No.51 September 2015 FREE ALLOCATION IN THE EUROPEAN ...€¦ · relating to the first phases of the EU ETS have not shown any evidence of carbon leakage (Reinaud, 2008; Sartor et

No.51 September 2015

FREE ALLOCATION IN THE EUROPEAN EMISSIONS TRADING SYSTEM

(EU ETS): IDENTIFYING EFFICIENT MECHANISMS THROUGH TO 2030

Matthieu Jalard1 and Emilie Alberola2

In a world with asymmetrical climate policies, the conclusions of the European Council of 23

October 2014 agreed on continuing the allocation of free CO2 emissions allowances beyond 2020

to industrial sectors in the EU ETS. This statement has been confirmed in the European

Commission’s proposal to revise EU ETS directive for phase IV disclosed in July 2015. The stated

objective is to ensure that the most efficient industrial installations do not face undue carbon costs which

would lead to carbon leakage. Furthermore, free allocations should not undermine the incentive to cut

CO2 emissions, lead to distortions or windfall profits and reduce the auctioning share of allowances.

From 2013 to 2020, the allocation of free allowances has been defined according to harmonized

European rules based on benchmarks (carbon intensity targets) and historical output adjusted to the free

allocation cap by applying the Cross-Sectoral Correction Factor (CSCF). What would be the impact of

pursuing the current mechanism through to 2030? Does the EU Commissions’ proposal of 15th

July respond to the Council’s requirements? Which alternative mechanisms could do so?

This study examines four scenarios and their potential consequences.

- Scenario 1 continues the current free allocation mechanism until 2030. The volume of free

allocations thus calculated would be higher than the available free allocation cap and would need to

be reduced by a Cross-Sectoral Correction Factor (CSCF) of 66% in 2030. Carbon costs will thus

increase for all installations, regardless their exposure to carbon leakages, reducing the protection of most

exposed sectors, while widely allocating sectors with limited exposure.

- Scenario 2 analyses the proposal to implement an allocation mechanism based on recent

industrial output combined with appropriate updating of benchmarks. This allocation method is

more effective in combating carbon leakage, enables to avoid over-allocations and perverse threshold

effects, but it further mutes the carbon price signal to consumers, and should be complemented by

additional mechanisms for more efficient use of materials. This mechanism would lead to a less

impacting CSCF of 71% in 2030 with a 1.4% annual growth assumption that would however depend on

the aggregate output level. Based on our estimates, this factor would be comprised between 62% and

82% in 2030, entailing an uncertainty about the net carbon cost borne by installations amounting to

10% of added value for the cement sector and 6% for the steel sector, with a 30€/tCO2 price assumption.

- In Scenario 3, a more targeted and focused allocation is presented, which better reflects the

exposure to carbon leakage risks. It is proposed to use differentiated allocation rates either based on

carbon cost and trade intensity thresholds like implemented in California for example, or based on

targeted maximum carbon costs for each sectors depending on trade intensity. This would enable to

reduce the allocation volume and overcome the ex post correction and the uncertainty coming with it, and

to mitigate carbon costs more efficiently for exposed sectors.

- The Scenario 4 assesses the European Commission’s proposal, which could be leading to a 30%

reduction of allocation volume to all installations by 2030. Focusing allocation to exposed sectors, and

enhancing flexibility in the supply of free allowances through a dynamic New Entrant Reserve could be

levers help combat carbon leakages more efficiently and maintain incentives to reduce emissions.

1 Matthieu Jalard is Research fellow in the 'CO2 and Energy Markets' unit: [email protected]

2 Emilie Alberola is Research Unit Manager of the 'CO2 and Energy Markets' unit – [email protected]

mailto:[email protected]:[email protected]

Climate Report No. 50 – Free Allocation in the European Emissions Trading System (EU ETS):

identifying efficient mechanisms through to 2030

2

ACKNOWLEDGEMENTS

The authors would like to thank all those who helped with preparing this report. Without

claiming any responsibility for the content, the authors wish to acknowledge in particular the

exchange of views with

Frederic Branger (CIRED), Sarah Deblock and Stefano Di Clara (IETA), Yue Dong and

Maxime Durande (French Environment Ministry DGEC), Pierre Guigon (World Bank,

Partnership for Market Readiness) Jean Giraud (French Ministry of Finance, Direction

Générale du Trésor – DG Budget), Fréderic Lehmann (French Ministry of Economy and

Industry, Direction Générale des Entreprises – DG Companies), Jean Pierre Ponssard

(Ecole Polytechnique), Stefan Schleicher (University of Graz, Austria).

The authors would like also to thank the CEPS and Business Europe for giving opportunity

to present preliminary results and for the helpful comments provided.

The authors would also like to thank the members of the CDC Climat Research Team.

The authors take sole responsibility for findings or ideas presented in this report as well as

any errors or omissions.

Publication director: Benoît Leguet - ISSN 2101-4663

To receive regular updates on our publications, send your contact information to [email protected]

Press contact: Maria Scolan - + 33 1 58 50 32 48 - [email protected]

This publication is fully-funded by Caisse des Dépôts, a public institution. CDC Climat does not contribute to the financing of

this research.

Caisse des Dépôts is not liable under any circumstances for the content of this publication.

This publication is not a financial analysis as defined by current regulations.

The dissemination of this document does not amount to (i) the provision of investment or financial advice of any kind, (ii) or of

an investment or financial service, (iii) or to an investment or financial proposal of any kind.

There are specific risks linked to the markets and assets treated in this document. Persons to whom this document is directed

are advised to request appropriate advice (including financial, legal, and/or tax advice) before making any decision to invest in

said markets.

The research presented in this publication was carried out by CDC Climat Research on an independent basis. Organisational

measures implemented at CDC Climat have strengthened the operational and financial independence of the research

department. The opinions expressed in this publication are therefore those of the employees of CDC Climat Research alone,

and are independent of CDC Climat’s other departments, and its subsidiaries.

The findings of this research are in no way binding upon, nor do they reflect, the decisions taken by CDC Climat’s operational

teams, or by its subsidiaries. CDC Climat is not a provider of investment or financial services.

mailto:[email protected]:[email protected]

Climate Report No. 50 - The European Emissions Trading System (EU ETS) and

free allocation through to 2030: identifying efficient mechanisms

3

ACRONYMS

EU ETS: European Union Emission Trading Scheme

CLEF: Carbon Leakage Exposure Factor

CSCF: Cross Sectoral Correction Factor

NAP: National Allocation Plan

NIM: National Implementation Measure

CCS: Carbon Capture and Storage

GHG: Greenhouse gases

MSR: Market stability Reserve

NER 300: New Entrant Reserve, 300 million allowances earmarked for innovation

TI: Trade Intensity

CC: Carbon cost

OBA: Output-Based Allocation

HA: Historical Allocation



4

TABLE OF CONTENTS

INTRODUCTION: THE CHALLENGE OF LIMITING THE RISK OF CARBON LEAKAGE WHILE APPLYING CARBON PRICING 5

I. FREE ALLOCATION IN PHASES 2 AND 3 (2008-2020) OF THE EU ETS: WHICH LESSONS CAN BE DRAWN? 6

1. The European regulatory framework to combat carbon leakage: free allocation of allowances based on historical production levels 6

2. The efficiency of free allocation mechanisms prevailing in phases II and III put into question 9

3. Insights from academic literature concerning output based allocation 13

II. SUSTAINABLE ALLOCATION OF FREE ALLOWANCES THROUGH TO 2030: EVALUATION OF THREE SCENARIOS 15

1. Scenario 1: Continuing the current free allocation method until 2030 16

2. Scenario 2: Implementation of output based allocation 18

3. Scenario 3: Alternative designs for Output based allocation 21

IV. CONCLUSION 33

1. The importance of strengthening the EU ETS carbon price signal 33

2. Effective mitigation of the risks of carbon leakage for exposed sectors to strengthen the credibility of the price signal and political commitment 33

3. A coherent strategy for low carbon innovation in addition to a price signal 35

V. ANNEXES 36

Annex 1: The impact of first applying the CLEF on the amount of free allocation 36

Annex 2: Calculation method for Output-Based Allocation 37

Annex 3: Uncertainty and free allocation volumes 41

Annex 4: Allocation rates in the framework of targeted maximum carbon costs 43

REFERENCES 44

THE ‘CLIMATE REPORTS’ SERIES FROM CDC CLIMAT RESEARCH 47



5

INTRODUCTION: THE CHALLENGE OF LIMITING THE RISK OF CARBON LEAKAGE WHILE

APPLYING CARBON PRICING

Despite the growing urgency of climate change, international climate negotiations have postponed the

prospect of a climate agreement which would implement a globally harmonized framework to limit global

greenhouse gases emissions. As a result climate policies will remain largely sub-global in the years to

come, giving rise to unilateral initiatives which internalize the costs of GHG emissions, such as the EU

ETS which covers the equivalent of 2GtCO2e of emissions from the European industrial and energy

sectors.

However, global cost-effectiveness of unilateral action is reduced by the lack of flexibility in the

geographical distribution of GHG emissions reductions and may be further undermined by the

phenomenon of carbon leakage. The carbon cost differential between two regions is indeed likely to lead

to a delocalisation of production towards jurisdictions which are bound by weaker environmental

constraints. Such carbon leakages would reduce the environmental benefits of the policy and would have

a negative impact upon the economy in question.

The economic literature has taken a close look at this phenomenon. So-called 'ex-ante' partial or general

equilibrium models generally present carbon leakage rates ranging from 5% to 20% (Branger et al. 2014),

but the diversity of underlying assumptions on the elasticity of demand for energy or substitution between

local and foreign goods makes it difficult to compare and interpret results. To date, empirical studies

relating to the first phases of the EU ETS have not shown any evidence of carbon leakage (Reinaud,

2008; Sartor et al. 2012, Branger et al., 2013). Indeed energy and carbon costs do not appear to influence

international trade as much as other factors, such as proximity of demand, or the institutional framework

(Sato, 2015). However, to date, observed CO2 prices have been low and protection mechanisms have

been very generous.

Several studies show that climate policies can induce, in some cases, two symmetrical phenomena

related to carbon leakage and competitiveness losses that are likely to offset them, at least partially.

These are additional GHG emission reductions induced by the diffusion of low-carbon technologies and

policies (so called spill-over effect, Dechezleprêtre, 2008, 2012), and the positive competitive impact

provided by the first mover advantage (Pollit, 2015). On a broader basis, the Porter Hypothesis (1995)

argues that beyond the short-term costs, climate policies are, from a dynamic point of view, likely to

stimulate additional innovation efforts increasing productivity, which would not be made otherwise due to

unavailability of information or risk aversion. Concerning Europe, this hypothesis is supported by

Constatini et al. (2011) who made use of a gravity model to show that the EU-15 environmental policies

tended to support innovation and exports rather than undermine industrial competitiveness over the period

1996-2007. These results argue for a European industrial renaissance oriented towards resource efficient

and green goods that will be highly valued by future markets.

As such, it seems than carbon leakages and competitiveness losses are, more than a technical issue, a

political obstacle to implementing ambitious and economically efficient climate policies. In terms of the

example of EU ETS, they have led to very generous free allocation and a significant inflow of international

credit. They have also accentuated the fall in the price of allowances, increasing the cost of long-term

decarbonisation. The generous allocation of allowances also entailed windfall profits, discrediting the

climate policy. It further reduced the revenue generated by the scheme, limiting the possibility of reducing

pre-existing distortive taxes and further increasing the cost of the environmental policy (Goulder, 2013).

Thus, specific and targeted measures aiming to protect the most exposed sectors to the risk of carbon

leakages are required to encourage the acceptability and credibility of climate policies and eventually to

strengthen their ambition and reduce their long-term costs. The conclusions of the European Council in

October 2014 set a more restrictive cap through to 2030 with an annual reduction of 2.2% from 2020. It

will contribute to the long-term visibility and strength of the price signal. This new cap, combined with a

Market Stability Reserve (MSR) correcting market imperfections relating to the inflexibility of supply and

the short-sightedness of stakeholders who do not take into account long term scarcity, is likely to lead to a



6

carbon price trajectory which is more in line with the European Union's long-term decarbonisation target

for 2050, and prevent a deviation in relation to an optimal abatement trajectory (Climate Strategies, 2015).

This strengthening of the EU ETS has led the European Council to commit to pursue free allocation post

2020 so that the most highly performing installations do not face any unwarranted carbon if it can be a

source of carbon leakage. This mitigation of carbon costs must however not weather carbon efficiency

incentive and associated investments triggering in the innovative technologies required for deep, long-

term decarbonisation of the industrial sectors. Moreover, according to the conclusions of the European

Council in October, free allocation must not lead to sectoral distortions or windfall profits resulting from the

over-allocation. Finally, the allocation of free allowances must be sustainable and predictable for industry,

within a context of a dwindling free allocation budget in order to preserve the share of auctioned

allowances. Which free allocation mechanisms could be implemented to respond to these specifications?

In a first part, main lessons drawn from Phases II and III of the EU ETS are outlined. Although current

allocation mechanism has effectively mitigated the carbon costs for all industrial sectors, thus protecting

those most at risk, the rigidity of the current regulation is entailing significant distortions between sectors

and giving rise to perverse incentives that fail to reward properly carbon efficiency improvements. In the

current context, where it is proving difficult to implement border carbon adjustment mechanisms, an

allocation method which is more reactive to output fluctuations and technological changes appears to be

appropriate.

In a second part durability and predictability of different free allocation mechanisms through to 2030 are

examined, with the definition of three scenarios.

Scenario 1 extends the current free allocation mechanism until 2030;

Scenario 2 analyses the (?) implementation of the proposed output based allocation and appropriate

updating of the benchmarks;

Scenario 3 explores several design options aiming to improve the efficiency and durability of the

mechanism presented in the second scenario, through implementing an allocation reserve, eliminating

the free allocation cap (a priori incompatible with point 2.9 of conclusions of the European council of

October, 2014 which recommends that the share of auction allowances be not reduced) and granting

targeted and gradual free allowances depending on exposure to the risk of carbon leakage.

Scenario 4, assesses the European Commission’s proposal for the revised EU ETS directive disclosed

on 15th July 2015in light of the Council’s specifications.

I. FREE ALLOCATION IN PHASES 2 AND 3 (2008-2020) OF THE EU ETS: WHICH LESSONS

CAN BE DRAWN?

1. The European regulatory framework to combat carbon leakage: free allocation of

allowances based on historical production levels

Since 2005, more than 11,000 European installations spread across 31 countries have been subject to

the EU ETS. They are obliged to measure their CO2 emissions every year and must cover them with

emission allowances.

During Phases I and II of the EU ETS, emission allowances were made freely available to installations

covered by the mechanism. The level of allowances for installations was determined nationally, in line with

National Allocation Plans (NAPs), based on observed CO2 emission levels.

The revised EU ETS directive of 2009 (2009/EC/29) describes the new regulations for allocating

allowances in Phase III, beginning in 2013: auctioning becomes the main method for acquiring emission

allowances. However, exceptions to this rule were granted through the allocation of free allowances in

order to prevent the risk of carbon leakage.



7

Firstly, it was decided to allocate a quantity of free allowances to industrial installations corresponding to

the sectoral benchmarks (average CO2 emissions per ton of output of the 10% most efficient European

installations), multiplied by the historical activity level. This quantity is eventually reduced by the Carbon

Leakage Exposure Factor (CLEF), determined by the European Commission, equal to 80% in 2013, then

reducing linearly to 30% in 2020.

Table 1 – Value of the Carbon Leakage Exposure Factor from 2013 to 2020 in the EU ETS

Source: EU ETS Directive (2009/29/EC)

Secondly, a list of sectors deemed to be highly exposed to the risk of carbon leakage for the years 2013

and 2014 was drafted in 2009. For installations in these sectors, the CLEF value is equal to 100%.This list

was drafted on the basis of precise quantitative indicators, but may be completed after a qualitative

evaluation. This list must be updated every five years. On 27th October 2014, the Commission adopted a

new list for the period 2014-2019 which includes 146 industrial sectors defined with revised NACE codes,

of a total of 236 and covering around 95% of 2013 industrial emissions.

Pursuant to Article 10a of the EU ETS Directive revised in 2009, a sector is considered to be exposed to

the risk of carbon leakage if it meets at least one of the following three quantitative criteria:

𝐶𝑎𝑟𝑏𝑜𝑛 𝑐𝑜𝑠𝑡 =𝐷𝑖𝑟𝑒𝑐𝑡 + 𝑖𝑛𝑑𝑖𝑟𝑒𝑐𝑡 𝑐𝑎𝑟𝑏𝑜𝑛 𝑐𝑜𝑠𝑡

𝐺𝑟𝑜𝑠𝑠 𝑣𝑎𝑙𝑢𝑒 𝐴𝑑𝑑𝑒𝑑 > 30%

𝑇𝑟𝑎𝑑𝑒 𝐼𝑛𝑡𝑒𝑛𝑠𝑖𝑡𝑦 = 𝐼𝑚𝑝𝑜𝑟𝑡𝑠 + 𝐸𝑥𝑝𝑜𝑟𝑡𝑠

𝑃𝑟𝑜𝑑𝑢𝑐𝑡𝑖𝑜𝑛 + 𝐼𝑚𝑝𝑜𝑟𝑡𝑠 − 𝐸𝑥𝑝𝑜𝑟𝑡𝑠 > 30%

𝑇𝑟𝑎𝑑𝑒 𝐼𝑛𝑡𝑒𝑛𝑠𝑖𝑡𝑦 > 10% & 𝐶𝑎𝑟𝑏𝑜𝑛 𝑐𝑜𝑠𝑡 > 5%

Figure 1 - Total allowance allocations (MtCO2) in 2013 for sectors deemed to exposed to a risk of carbon

leakage for the period 2015 to 2019 and carbon costs (assuming a 30€/tCO2e price1)

Source: CDC Climat Research based on European Commission 2014, EUTL

1 The carbon cost is retrieved from the Annex to the European Commission Decision on determining, pursuant to Directive

2003/87/EC of the European Parliament and of the Council, a list of sectors and subsectors which are deemed to be exposed

to a significant risk of carbon leakage for the period 2015 to 2019. It is equal to the sum of the direct costs (emissions

multiplied by the Auctioning Factor and the CO2 price) and the indirect costs (electricity consumption times average

emissions factor) divided by the sectorial value added.

Year 2013 2014 2015 2016 2017 2018 2019 2020

Carbon Leakage Exposure Factor 80% 73% 66% 59% 51% 44% 37% 30%



8

Table 2 – Sectors deemed to exposed to a risk of carbon leakage for the period 2015 to 2019

Source: CDC Climat Research calculations based on European Commission 2014, EUTL

Against this backdrop, National Implementation Measures (NIMs) were defined by Member States,

presenting the quantities of allowances to be freely allocated following defined rules until 2020. These

NIMs were approved by the European Commission Decision (2013/448/EU) in September 2013.

However, the number of free allowances to be granted was higher than the allocation cap. A Cross-

Sectoral Correction Factor (CSCF) was applied, reducing the quantity of allowances granted to each

installation.

Table 3 – Value of the Cross-Sectoral Correction Factor from 2013 to 2020 in the EU ETS

Source: European Commission Decision (2013/448/EU)

The final allocation received by an installation ends up calculated as shown below1:

Source: CDC Climat Research, based on the European Commission Decision (2013/448/EU)

1 The Output correction factor applied to final allocation enables the free allocation level to be readjusted: as soon as the

annual production level of an installation falls below 50%, 25% or 10% of the reference output level, the allocation received

the following year is reduced respectively by 50%, 75% and 100%. The free allowance allocation method established during

Phase III is therefore hybrid: it retains a dimension of so-called historical allocation, but enables readjustments in cases output

variations are too large compared to the reference output levels.

Criteria Number of sectors

Allocation

(MtCO2)

Carbon cost > 30%

4

210

Carbon cost > 5% and trade intensity > 10%

20

496

Trade intensity > 30%

133

148

List total

146

712

Industry total

236

755

Year 2013 2014 2015 2016 2017 2018 2019 2020

CSCF 94.27% 92.63% 90.98% 89.30% 87.61% 85.90% 84.17% 82.44%



9

2. The efficiency of free allocation mechanisms prevailing in phases II and III put into

question

Mechanisms established to date have largely mitigated the carbon costs caused by

the carbon price in the EU ETS.

Installations subject to the EU ETS face a direct carbon cost equal to CO2 emissions, multiplied by the

average carbon price. However, the allocation of free allowances mitigates this cost. Net carbon cost is

thus defined as the difference between the allocation of allowances and emissions, multiplied by the

observed carbon price.

In Figure 2, gross carbon cost in relation to added value for all EU ETS sectors - the height of the

rectangles - is given, as is the net mitigated cost - the black line - by allowance allocation - the width of the

rectangles - under the assumption of of a €5/tCO2e carbon price. As illustrated, net carbon cost has been,

for most sectors, lower than 1% of sectoral added value in 2013. For some sectors, carbon cost has been

negative: this means that free allocation was higher than observed emissions. Moreover, this calculation

takes into account neither the potential repercussion of carbon costs to the end consumer in certain

sectors, nor the use of international offsets reducing the compliance cost. The perceived cost could

therefore have been further mitigated.

Figure 2 - Mitigated carbon costs of EU ETS sectors in 2013 (5€/tCO2e)1


Therefore, free allocation mechanism has been effective in mitigating carbon costs. Moreover, as

illustrated in Box 1, given a carbon cost borne by an installation, several factors naturally mitigate the risks

of carbon leakage, and empirical studies highlight that carbon cost actually has very little influence on

international trade flows (Sato et al. 2015). As such ex post econometric studies have not revealed

statistically significant evidence of carbon leakage (Reinaud, 2008; Sartor et al, 2012 ; Branger et al,

2013,

1 The net carbon cost is defined as the compliance cost minus the value of freely allocated allowances :

𝐷𝑖𝑟𝑒𝑐𝑡 𝑛𝑒𝑡 𝐶𝑜𝑠𝑡 = ( 𝐸𝑚𝑖𝑠𝑠𝑖𝑜𝑛𝑠 − 𝐴𝑙𝑙𝑜𝑐𝑎𝑡𝑖𝑜𝑛) ×𝑃𝐶𝑂2

𝑉𝐴= (1 −

𝐴𝑙𝑙𝑜𝑐𝑎𝑡𝑖𝑜𝑛

𝐸𝑚𝑖𝑠𝑠𝑖𝑜𝑛 ) × 𝐷𝑖𝑟𝑒𝑐𝑡 𝐶𝑜𝑠𝑡



10

It is important to note that the issue of allocation mainly concerns around ten energy intensive sectors,

with potentially high carbon costs. The ten sectors with the highest carbon costs represent 550 million free

allowances granted, of the 712 total granted to the sectors on the list of sectors deemed to be exposed to

a significant risk of carbon leakage in 2013.

Box No. 1: Carbon cost and carbon leakage

To what extent the carbon cost resulting from climate policy likely cause carbon leakage? Carbon leakage

is defined as the transfer of a production activity covered by the EU ETS in Europe to another geographic

zone which is not subject to the same regulations in terms of GHG emissions. Compliance with the EU

ETS is, indeed, likely to lead to an additional cost which confers a comparative advantage upon

competing installations. Production outside the EU ETS is thus likely to gain market shares. The carbon

leakage rate is defined as the increase in CO2 emissions outside EU ETS caused by the climate policy

divided by the emissions reduction in Europe caused by the EU ETS. This means that the carbon leakage

rate may be higher than 100% when the substitute production is more efficient in terms of emissions.

A carbon cost is not necessarily synonymous with carbon leakage: various mechanisms are naturally

mitigating the cost differential. Initially, free allocation mechanisms, as implemented within the EU ETS,

directly contribute towards reducing the carbon cost per unit produced. Moreover, there is a potential to

reduce CO2 emissions to a cost per unit which is lower than the price given by the market, which may be

significant, reducing the cost of compliance. Besides, depending on the characteristics of the considered

market, particularly the price elasticity of demand and production, as well as the degree of competition,

the producer can pass on some of the carbon cost to the end consumer. Finally, net carbon cost must be

compared to competitors' carbon costs which, even if they are not subject to an explicit carbon price

signal, often bear an implicit carbon cost resulting from more diffuse climate policies. This is the implicit

carbon cost resulting, for example, from additional costs following renewable energy objectives or

restrictions on the construction of coal-fired power plants. This implicit cost may be on the same order of

magnitude, or even higher, than the explicit cost resulting from the EU ETS. It should be emphasized that

the observed cost differential of carbon ends up having little importance in comparison with other factors

such as energy prices, regulations, institutional framework, infrastructures, and proximity of demand,

which are the main determinants of international trade flows.

Free circulation of goods is not the only source of carbon leakage. The asymmetry of climate policies is

also likely to affect capital flows, encouraging investment in countries where the carbon cost is predicted

to be lower. This phenomenon is known as 'investment leakage'. Finally, global energy prices are a

second lever for carbon leakage. When installations in a geographic zone reduce their energy

consumption to comply with a climate policy, this will have a downward pressure on global energy

markets, which could increase consumption and emissions in third zones. The GHG emissions reduction

through CCS technologies is one possible solution to counter this phenomenon (Quirion, 2011).



11

Figure 3 - Factors mitigating carbon costs

Source: Ecorys, 2013

Allocation of free allowances by benchmark according to harmonized rules has

reduced excess allocations as well as distortions between sectors and countries.

Between 2005 and 2012, every Member State had an allocation budget for their eligible installations

depending on historic observed emission levels. This allocation method led to significant allocation

surpluses: during Phase II, industry was allocated a quantity of allowances corresponding to, on average,

130% of its actual CO2 emissions. In addition, the allocation level was very unequal across sectors. In

2009, the allocation rate, defined as the allocation divided by emissions, was nearly 200% for the steel

sector, compared to 100% for the refining sector. This allocation method led to windfall profits for some

installations, as well as to distortions between sectors.

From 2013, the implementation of harmonized European-wide rules, allocating free allowances according

to benchmarks and historic output levels, considerably reduced allocation surpluses and, to a lesser

extent, distortions between sectors. As illustrated in Figures 4 and 5, the allocation rate was, on average,

only 100% for industrial sectors in 2013 and differences between sectors tend to reduce.

Figure 4 - Allocation of allowances divided

by output based CO2 emissions:

reduction in surpluses in Phase III

Figure 5 - Allocation of allowances divided by output

based CO2 emissions:

distortions between sectors




12

However, due to the rigidity of the rules, some sectors still enjoyed significant surpluses in 2013: the steel

sector was allocated up to 140% of its emissions and 120% in the case of the cement sector. Indeed,

allocation is proportional to the reference historical output levels, and for some sectors, industrial output

has fallen compared to pre-crisis levels. Free allocation has not significantly reduced, insofar as most

installations continue to produce above the 50% historical output threshold. To a lesser extent, allocation

differences between sectors result from the different distributions of installations' carbon efficiencies in

relation to benchmarks (See Annex 2).

The current free allocation mechanism has reduced the incentive to carbon

efficiency

Beyond unjustified distributional effects, allocation surpluses are likely to damage the efficiency of the EU

ETS. Using industrial data, Zachmann et al. (2011) showed that over-allocations are prone to reduce

installations' efforts to reduce emissions. These empirical results are in contrast with the economic theory

which states that installations equate the observed CO2 price with their marginal abatement costs,

regardless of the volume of free allowances. He concludes that too high allocation levels tend to mask the

price signal observed by market participants. The economic efficiency of the EU ETS, which is based on

spatial and temporal flexibility enabling to exploit lower cost emission reductions potential and an optimal

abatement trajectory, is thus reduced. This also means that the opportunity cost of free allowances is not

fully passed through to consumers as theory would predict. On the one hand, this means that free

allocation is likely to help industries retain their market shares, but on the other hand, it is muting carbon

price for intermediate and final consumers,

Last but not least, the current mechanism, which is correcting allocation according to output thresholds, is

giving rise to strategic behaviours, ultimately encouraging certain installations to emit more CO2 per unit

produced. When the annual production level of an installation falls below 50%, 25% or 10% of the

historical output level, the allocation received the following year is reduced respectively by 50%, 75% and

100%. The rational for using these thresholds is to reduce potential allocation surpluses identified during

the preceding phases in the event of a large output reduction. However, it has been shown that some

installations, particularly in the cement sector where demand remains low, increased their output levels in

2012 to reach these thresholds and to benefit from a higher volume of free allocation. This is financially

attractive as long as the difference in allowances received remains higher than additional allowances to

be returned corresponding to increases in output.

Using a counterfactual scenario, Branger et al. (2014) show that strategic behaviors of cement plants in

order to reach the 50% historical output threshold entailed an increase in European clinker production of

6.4Mt in 2012, i.e. an emissions increase of 5.8 MtCO2e. The total allocation for the cement sector was

138 MtCO2e in 2012, slightly lower than the counterfactual level which would have been reached without

the threshold, estimated at 144.5 MtCO2. An allocation mechanism strictly proportional to output levels

would have reached an aggregate level of only 98.2 MtCO2e. The effectiveness of the thresholds to limit

over-allocations was, therefore, very limited. These have, moreover, led to significant operational

distortions. According to the study, the increase in clinker production was compensated for by either the

increase in exports mainly to African countries, or an increase in the quantity of clinker used per ton of

cement. In the second case, historical allocation corrected with thresholds thus provided a perverse

incentive to eventually increase CO2 emissions per ton of product.

Regarding these lessons, it seems necessary to increase the flexibility of the current free allocation

mechanism and to make it more responsive to installations' output fluctuations. The economic literature

provides a detailed analysis of the different allocation methods and, in the absence of the border carbon

adjustment, suggests that output based allocation (OBA) would be more efficient, rather than historical

allocation (HA).



13

3. Insights from academic literature concerning output based allocation

In the absence of a harmonized price signal on the international scale, the economic literature suggests

(Demailly and Quirion, 2006; Monjon, 2009; Fisher, 2009) that auctioning allowances for all sectors,

combined with a border carbon adjustment, is the most cost-effective way of implementing a unilateral

climate policy. This would, indeed, equalize the carbon costs while efficiently enabling the pass through of

carbon cost throughout the whole value chain. The incentive to reduce CO2 emissions remains, both

through more efficient production and through substitution of domestic consumption with products with

lower CO2 emissions. However, such a mechanism raises concerns in terms of administrative cost,

compatibility with international trade regulations (Branger, 2013) and equitable sharing of abatement costs

(Böhringer, 2012). A border carbon adjustment mechanism could be seen as veiled green protectionism

and could trigger a trade war, instead of incentivizing the implementation of similar climate policies. In the

case of Europe, the acquisition of allowances for importers according to Best Available Technologies

carbon intensities, as well as recycling revenues raised for funding mitigation and adaptation in

developing countries (Godard, 2009; Neuhoff, 2007; Branger, 2013) is the most plausible solution to

comply with GATT regulations (so called 'most favoured nation' and 'national treatment') while equitably

distributing the revenue raised. However, this would not allow discriminating against the less carbon

efficient producers worldwide, increasing the cost of the policy compared to an efficient outcome.

In light of the difficulties around implementing a border carbon adjustment, Demailly (2008) Quirion (2009)

and Fisher (2004) suggest an output based allocation, which is more efficient to combat carbon leakage

than historical allocation, such as that implemented within the context of the EU ETS. Historical allocation

has a tendency to preserve industrial competitiveness, seen as the ability to generate profits. Output

based allocation, by encouraging production, can better preserve competitiveness, defined as the ability

to retain market shares, and will thus be more effective to combat carbon leakages. However, the cost of

the climate policy is likely to increase, because the marginal carbon cost borne by installations will

therefore vary depending on sectors. This can give rise to inefficiencies in allocating abatement efforts

and lower transmission of the carbon price signal, leading to excessive consumption of polluted goods. In

comparison with an optimal decarbonisation trajectory, this would entail the use of additional and more

costly abatement options to achieve the same reduction target.

Box No. 2: Output based and historical allocation

The simplified model described above, inspired by Demailly (2006), illustrates the different incentives

provided by HA (historical allocation) and OBA (output based allocation

A price taker and profit maximizing installation has a horizontal marginal revenue curve (MR) and an

increasing marginal cost curve (MC) as illustrated below. Without a carbon price signal, it produces a

quantity q* equalizing these two values, with a carbon intensity i.

In the context of implementing an emission trading system, with historical free allocation (HA), the

installation will pass through the opportunity cost of allowances in its marginal cost : producing one more

unit of output will not give rise to additional allocations, and conversely reducing the production would

enable to sell the corresponding allowances in the carbon market. Its marginal production cost will thus be

increased by 𝑃𝐶𝑂2 × 𝑖𝐻𝐴, leading to a lower production equilibrium 𝑞𝐻𝐴∗ . This equilibrium is the same as in

the case without free allocation. This is a common result stating that initial allocation does not change

output and abatement decisions, but has only a distributional impact.

In the case of an output based allocation, however, the behaviour of market participant is directly

impacted by the free allocation mechanism. Producing one more output gives rise to more allowances

freely allocated. The marginal production cost is increased by 𝑃𝐶𝑂2 × 𝑖𝑂𝐵𝐴, but the marginal revenue

increases by 𝑃𝐶𝑂2 × 𝐵 following the allocation of additional quotas. The new equilibrium 𝑞𝑂𝐵𝐴∗ is higher than

the previous case.



14

Source: CDC Climat Recherche (2015) d’après Demailly (2006)

Ultimately, in the context of output based allocation, the marginal cost of production is therefore reduced

by the marginal allocation and output levels are therefore higher.

Regarding the EU ETS as a whole, activity levels would tend to be higher with OBA. Meanwhile, the

carbon efficiency of production should necessarily be improved in order to achieve the same emission

reduction target. This would entail the following points:

1. Output based allocation is less likely to reduce industrial activity;

2. Output based allocation is a more effective means to protect against carbon leakage

3. Output based allocation leads to further reductions in the carbon intensity of production;

4. Output based allocation effectively counters over-allocation;

5. Output based allocations reduce the ability to pass through the carbon price signal for allocated

sectors, and thus combats windfall profits, but reinforces the need to implement parallel mechanisms

to ensure the price signal is transferred to consumers and gives incentive to consume less polluting

goods;

6. The equilibrium price will, in theory, be higher, as will the overall economic cost of the economic

policy: the lever for reducing the carbon intensity of production is used to a greater extent and the

reduction in the production of polluting goods is little used. However, this is compensated for by the

fact that the taxes applied to the factors of production lead to pre-existing distortions, reducing the

production of goods in relation to the efficient economic outcome in certain markets (Fisher, 2004).

The high market concentration for energy intensive goods is also likely to lead to market powers and

production levels which are too low, aimed at maximizing profits. However, empirical evidence show

that the carbon cost is not fully passed through anyway in the framework of historical allocation in the

EU ETS



15

Table 5 - Comparison of various allocation methods.

Grandfathering Benchmarking based on historical

output

Output based (dynamic) allocation

Border Trade Adjustement

Leakage protection

- - + ++

Windfall profits and distorsions

- - - + ++

Incentive to carbon efficiency

- - - + ++

Price signal transmission

- - -- ++

Administrative costs

++ + - --

Source: CDC Climat Research (2015) based on Demailly (2008), Quirion (2009), Monjon (2011), Fisher (2004)

In conclusion, given Europe's structural overcapacity and current low demand, the strategic industrial

policy to specialize in resource efficient goods corresponding to the markets of the future, the need to

reduce over-allocation, windfall profits and strategic behaviours, the implementation of an output based

allocation seems to be particularly appropriate within the context of the EU ETS.

To limit the effect of increasing production, this approach should be complemented by (i) appropriate

revision of benchmarks reflecting the technological changes observed for each sector which will enable

the marginal net carbon cost to remain sufficiently incentive and (ii) a mechanism for transmitting the

carbon price cost to consumers in the most energy intensive sectors, such as the inclusion of

consumption including (Climate Strategies, 2014), proposed by implementing an excise tax on

consumption depending on the commodity content (cement, steel) of the final product.

II. SUSTAINABLE ALLOCATION OF FREE ALLOWANCES THROUGH TO 2030: EVALUATION OF

THREE SCENARIOS

While the debate on the preparation of Phase IV (2012-2028) of the EU ETS is underway, the issue of the

sustainability of the free allocation mechanism has been raised. Does the current mechanism ensure that

installations exposed to the risk of carbon leakage do not bear undue carbon costs, as the Council

committed to in October 2014?

As the free allocation cap reduces each year with the emission cap, the Cross-Sectoral Correction Factor

will necessarily sever a growing share of installations' free allocation, regardless of their exposure to

carbon leakages. What impact will this have upon the net carbon cost borne by these installations

between now and 2030? A scenario based approach aims to shed light on this question and discuss what

could be the features of allocation mechanisms addressing policy objectives formulated by the European

Council in 2014. These policies include: limiting carbon cost for the most efficient installations which are

exposed to carbon leakage, maintaining an economic incentive to reducing CO2 emissions, and avoiding

distortions between sectors and between countries.



16

Figure 6 - Free allocation cap until 2030 with the implementation of the backloading in 2014-2016

Source: CDC Climat Research (2015) based on data from EUTL, European Commission

The study examines four scenarios:

Scenario 1 extends the current free allocation mechanism until 2030;

Scenario 2 analyses the implementation of output based allocation and appropriate updating of the

benchmarks;

Scenario 3 analyses the implementation of output based allocation, with additional mechanisms to

improve its efficiency and durability, and contain the uncertainty of the CSCF.

Scenario 4 examines the implementation of the European Commission’s proposal for a revised EU

ETS directive

Each scenario is evaluated against several criteria: effective mitigation of the carbon cost for exposed

industries, the level of uncertainty entailed by ex-post adjustments of the amount of free allowances (i.e.

application of the Cross-Sectoral Correction Factor) and, in relation to the results of the previous part of

the study, effectiveness of protection against carbon leakage, economic incentive to reduce CO2

emissions per unit produced, and restricting windfall effects and distortions.

1. Scenario 1: Continuing the current free allocation method until 2030

The first scenario considers extending, current allocation rules until 2030. The underlying assumptions

being that:

The free allocation cap is reduced by 2.2% per year after 2020, so that the auctioning share remains

constant:

The list of sectors deemed to be exposed to carbon leakages during the 2020-2030 period remains

identical to those identified for 2015-2019;

The preliminary allocation attributed to an installation is equal to the benchmark multiplied by the

unchanged historical output level;

Benchmark values are assumed to be constant;

The Carbon Leakage Exposure Factor decreases linearly and stops in 2027;

There are two ways to compute the CSCF: (i) pursuing the method used until 2020, which consists of

dividing the free allocation cap by the preliminary allocation (the final allocation is subsequently equal to

the product of the CLEF, the CSCF and the preliminary allocation and will therefore be lower than the free

allocation cap as illustrated in annex 1) or (ii) applying first the coefficient of exposure to carbon leakage

(CLEF) to the preliminary allocation, then to calculate the CSCF correction factor.



17

Under the first method, the final allocation volume is lower than the free allocation cap over the period

considered and the cross-sectoral correction factor CSCF reaches the value of 61% in 2030. Logically, it

would be preferable to apply the second method, with which free allocation remains equal to the free

allocation cap. The CSCF correction factor reaches the value of 66% in 2030, implying that the final

allocation will be reduced in 2030 only by 34% instead of 39% calculated previously.

Figure 7 -Value of the CSCF in Scenario 1

Source: CDC Climat Research (2015) based on European Commission EUTL data

With this CSCF, the final free allocation attributed to each installation covered by the EU ETS can be

assessed. The CO2 emissions of each installation in 2030 are subsequently estimated, under the

assumption of 1.4% annual growth in production and annual efficiency gains of 1% (ADEME, 2014). The

net carbon cost mitigated by free allocation is then computed at sectoral level, as illustrated in Figure 8.

Figure 8 - Mitigated carbon costs in Scenario 1 in 2030 (30€/tCO2e)

Source: CDC Climat Research (2015) based on European Commission EUTL

The net cost of carbon increases for all sectors under this scenario, regardless of their actual exposure to

carbon leakage. This is due to the fact that the allocation gradually decreases with the CSCF correction



18

factor. Some exposed sectors such as the fertilizer sector, lime undergo an increase in their net cost

exceeding 15% of value added, while some slightly exposed sectors see their cost of carbon strongly

attenuated in an inefficient way.

Beyond distortions, persistent over-allocations and perverse incentives, the continuation of the current

allocation method would entail high costs for certain exposed sectors, and thus does not meet the goals

formulated by the European Council. The protection against carbon leakage would not be optimized, as

well as the incentive to reduce CO2 emissions per unit produced.

2. Scenario 2: Implementation of output based allocation

Implementing the mechanism in the base case of 1.4% economic growth

Scenario 2 analyses the sustainability of the implementation of output based allocation in the context of

Phase IV. The underlying assumptions are:

The free allocation cap decreases by 2.2% per year after 2020, so that auctioning share remains

constant:

The list of sectors deemed to be exposed to carbon leakages during the 2020-2030 period remains

identical to those identified for 2015-2019;

The preliminary allocation attributed to an installation is equal to the benchmark multiplied by the

actual output level. In practice, output updates could be conducted every two or three years, for

example, or be subject to ex post adjustment as it is common in the power or water industry.

Benchmarks are assumed to gradually decrease along with observed sectoral technological

progresses (1% per year for industrial installations)

The Carbon Leakage Exposure Factor decreases linearly and stops in 2027.

The method for calculating the CSCF is identical to that used in Scenario 1 after 2020: the CLEF is

first applied to the preliminary allocation, then the corresponding level of allocation is compared to the

free allocation cap to obtain the CSCF value.

In the base case of 1.4% annual GDP growth, the output level of industrial installations is assumed to

grow 1.4% per year, as from 2015. The carbon intensity of installations is assumed to decrease by

1% per year.

The electricity mix is assumed to follow the pathway indicated in the Climate and Energy Package

2030 impact assessment: 45% of renewable energy in 2030, and growth in electricity production by

0.6% per year in the base case of 1.4% GDP growth, as illustrated figures 9;

Figure 9 – Projected CO2 emissions

in the electricity sector

Figure 10 - Projected CO2 emissions in industry

sectors


Supply and demand for allowances are set out in the graph 11 and 12, taking into account the

implementation of the MSR (Market Stability Reserve proposed by the European Commission (COM

(2014) 20 final)) from 2019, and the placement of allowances from the backloading into the MSR as

settled in the agreement resulting from the trilogue negotiations on May 5th, 2015.



19

Figure 11 - Supply and demand for allowances:

annual growth of 1.4%

Figure 12 - Supply and demand for allowances:

annual growth of 0.5 %


Actual production data is not readily available for individual installations and sectors. As a result, it is

preferable to calculate the allocation for each installation using their emissions data and carbon

intensities.

𝐴𝑙𝑙𝑜𝑐𝑎𝑡𝑖𝑜𝑛 = 𝐵𝑒𝑛𝑐ℎ𝑚𝑎𝑟𝑘 × 𝑃𝑟𝑜𝑑𝑢𝑐𝑡𝑖𝑜𝑛 = 𝐵𝑒𝑛𝑐ℎ𝑚𝑎𝑟𝑘

𝐶𝑎𝑟𝑏𝑜𝑛 𝑖𝑛𝑡𝑒𝑛𝑠𝑖𝑡𝑦× 𝐸𝑚𝑖𝑠𝑠𝑖𝑜𝑛

The benchmark divided by carbon intensity ratio can be retrieved for each installation by dividing the

preliminary free allocation received in 2013 by historical CO2 emissions which serve as the reference for

allocation in Phase III. This data can be obtained through the EUTL database. The European Council

meeting in October 2014 advocated for the updating of benchmarks during Phase IV, so that they

adequately reflect actual sectoral emission trajectories. This leads us to make the hypothesis that the

benchmark divided by carbon intensity ratio remains constant until 20301. This ratio was calculated for the

8 000 industrial facilities- where relevant data was available. The ratios were then aggregated by sectors

defined at NACE 2 level. Subsequently, sectoral allocation levels can be estimated based on emission

projections through to 2030.

The ex-post correction factor (CSCF) is computed by dividing the aggregated preliminary allocation by the

free allocation cap. The calculation method is specified in Annex 2. The corresponding trajectory of the

CSCF values in Scenario 2, assuming 1.4% annual growth is set out in Figures 13 and 14.

Figure 13 - CSCF values, Scenarios 1 and 2

Figure 14 - CSCF values, Scenario 2 without

benchmark revision


1 On can object that less efficient facilities are likely to close while the improvement of the average efficiency of the 10% most

efficient installations will remain weak. The average sectoral carbon intensity would then be likely to decrease faster than the

benchmark decline The quantification of such an effect on the evolution of the ratio requires a detailed analysis at the sectoral

level



20

The CSCF reaches 95% in 2021 and decreases to 71% in 2030. This is higher than the level it reaches in

Scenario 1 for several reasons:

Until 2024, the total output level under the EU ETS is lower than the historic reference level used in the

context of historical allocation. As a result, installations receive a larger free allocation volume under

historical allocation; the CSCF further corrects the imbalance between the allocation cap and the

preliminary allocation.

Revision of benchmarks, (along with the assumed carbon efficiency improvement trajectory of 1% per

year) reduces the preliminary allocation and thus the impact of the CSCF. This is why it is higher in

Scenario 2 (output based allocation with benchmark revision) in comparison with Scenario 1 throughout

the entire period in question. However, in the case of output based allocation without revising the

benchmarks, the CSCF is lower from 2024, when the output level exceeds the level of historical output.

Aggregated production level and uncertainty over the CSCF

In the context of historical allocation addressed in Scenario 1, the CSCF value is independent of the

aggregate output level of installations: only the aggregate historical level is taken into account. This may,

therefore, be calculated well in advance, as the European Commission did for the period up to 2020, and

as done through to 2030 in the Scenario 1 above.

Conversely, in the context of output based allocation, the CSCF can (in theory) only be calculated at the

end of the year when the aggregate output level is known: it acts as an ex-post correction to ensure the

free allowance cap is not exceeded. This ex-post correction is likely to spur uncertainty with regard to the

quantity of free allowances received annually by the installations, reinforcing existing uncertainties over

the carbon price.

In the case of strong uncertainty over the abatement cost curve (i.e. output level), the variability of the

induced CO2 price is likely to spur economic efficiency losses that may be important in the context of an

ETS - more than in the context of a tax as long as the abatement cost curve is steeper than the marginal

damage curve, which is the case of CO2 emissions.

This dynamic can be strengthened in the framework of output based allocation subject to a strict free

allocation cap: if output increases, then (i) the scarcity of short-term allowances increases as well as the

CO2 price, (ii) the amount of allowances distributed free of charge per output unit decreases as soon as

the free allocation cap is reached. The interaction of these two factors increases the uncertainty regarding

the cost of carbon, and is not favorable to carbon investments, which are the most capital intensive and

therefore more risky.

The study does not address the uncertainty regarding the price dynamics and the scarcity of emission

allowances, but focuses on the potential uncertainty upon the quantities of allocated allowances for

industrial sectors.

A large number of random annual growth scenarios are modelled between 2015 and 2030 - according to

a standard normal distribution of 1.4% and with a spread of 0.5% (see Annex 3). This enables to set out a

profile of CSCF values in 2030, according to certain probabilities. The more dispersed the values of this

CSCF factor are, the greater the net carbon cost uncertainty will be for industrial sectors. In the case of an

output based allocation with the described assumptions, the CSCF is estimated between 62% and 82% in

2030 with a 90% probability.

Figure 15 represents the mitigated net carbon cost for industrial sectors in the EU ETS, assuming a 1.4%

annual growth rate from 2015, as well as a cost variation range estimated with a 90% probability.

Compared to Scenario 1, the net carbon cost profile is very different: the output level is indeed updated

year upon year, and the sectors benefiting from over-allocations due to lower output levels than the

historical levels experiment a high rise in their net carbon cost. For the cement sector, the net carbon cost

almost doubles from 9% to 16%. Other exposed sectors, such as fertilisers, receive more free allowances

in the context of this scenario, and their net carbon cost decreases from 15% to 10%.



21

Figure 15 - Mitigated carbon costs in Scenario 2 in 2030 (30€/tCO2e)

Source: CDC Climat Research (2015) based on European Commission EUTL and Eurostat data

The uncertainty range of the net cost (with a probability of 90%) reaches very high values for energy

intensive sectors: 10% of added value for cement, 8% for steel and 5% for refining. This uncertainty would

ne be sustainable in the context of the industrial investments necessary for a transition to a low carbon

economy.

3. Scenario 3: Alternative designs for Output based allocation

Building on the more flexible output based allocation described in the second scenario, the scenario 3

presents additional mechanisms and alternative designs enabling to improve its efficiency and durability

and mitigate the uncertainty caused by the ex post correction of free allocation: (i) implementation of an

allocation reserve, (ii) removing the annual allocation cap and (iii) Making free allocation more targeted

and focused.

Implementing a free allowance allocation reserve to increase flexibility

To manage the uncertainty related to an ex post correction of free allowances, a first solution would be to

give greater temporal flexibility to the free allocation mechanism by implementing a reserve. Therefore,

when annual output falls and induces a fall in allocation needs to a lower level than the cap, which would

correspond to a CSCF higher than 100%, the corresponding difference is placed in the reserve.

Allowances placed in the reserve are made available if, in the following years, a large increase in output

entails allocations needs higher than the free allocation cap.

This would induce two main effects: (i) the temporal flexibility enables to better adjust supply of free

allowances and not to "lose" unallocated quotas in years or low production, and thus reduce uncertainty

concerning the quantities (ii) symmetrically to the MSR, the reserve plays a stabilizing role on prices.

When production increases, the supply of allowances is adjusted accordingly, containing the price

increase. Gradually, as the reserve is drained, a scarcity signal is conveyed to installations, which might

have more time to intensify their abatement efforts and adjust their production accordingly.



22

The way the reserve operates is illustrated below in Figure 17: annual fluctuations in demand for

allowances trigger inflows and outflows of allowances into the reserve. The CSCF thus obtained with a

reserve is, on average, higher than that in the 2021-2030 period in relation to the case without a reserve.

Figure 16 - Annual fluctuations

in the demand for allowances

Figure 17 - Example of CSCF trajectories with

and without a reserve


As in the previous case, the potential impact of the reserve on allowances prices is left out of the scope of

the study, which focuses on the quantity effect. The average annual growth value from 2015 until 2030

follows a standard normal distribution of 1.4%, as well as annual fluctuations in demand as of 2020.

Moreover, several initial supply assumptions are suggested, reflecting the possibility of using allowances

not used in Phase III, from backloading,1 or the new entrants reserve (NER).

Figure 18 - Variations in the CSCF value with an allowance reserve


The reserve increases the average realized value of the CSCF, but also has the effect of increasing the

spread of its values. This can be seen in the above graph, the variation range of the CSCF with a

probability of 90% increases with the reserve (initially empty) by nearly 5% compared to the case without

a reserve. This is due to the fact that only CSCF values which are close to 100%, corresponding to the

case where allocation is close to the cap, may benefit from the flexibility offered by the reserve. For lower

values, around 70%, free allocation demand remains significantly higher than the free allocation cap, so

that the flexibility offered by the reserve has no effect.

1 Following European Commission regulation No. 176/2014, the auctioning of 900 million allowances between 2014 and 2016

is postponed to 2019 and 2020.



23

Moreover, placing 900 million allowances into the reserve initially, for example from back-loading, would

not significantly reduce the uncertainty concerning the CSCF value, which would vary anyway between

80% and 99% with a probability of 90%. At least 1,500 million allowances would need to be placed into

the reserve, including allowances from back-loading and allowances which were not allocated in Phase III,

to significantly reduce uncertainty (The CSCF would vary between 95% and 100% with 90 % probability).

Removing the free allowance cap

A simple approach would be to remove the annual free allocation cap. The amount of allowances freely

allocated per unit of output would then be known with certainty, determined by the values of benchmarks.

This would however be likely to reduce the volumes of allowances auctioned, which would be detrimental

to the visibility of revenues raised by Member States This approach would as a result not be consistent

with Point 2.9 of the conclusions of the European Council in October 2014 stating that the share of

allowances auctioned should not be reduced compared to Phase III. Furthermore, this could have a

further distributional impact between sectors which receive free allowances and sectors which ensure

their compliance through purchasing auctioned allowances. The electricity sector could face prolonged

deficits in allowances, particularly because the stability reserve mechanism will already absorb a

significant number of allowances auctioned during the 2020 to 2030 decade.

Figures 19 and 20 illustrate the supply and demand resulting from output based allocation without any

free allocation cap (with a 1.4% annual growth assumption). The cumulative effect of the MSR

implemented in 2021 along with back-loaded allowances flooding into the market, and the additional free

allocation above the cap, results in an auctioning deficit of nearly 200 MtCO2e for the compliance of the

electricity sector each year of Phase IV.

Figure 19 - Supply and demand without

the free allowance cap – MSR 2021

Figure 20 - Auction deficit

for the electricity sector – MSR 2021


The implementation of the MSR from 2019 onwards combined with the placement of back-loaded

allowances straight into the reserve would however considerably reduce these interaction effects.



24

Figure 21 - Supply and demand without

the free allowance cap – MSR 2019

Figure 22 - Auction deficit

for the electricity sector – MSR 2019


Making free allocation more focused and targeted

The two solutions presented above are not, in themselves, fully satisfactory. Placing 1.5 billion allowances

into the reserve has unwarranted distributional impacts, as would the removal of the free allocation cap. A

prolonged auctioning deficit of 200 million allowances for the power sector would lead to great price

instability that would eventually be passed through to industrial installations. These alternative designs

should in any case, go hand in hand with a more focused and targeted allocation process, reflecting the

actual exposure to carbon leakages.

Gradual allocation with a more targeted carbon leakage list : CDC Climat Research’s proposal

One relevant solution would be implement a more targeted and tiered carbon leakage list, so that free

allocation volumes better reflect real exposure to carbon leakage.

The current list of sectors deemed to be exposed to a significant risk of carbon leakage covers 95% of

industrial emissions. Eligible installations receive the total amount of preliminary allocation only reduced

by the CSCF. This in/out mechanism allocates a large quantity of free allowances to sectors which are not

exposed. It is politicizing discussions concerning the revision of the list of sectors, although they should be

based essentially on technical grounds.

Implementing a more targeted list of sectors, which would gradually allocate allowances depending on

exposure to carbon leakage risks, would significantly reduce the initial allocation volume. The example of

criteria and thresholds below determining whether a sector belong to the list would enable a more tiered

allocation. According new criteria defined by CDC Climat Research, this method would have only

allocated 445 MtCO2e in 2013, compared to 712 MtCO2e under the current EU ETS list.

1. High risk - Carbon cost > 25% and trade intensity > 15% 100% of free allocation

2. Medium risk - Carbon cost > 15 % and trade intensity > 5% : 70 % of free allocation

3. Low risk - Carbon cost > 5 % and trade intensity > 10%: 40% of free allocation

Removing the 'degree of competition superior to 30%' criterion, which applies to sectors with a carbon

cost lower than 5% of added value, and therefore have very low exposure to the risk of carbon leakage,

would enable the allocation volume to be reduced by 62 MtCO2e. Allocating only 60% of defined

preliminary allocation for sectors corresponding to the second criterion (moderately exposed), then 30%

for sectors corresponding to the third criterion 3 (limited exposure), would enable the volume of free

allocations to be reduced by an additional 270 MtCO2e.

In the base case corresponding to annual growth of 1.4%, the total amount allocated to sectors in the list

would be 409 MtCO2e in 2030, lower than the free allowance cap of 500 MtCO2e. Annual average growth



25

of 2.2% would be needed for the allocation volume to once again exceed the allocation cap, requiring the

use of the CSCF. With an annual growth rate of 3%, the CSCF would be 89%.

In 2030, the mitigated carbon cost would be lower than 10% of sectoral added value for all sectors on the

list, which would meanwhile cover 88% of industrial CO2 emissions. Moreover, as long as average annual

growth rate in the 2015 to 2030 period remains below 2.2%, the CSCF ex post correction is rendered

unnecessary. This enables a greater stability in net anticipated cost for the sectors, varying from less than

1% of added value for all sectors with 90% probability (not taking into account the price uncertainty).

Figure 23 - Mitigated carbon costs within Scenario 3 with the use of a more targeted and gradual list

Source: CDC Climat Research (2015) based on European Commission, EUTL

Table 4 presents the allocation volumes in 2030 for the seven most energy-intensive sectors. They would

receive up to 356 MtCO2e free allowances of a total of 409 MtCO2e allocated. Meanwhile, these sectors

account for only 1,559 installations of the total 8,210 industrial installations in total. In order to mitigate the

administrative burden of output based allocation, it could thus be relevant to apply the mechanism only to

these identified sectors. The 53 MtCO2e left could be allocated to the 6,651 remaining installations

according to output levels which are updated less regularly.

Table 4 – Value of the Carbon Leakage Exposure Factor from 2013 to 2020 in the EU ETS

Source: CDC Climat Research (2015) based on European Commission and EUTL

Sectors Allocation in 2030

(MtCO2e) Number of

installations

Manufacture of lime and plaster 19 277

Manufacture of cement 77 320

Manufacture of coke oven products 5 28

Manufacture of fertilisers and nitrogen compounds 42 138

Manufacture of basic iron and steel and of ferro-alloys 132 528

Manufacture of refined petroleum products 77 179

Aluminium production 4 97

Total identified sectors 356 1567

Industry total 409 8,210



26

Ex post allocation to contain carbon costs below targeted values

Another design option could be to define targeted carbon costs for each sector according to their

corresponding trade intensity. A sectoral allocation coefficient may subsequently be defined ex post,

based on sectoral emissions, value added, and the average price of carbon recorded, that would ensure

that the targeted cost of carbon is not exceeded.

The maximum carbon cost could for example (Branger 2015) be set at 10% when the trade intensity gets

close to zero, and at 5% when the trade intensity is higher than 15%, and linear in-between as shown in

the red dotted line figure 24. According to data published by the European Commission in 2014, with a

30€/tCO2e carbon price, only 24 sectors located above this maximum carbon cost frontier would need to

be allocated. Among them, the main energy-intensive sectors including cement, steel and refining.

Figure 24 - Carbon costs distribution and the

targeted carbon cost frontier

Figure 25 - Carbon costs contained below targets

after allocation

Source: CDC Climat Research (2015) based on European Commission and EUTL

The chart in Figure 26 indicates estimations of allocation volumes that would be necessary to achieve the

targeted carbon costs for every sector. With the following assumptions of a carbon price at 30€/tCO2 and

an annual growth of 1.4%, 481 million allowances would be allocated in 2030, lower than the free

allocation cap in this horizon. Allocation figures concerning the 24 concerned sectors are provided in

annex 4.

Figure 26 – Mitigated carbon costs and allocation volumes

in the framework of implementing targeted carbon costs




27

Depending on the evolution of carbon prices, the sectoral free allocation rates would be adjusted in order

to stabilize the carbon costs. We calculate that under the assumptions of a 1.4% annual growth, free

allocation would remain below the cap as long as the CO2 price does not exceed 45€/tCO2e. Chart in

figure 27 illustrates how sectoral allocations would be adjusted according to the CO2 price. This

estimation does not take into account the feedback effect of carbon price increase on installations’

behaviors. A price increase would entail increased abatement efforts. As a result, fewer free allowances

would be necessary in practice to stabilize sectoral carbon costs.

Figure 27 – Sectoral allocation volumes and carbon prices

Source: CDC Climate Research (2015) based on European Commission, EUTL

4. Scenario 4: The European Commission’s proposal

New rules for a new ambition

The European Commission has proposed to continue using benchmark-based allocation in Phase III and

a free allocation budget of 40.4% (43% including the 400 million allowances from the innovation fund) of

the emissions cap within the period. This 43% level comes from the average share of free allowances in

Phase 3.

Furthermore, 400 million allowances will be placed in a New Entrant Reserve and made available for new

entrants and significant production increases, of which :

250 million allowances come from the Market Stability Reserve, corresponding to the amount not

allocated during phase III due to partial cessations of activity (according to the EC, 196 million

allowances have not been allocated for free in the 2013 – 2016 period due to partial cessations of

activity)

150 million allowances from the allocation budget that will not be allocated in Phase III due to the

application of the Carbon Leakage Exposure Factor declining from 80% to 30%, meaning that the

final allocation remains below the free allocation cap in phase III.



28

Figure 27 – Free allocation budget in phases III and IV


According to estimated industrial emissions1, the cumulated deficit of allowances will amount to 1450

million allowances in phase IV. However, if the 400 million allowances from the NER are released

throughout the period, the cumulative deficit would amount to only 1000 million allowances.

Table 5 - Free allocation and estimated emission between 2021 and 2030

Year 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 TOTAL

Free allocation 715 695 676 656 636 617 597 578 558 539 6 267

Estimated

emissions

758 761 764 767 770 773 777 780 783 786 7 720

Estimated

deficit

43 66 88 111 134 156 180 202 225 247 1 453

Source: CDC Climate Research (2015) based on European Commission

Installations deemed to be exposed to carbon leakages will receive up to 100% of benchmark-based

allocation, while other installations will receive only 30%.

Benchmark-based allocation will be determined for periods of five years.

In the period 2021-2025 and 2026- 2030, allocation will be determined based on updated activity

levels respectively from the years 2013-2017 and 2018-2022. In case of significant production

increases, activity levels will be adjusted by applying the same thresholds and allocation adjustments

that apply to partial cessations of operations. Allowances not allocated to installations due to

closures or partial cessation of operations shall be added to the New Entrants Reserve instead of

being auctioned.

Benchmark values will also be reduced in Phase IV relative to the current value which is based on

2007-08 data. It will decline by 1% each year between 2008 and the middle of the relevant free

allocation period, unless there is evidence that the values of a benchmark differ from the default

annual reduction by more than 0.5%, higher or lower. Benchmarks will be updated twice in Phase IV

of the EU ETS. The first update will provide stable values that will be used from 2021-2025. The

second update will concern the benchmark values applied as of 2026 and these values will in turn be

kept stable until 2030;

1 Assuming a 1.4% annual growth rate of activity levels and a 1% annual efficiency improvement



29

A sector is deemed to be at risk of carbon leakage if the multiplication of the two below factors exceeds

0.2:

Their trade intensity with third countries (calculated as the ratio between total value of exports to third

countries plus the value of imports from third countries and the total market size of the European

Economic Area - calculated as the annual turnover plus total import from third countries);

Their emission intensity (measured in kg/CO2 divided by the Gross Value Added).

The Figure 28 outlines the position of different sectors compared to the frontier between the two

categories of sectors. It has been calculated with data from the European Commission concerning the

2015-2019 carbon leakage list1. With the 0.2 threshold proposed, Sectors representing 93% of industrial

emissions are in the carbon leakage list.

Figure 28 – Distribution of sectors compared to the carbon leakage list frontier

Source: CDC Climat Research, based on European Commission

Potential impacts of the proposal

The Commission decision could lead to a 30% uniform reduction of allocation volumes by

2030, with levers to make free allocation more targeted to exposed sectors

In the proposal, benchmarks are reduced 1% per year from 2008 onwards. This will lead to a decrease of

free allocations to each sector, regardless of their exposure to carbon leakage. This automatic update of

benchmarks is equivalent to applying a uniform correction factor of 85% during the 2021 to 2025 period,

and of 80% during the 2026 to 2030 period. As such, it does not enable the distribution of free allowances

to those sectors most at risk, and does not improve the efficiency of the allocation method.

With the carbon leakage list proposed, a 1.4% annual growth until 2022 (reference year for the update of

activity levels in the period 2026 to 2030), a 1% annual decrease of benchmark values, the preliminary

allocation is estimated to be on the order of magnitude of 608 million allowances in the 2021-2025 period,

lower than the free allocation budget2, and thus no CSCF would be needed. Then the preliminary

allocation is estimated to be 620 million of allowances in the 2026-2030 period, higher than the free

allocation cap. This would entail a CSCF decreasing from 100% in 2026 to 86% in 2030. This CSCF

would come on top of the uniform reduction of 20% of the benchmarks. As such, the allocation would be

uniformly reduced by 30% in 2030, and the allocation rate would be of 70% in this time frame. With a

0.5% revision of all benchmarks, the CSCF reaches 79% in 2030, but the allocation rate remains at 70%

in the end. With a 0% revision of benchmarks, the CSCF is estimated at 70% in 2030.

Figure 29 – Preliminary allocation and the free

allocation cap until 2030

Figure 30 – Values of the CSCF and the free

allocation rate of industrial sectors

1 A 30€/tCO2 has been assumed for this calculation, but the results do not depend on this assumption.

2 Free allocation to the heat sector is assumed to decrease from 48MtCO2e in 2020 to 0 in 2027



30

As a result, free allocation does not seem to be targeted enough to the sectors most exposed sectors

which might face high carbon costs in the 2030 horizon.

Figure 31 – Estimated carbon costs in different sectors

Source: CDC Climat Research, based on European Commission

Building on the European Commission’s proposed mechanism, a more focused carbon leakage list could

be implemented. With a 0.8 coefficient to be lower than the product of trade intensity and emissions

intensity, instead of the 0.2, the list would cover only 78% of 2013 emissions as outlined in figure 32.

Figure 32 – Volume of the carbon leakage list for diff

Documents

No.51 September 2015 FREE ALLOCATION IN THE EUROPEAN ...€¦ · relating to the first phases of the EU ETS have not shown any evidence of carbon leakage (Reinaud, 2008; Sartor et