Carbon Leakage in the context of increasing the EU greenhouse gas emission reduction targets – the ways the EU and global emission behave and what influences its scale

Table 1 shows the different scenarios analysed together with the respective emission reduction targets assuming for the EU States.

Important assumption of the study is that in order to better reflect and analyse the channels of carbon leakage phenomenon, we have not included emission reduction targets (NDCs) for the other regions/states (outside the EU). This should be borne in mind while analysing the results.

Scenarios assume implementation of the EU’s climate policy targets for GHG’s emission reductions by 20% in 2020 and by 40% in 2030 relative to 1990. These targets concern emission from all sectors of the economy.

Total GHG’s emission reductions were split between the sectors covered by the EU ETS and non-ETS. In the EU ETS, emission will be reduced by 21% in 2020 and by 43% in 2030 respectively, relative to 2005. Sectors included in EU

ETS (in EU-28, and three EFTA countries) have to surrender enough emission allowances to cover all its emission. The amount of emission allowances available in the market in the EU ETS results directly from the adopted reduction target.

The non-ETS sectors would need to cut emission by 10% in 2020 and by 30% in 2030, relative to 2005. Distribution of the non-ETS reduction target between member states was set according to EC Decision 2009/406 /EC

For details, see EC Decision 2009/406 /EC.

Figure 1

The target in 2020 of 20% cut in greenhouse gas emission (from 1990 levels) stay in GHG45/MSR and GHG45/MSR/ ETP scenarios unchanged. However, we assumed an increase of the GHG emission reduction target in 2030 to 45% for the EU compared to 1990. For the EU ETS sectors, the new GHG’s emission reduction target in 2030 was set at the level of 48% compared to 2005 and for the non-ETS sectors at the level of 36% compared to 2005.

See Ingvild Sorhus (2018).

In Figure 2, we have presented the redistribution of emission reduction target -36% in non-ETS sectors between Member States compared to previous scenarios with -30% target.

Figure 2

We have assumed that the reduction of volume of GHG’s emission allowances in the EU ETS due to the implementation of MSR will have an impact on the achievement of the emission reduction targets in 2020 and 2030 (in the scenarios with MSR). Based on the emission projections from the EU Reference Scenario 2016, we estimated that by 2020, around 760 million emission allowances (EUA’s) will be transferred from auction and put into the MSR. In the next period, the MSR will have a slightly less effect on the market – during the period 2021–2030, around 690 million allowances will be transferred to the reserve.

There is a very small difference (less than 1%) between the scenarios with different targets in EU ETS (-43% and -48% in 2030 compared to 1990) in the number of allowances transferred to the MSR.

The number of allowances withdrawn from the market is roughly equal to the number of allowances that will be sold in auctions during the two-year period in the EU ETS. The transfer from auctions to the MSR during the periods, such a large number of emission allowances will correspond to an annual reduction of the allowances in the EU ETS, on average 95 million in the period 2013–2020 and 69 million in the period 2021–2030 which, for example, would reflect an increase of the reduction targets in the EU ETS, only additionally in 2020 by 4% and in 2030 by 3%.

The analysed policy options (scenarios) were done on two versions of the model:

Simplifying, in terms of climate protection, carbon leakage is defined as ‘the increase in GHG’s emission outside a region as a direct result of the policy to cap emission in this region’.

See Julia Reinaud (2008)

Where:

CL_i – carbon leakage rate in sector i,

ΔEαi$\Delta {{\text{E}}^{\alpha }}_{i}$– change (decrease) the GHG’s emission in a region α and sector i, where climate policy is present,

ΔEβi$\Delta {{\text{E}}^{\text{ }\!\!\beta\!\!\text{ }}}_{i}$– change (increase) the GHG’s emission in a region

β and sector i, where no climate policy is present or the activities to reduce emission are negligible.

Carbon leakage can be also decomposed into channels distinguished by a specific driving factor. Three main channels have been identified:

Competitiveness channel has so far received the most attention in the literature. Through this channel, the competitiveness of the energy intensive industries would be weakened due to more stringent carbon mitigation policies. This can induce the potential relocation of the affected industries into the non-abating regions. Impact on industrial competitiveness will be visible through changes in the trade patterns

Analysis by Babiker (2001) highlights the importance of the trade channels over capital mobility.

Carbon intensity channel refers to decline of the world fossil fuel prices induced by the fall in demand for fossil fuels in the abating countries. Lower fossil fuel prices might induce an increase of the fossil fuel based energy demand in the non-abating regions. Literature does not appear conclusive on the strength of this channel as also strategic interactions might play an important role such as decisions of the fossil fuel supplying countries, see, for example, Criqui and Mima (2012).

Using an index number decomposition

There is a broad literature considering suitability of several index number forms for decomposition, for an overview see, for example, Agn et al. (2003). Our decomposition is based on Tan et al. (2018).

Where:

ΔC_i – change in carbon emission,

ΔEX_i – change in exports,

ΔIM_i – change in imports,

ΔDD– change in domestic demand,

Clisc$Cl_{i}^{sc}$_i – carbon intensity in the analysed scenario,

Clib$Cl_{i}^{b}$– baseline carbon intensity in the analysed scenario,

Qisc$Q_{i}^{sc}$– GDP in the analysed scenario,

Qib$Q_{i}^{b}$– baseline GDP.

In that framework, the three channels are calculated as follows:

Moreover, carbon intensity channel may be further decomposed into changes in energy demand and changes in the structure of the economy. Therefore, changes in internal production structure (such as shifts between different domestic commodities) are also included, as well as changes in the structure of use of fossil fuels or changes in emission intensity of fuels as such.

Consequently, the carbon intensity component of change in emission can be further decomposed to changes in sectoral structure and energy intensity, using additive logarithmic mean divisia index decomposition, see Ang et al. (2013):

Cs,ibandCs,isc$C_{s,i}^{b}\,\text{and}\,C_{s,i}^{s\,c}$– sectoral carbon emission in baseline and scenarios,

CIs,ibandCIs,isc$CI_{s,i}^{b}\,\text{and}\,\,CI_{s,i}^{sc}$– carbon intensities in baseline and scenarios,

shs,ibandshs,isc$sh_{s,i}^{b}\,\,\text{and}\,\,\,sh_{s,i}^{sc}$– shares of sector in output (GDP) in baseline and scenarios,

L(x,y)=x−ylnx−lny$L\left( x,y \right)=\frac{x-y}{\ln x-\ln \,y}$is the logarithmic average.

The first component of the sum is a part of the change in carbon intensity that is due to the changes in the share of a given sector in output and the second is the contribution of changes within sector carbon intensity. Such decomposition will allow to say whether the changes in aggregate carbon intensity are achieved due to the changes in sectoral structure of the economy or due to the efforts of enterprises either to reduce energy use or change in the internal use of fuels.

In general, under the set assumptions, we expect a fall in the EU total GHG’s emission until 2030 in all scenarios. The difference between 2011 (base year) and 2030 is equal to 1054 million tonnes in the non-technical progress version of the model and 1102 million tonnes in the version where technological progress is modelled. Therefore the difference is very small, because most of the sectors emit as much as they are allowed regardless of the change in the technologies. In that context, the reduction in carbon emission is forced by regulations and the technology itself makes it easier to comply. The real difference lies in the GHG’s emission in non-EU regions, where technology helps to reduce carbon emission faster and more efficiently.

Increase of the GHG’s emission reduction target in the EU to 45% (modelled as more stringent constraint to emission, as described above) impacts more heavily in 2030 than the introduction of the MSR, which altered emission pathway only slightly. In scenarios without technical progress, introduction of MSR decreases emission only in the year 2030 by 2% in comparison to the baseline scenario, and additionally, further tightening of the reduction target to 45% decrease emission by another 8% in the year 2030 only (10% in total compared to baseline scenario). As long as energy-saving technical progress is included already in the baseline, the impact of policies (both MSR and tightening reduction targets) is very similar, as sectors comply to the regulations regardless of the technologies they have in place – consequently, more stringent regulations are a vital part of the emission reduction policies.

In the case of lack of binding emission reduction targets for the other than the EU regions of the world, some carbon leakage would still be observed because their total emission would not be capped in any way. Alternatively, if there will be a nationwide emission cap, carbon leakage

Figure 3

Figure 4

would not exist, because their EU emission would equal their limits regardless of the actions. Therefore, the question of how binding are NDCs of the other regions of the world or in what scale they will be implemented is of crucial importance when carbon leakage is considered. Unfortunately, CGE model is not the best tool to extensively analyse how binding are emission constraints for the governments outside the EU. Consequently, in this paper, we assumed that regions other than the EU have no emission targets to show what theoretically would be carbon leakage in such a situation. We will present additional scenarios with binding reduction targets according to NDC’s in the forthcoming paper. Nevertheless, this is an important caveat and must be kept in mind, while analysing the results, the purpose of which is to show the effects of the carbon leakage and factors affecting the scale of this phenomenon. Also, that (unrealistic) assumption is the reason, why these results should not be interpreted as

Figure 5

Figure 6

a concrete answer on the question on whether the carbon leakage will occur or not (as the answer is ‘it depends’), but rather as a demonstration on how important assumptions are for the results of carbon leakage.

The magnitude of increase in GHG’s emission in the rest of the world is rather small and do not exceed 1%. However, due to the higher base, the amount of ‘leakage’ is quite substantial in absolute terms. For instance, introduction of MSR in version without taking into account energy-saving technical progress increases GHG’s emission by 0.2% and 45% reduction target by another 0.1% in 2025. In the ‘technical progress’ version of the model, these amounts are very similar, but the base is lower. However, even such a small value of GHG’s emission reduction outside the EU is enough to outweigh the emission reduction benefits of MSR/higher target for the EU in 2030 in some scenarios.

The leakage rates (measured as a ratio of increased emission in rest of the world to avoid emission in Europe) are quite high and range from 30% in the most restrictive scenario in 2015 up to 361% in 2030 in the least restrictive scenario. In general, leakage rates are higher in versions without technical progress in 2015–2025, but this changes in 2030, where leakage rates are higher in the scenarios with technological progress. When energy-saving

Figure 7

Figure 8

technical progress is taken into account, both EU and non-EU countries are more energy efficient and therefore their production is less emission intensive. The scale of the leakage is still significant mostly due to the assumption on no reduction targets in the rest of the world.

In this section, we decompose the carbon leakage into the carbon intensity, competitiveness and demand channels.

Figure 9

Figure 10

Our results indicate that carbon intensity channel is the most important channel of carbon leakage, far more important than competitiveness or demand channels. In both versions of the model, the impact of carbon intensity channel on total change in emission is positive. This phenomenon is understandable, given the use of fuels in the EU countries is significantly reduced in scenarios in comparison to the baseline and the production of energy intensive goods is outsourced to rest of the world countries. Also, as there is no convergence in energy-efficiency, these goods are produced using less effective and older technologies. Consequently, these outputs are even more energy and emission intensive if produced abroad.

Figure 11

Figure 12

Domestic demand channel is slightly more important in scenario with 45% reduction if exogenous energy-saving technical progress is taken into account. In this case, changes to reduce in carbon production emission structure in the alone EU countries are not sufficient to meet the targets. Therefore, the domestic demand must be reduced. In general, the first response of the economy to intensity changes of GDP, in emission either through targets substitution is to reduce between emission the fossil fuels or through changes in the production patterns (e.g., changing technology to less emission intensive).

Outsourcing production elsewhere or reduction of the domestic demand is the next step, when reduction in emission intensity of the economy is not sufficient to meet the target.