ETS scenario. This is because the price of crude oil is fixed in the model and the inclusion of oil refining in the ETS increases the price of refined oil, causing a price-induced substitution toward more fuel-efficient vehicles. The ETS reduces the number of new car sales by 12,297 (0.5%) relative to BAU.
Changes in output relative to BAU for the ETS and ETS-Vehicles scenarios are reported in Table 7. In the ETS scenario, output for all sectors decreases except for lowcarbon electricity sectors without regulatory constraints (solar electricity, wind electricity, and other electricity). The largest proportional reductions in output occur in energy sectors (e.g., coal electricity generation). Among non-energy sectors, the chemical, rubber & plastic products sector experiences the largest proportional output decrease (6.6%).
Absolute decreases in output are largest for chemical, rubber & plastic products; iron and steel; refined oil products; and coal electricity. The output decrease for motor vehicles and parts (0.6%) is small relative to those for other sectors, as the GHG intensity for this sector is relatively low. Consequently, the negative impact of rising energy costs is partially offset by reduced demand for capital and labor from most other sectors. Changes in exports relative to BAU (not reported) follow a similar pattern to changes in output. Proportional exports decreases are largest for chemical, rubber and plastic products (8.6%) and iron and steel (9.4%), and the decline in exports of motor vehicles and parts is 0.6%. In the ETS-Vehicles scenario, output changes for all sectors except motor vehicles and refined oil are similar to those in the ETS scenario. The fuel economy standard has two opposing impacts on the output of the motor vehicles sector. First, by increasing the costs of new cars, the standard decreases the demand for motor vehicles (as illustrated by the decline in new car sales). Second, for each kilometer of travel, the standard forces consumers to spend more on powertrain capital and less on fuel, which increases demand for outputs from the motor vehicles sector. The 2.3% increase in motor vehicle output relative to BAU in the ETS-Vehicles (compared to a 0.6% decrease in the ETS scenario) indicates that the powertrain-share effect dominates the cost effect. The forced decrease in fuel expenditure per kilometer traveled in the ETS-Vehicles scenario leads to a larger decrease in output of refined oil products relative to BAU (6.7%) than in the ETS scenario (6.1%).  -5,580 -0.5 -6,279 -0.5 -4.6 -0.4 In the ETS-All scenario, as expected, the carbon price ($62.9/tCO2) and the welfare cost of meeting the NDC goal is lower than in other policy scenarios. Due to the lower carbon price, relative to other ETS scenarios, there is more electricity from coal and gas and less from wind and solar. There are smaller reductions in the output of energy-intensive sectors included in the KETS (e.g., relative to BAU, output of chemical, rubber and plastic products decreases by 4.9% in the ETS-All scenario and 6.6% in the ETS scenario).
Conversely, there is a large reduction in the output of the commercial transportation sector, an energy-intensive sector that was excluded from carbon pricing in other ETS scenarios.
Relative to the ETS scenario, output of motor vehicles and parts falls in the ETS-All scenario as the inclusion of household fuel purchases in the ETS reduces the demand for vehicles.

Conclusions
South Korea's NDC has been rated as "Inadequate" by the Climate Action Tracker (2017) on the basis that the proposed 2030 target would allow emissions to be more than double the 1990 level. In contrast, the EU is rated "Insufficient", a higher rating. These ratings are against what is required to achieve the 1.5 or 2.0 degrees C target of the Paris In some previous studies comparing vehicle standards and cap and trade, a quite large additional cost was incurred when vehicle standards were imposed Rausch and Karplus, 2014). These studies focused on costs of standards in the EU and the US, generally for the period up to about 2030, similar to our South Korea study. However, in the case of South Korea, we see some additional cost but much less than in these previous studies even though the basic structure of the model applied here is similar. There are a few reasons for this difference. First, the scenario construction among these studies was different. Rausch and Karplus (2014) compared achieving a given reduction via technology standards for vehicles and power generation versus achieving the same emissions level with an economy wide cap. \That both broadens the coverage from two sectors to the entire economy and substitutes a more efficient policy mechanism. Here we are comparing a cap and trade that does not include transportation with a policy that adds transportation via fuel standards. In principle by broadening the policy to include more sectors, it should reduce the cost, but instead the cost is increased-clear evidence of an inefficiency (partly because of the mechanism itself, and possibly also because the standard is more stringent than it should be). The Paltsev et al (2016)  A second reason for the smaller cost add-on is due to the fact that the carbon price needed to meet the target for South Korea is much higher than the near-term carbon price in, for example, the EU. The South Korean carbon price level in 2030 would, by itself, create incentives to adopt more efficient vehicles, and so the fuel standard is pushing in the right direction, albeit inefficiently. Paltsev, et al (2015), using a recursive dynamic model, and comparing regulatory approaches to cap and trade demonstrated that as the GHG price rose, eventually the added cost of the fuel standards largely disappeared.
Of course, our estimates depend on the added vehicle costs related to improving efficiency, relative to the cost of abating elsewhere in the economy, but our formulation here is similar to that in the earlier cited studies.
Our results show impacts on sectors generally as expected: fossil fuel dependent electricity production drops substantially, especially coal, and non-fossil electricity expands. Production in nearly all other sectors declines. The decline is in the range of 4.5 to 6.5% for more energy intensive sectors such as iron & steel and chemicals, rubber & plastics. For other less energy intensive sectors the decline in production is on the order of 0.5 to 2.0%. The value of production of motor vehicles and parts is one sector that actually increases under the fuel standards scenario. While we estimate that the number of vehicles sold falls, the cost of vehicles rise and so the total value of sales actually increases by a bit over 2% whereas under the cap and trade the value of vehicle sales fall by about 0.6%.
Vehicle sales and the value of sales falls when the sector's emissions are covered by the cap and trade measure instead of the vehicle standards.