JOHANNES KEPLER UNIVERSITÄT LINZ Altenberger Straße 69 4040 Linz, Österreich jku.at Eingereicht von Philipp Mahringer Angefertigt am Institut für Volkswirtschaftslehre Beurteiler / Beurteilerin a.Univ.-Prof. Dr. Franz Hackl März 2021 AN ANALYSIS OF THE EU EMISSIONS TRADING SYSTEM – INSIGHTS FROM THE EUROPEAN UNION TRANSACTION LOG Diplomarbeit zur Erlangung des akademischen Grades Magister der Sozial- und Wirtschaftswissenschaften im Diplomstudium Wirtschaftswissenschaften (180)
105
Embed
AN ANALYSIS OF THE EU EMISSIONS TRADING SYSTEM – …
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
JOHANNES KEPLER UNIVERSITÄT LINZ Altenberger Straße 69 4040 Linz, Österreich jku.at
Eingereicht von Philipp Mahringer Angefertigt am Institut für Volkswirtschaftslehre Beurteiler / Beurteilerin a.Univ.-Prof. Dr. Franz Hackl März 2021
AN ANALYSIS OF THE EU EMISSIONS TRADING SYSTEM – INSIGHTS FROM THE EUROPEAN UNION TRANSACTION LOG
Diplomarbeit
zur Erlangung des akademischen Grades
Magister der Sozial- und Wirtschaftswissenschaften
im Diplomstudium
Wirtschaftswissenschaften (180)
2
EIDESSTATTLICHE ERKLÄRUNG
Ich erkläre an Eides statt, dass ich die vorliegende Diplomarbeit selbstständig und ohne fremde Hilfe verfasst, andere als die angegebenen Quellen und Hilfsmittel nicht benutzt bzw. die wörtlich oder sinngemäß entnommenen Stellen als solche kenntlich gemacht habe. Die vorliegende Diplomarbeit ist mit dem elektronisch übermittelten Textdokument identisch. Linz, 15.03.2021
CHAPTER 2. THE ECONOMICS OF EMISSIONS TRADING 2.1. THE ECONOMICS OF TPP
This is especially true for transnational ETS such as the European Union’s approach, which interact
with a multitude of existing national policies. Whereas such overlaps are not necessarily detrimental to
overall GHG abatement, studies on the EU ETS reveal that under certain circumstances, a combination
of complementary policies may lead to undesired effects. In particular, national measures subsidizing
the reduction of GHG emissions in certain industry sectors grant companies a surplus of allowances,
resulting in a lowered allowance price while leading to increasing emission levels in other areas. Hence,
the overall abatement efficiency is compromised in comparison to an isolated perspective treating both
the ETS and complementary policies as independent. This phenomenon, which is commonly referred to
as the waterbed effect, can be attributed to the static nature of the emissions cap and constitutes one of
the most widely discussed aspects of emissions trading in recent publications. From a theoretical point
of view, said detrimental effects can at least be mitigated by adapting the ETS to the policy mix both
on a national and on an EU-wide level (Görlach, 2013; Rosendahl, 2019).
2.1 The Economics of Tradable Pollution Permits
QaggQ3Q2Q1
P*=MACagg
P
MAC 1
MAC 2
MAC 3
MAC agg.
emissionreduction
Figure 2.2: The abatement-based model as an adaption of a general oligopoly.
As fig. 2.2 illustrates, the ideal representation of an emissions market is very similar to a general
oligopoly, in which each market participant has an individual marginal abatement cost or MAC curve.
Following this simplified economic model, the MAC curve across all market participants, which is rep-
resented by the dashed line in the diagram, is obtained by horizontally aggregating the individual MAC
6
CHAPTER 2. THE ECONOMICS OF EMISSIONS TRADING 2.1. THE ECONOMICS OF TPP
functions. We assume that a regulator sets a certain amount of necessary emission reductions Qagg based
on cost benefit considerations. The price for one emissions reduction is then determined by the given level
of Qagg and the aggregated MACagg function. As the sample calculation presented in equation 2.1 indi-
cates, the individual quantity Qi is dependent on MACi, meaning that each market participant reduces
emissions based on their individual MAC curve. Accordingly, firms with high MACs are likely to buy
allowances to fulfill their compliance obligation, whereas firms with low MACs are expected to reduce
emissions, resulting in lower average costs compared to a command and control scheme with identical
emissions abatement.
Assume MAC1 = 9Q1, MAC2 = 6Q2, MAC3 = 3Q3
Then the aggregated MAC curve is Q = P
9 + P
6 + P
3 = 11P
18 or P = 18Q
11If we assume Qagg then P ∗ = 18Qagg
11and, therefore, Q1 = P ∗
9 , Q2 = P ∗
6 , Q3 = P ∗
3
(2.1)
Hanley et al. (2008, pp.130), in turn, propose a different, damage or emission-based perspective to
illustrate this mechanism, the fundamental principle of which is identical to the aforementioned model.
Introducing a cap on emissions, the authors set the supply of allowances equal to MACagg, so that the
equilibrium price P ∗ can be calculated at the intersection of the aggregated marginal benefit function
of emissions and the emission supply function E. Note, that marginal benefits from emissions (MAB)
and marginal cost of emissions reductions can be traced back to the same economic fact – the amount
of income which can be earned with one unit of emissions. As fig.2.3 indicates, the emissions abatement
Qagg is defined as the difference between Ef , which equals P = 0 in fig. 2.2, and the cap on emissions
(E).
7
CHAPTER 2. THE ECONOMICS OF EMISSIONS TRADING 2.1. THE ECONOMICS OF TPP
P*
E Ef
MAC
Σi MABi = MBagg
emissions
NUMBER OF PERMITSEMISSION TARGET
Qagg
permits
Figure 2.3: The emission-based model. Adapted from Hanley, N., Shogren, J. F., & White, B. (2008).
Environmental economics in theory and practice (2nd edition). Basingstoke, Palgrave Macmillan.
8
Chapter 3
The European Union’s Emissions
Trading System
3.1 The History of the EU ETS
Designed as a means of meeting the GHG reduction targets agreed upon in the Kyoto protocol in 1997,
the EU Emissions Trading System or ETS was introduced in 2005 on the basis of a directive of the
European Parliament issued in 2003. Both its theoretical foundation and its implementation were first
discussed in a research paper published by the European Commission in 2000, which details on several
aspects that are crucial to the understanding of the EU ETS: Elaborating on the scope of a potential
European ETS, its participants, the sectors to cover, the level of centralization necessary, or potential
means of allocating allowances, the European Commission’s Green Paper on greenhouse gas emissions
trading within the European Union (European Commission, 2000) provides a theoretical framework for
the political process eventually leading to the implementation of the ETS in 2005. Both Ellerman et al.
(2010) and Skjærseth et al. (2016) give a detailed account of the events and developments that preceded
its creation. Starting their observation in the the early 1990’s, Ellerman et al. (2010, ch.2) state that
the European Commission’s original intent had been to install an EU-wide carbon tax. Nevertheless,
this attempt failed in 1992 due to the EC’s inability to reach a unanimous decision supported by all
member states. While during the Kyoto conference in 1997, the EC had still largely been opposed to the
implementation of a transnational ETS, they embraced the concept only six months later, positioning the
European Union as a global leader in environmental politics. This leadership role was further strengthened
when in 2001, US president George W. Bush decided not to ratify the Kyoto protocol.
CHAPTER 3. THE EU ETS 3.1. THE HISTORY OF THE EU ETS
As Skjærseth et al. (2016, pp.35) point out, the European Commission was still expecting an inter-
national agreement on emissions trading to be reached at the United Nations Framework Convention on
Climate Change (UNFCCC) in Buenos Aires in 1998, when it published its first communication on the
subject titled Climate Change – Towards an EU post-Kyoto strategy (European Commission, 1998). As
agreed at Kyoto, the EU was obliged to cut its greenhouse gas emissions by 8% compared to 1990 until
the end of the first commitment period from 2008 to 2012. In addition, the Kyoto agreement required
the EU to "make demonstrable progress in achieving its commitment by 2005" (Skjærseth & Wettestad,
2016, p.4), highlighting the necessity to develop an effective, community-wide GHG abatement strategy.
Based upon the results of the Vienna European Council in 1998, the European Commission compiled a
second and more elaborate communication titled Preparing for the Implementation of the Kyoto Proto-
col (European Commission, 1999), which was eventually published in 1999. However, both publications
give only a vague indication of a potential, European ETS, which is why, according to Skjærseth et al.
(2016, p.37), the aforementioned Green Paper is to be considered the actual starting point of the emission
trading system’s development phase.
In fact, the Green Paper proposes a cap-and-trade system imposing a centrally defined cap on the
annual emissions of the industry sectors covered rather than a baseline-and-credit system operating on
an installation level. Originally, six industry sectors covering about 45% of CO2 emissions were supposed
to be included in the proposed ETS – electricity& heat production, iron&steel, refining, chemicals, glass,
pottery&building materials as well as paper&printing. With regard to the organizational structure of the
proposed ETS, the Green Paper avoids dogmatism by suggesting several alternative strategies: Both low
and high levels of community harmonization, which translate into grades of member state autonomy, are
considered as viable options. Concerning allowance allocation, however, the paper highlights the technical
superiority of auctioning over the grandfathering approach which has eventually been implemented and
used during phase I&II of the actual EU ETS. Furthermore, the paper considers the possibility of a
voluntary or opt-out system covering only those member states which are willing to participate (European
Commission, 2000).
According to Skjærseth et al. (2016, p.40-46), the next major step towards a European ETS was taken
in 2002 with the publication of the Directive establishing a scheme for greenhouse gas emission allowance
trading within the community and amending council directive 96/61/EC (European Commission, 2002),
or, in short, ETS Directive, which, although based largely on the guidelines formulated in theGreen Paper,
differed from its predecessor in several aspects. For instance, the scope of industry sectors covered was
narrowed down to four activities – energy, production&processing of ferrous metals, mineral industry and
other activities – omitting the chemicals sector. Apart from that, a decentralized approach was favored
with regard to the issuance of allowances – the proposal drafts the implementation of National Action
Plans or NAPs which were used in the actual ETS until 2012. As to allowance allocation, grandfathering
10
CHAPTER 3. THE EU ETS 3.1. THE HISTORY OF THE EU ETS
was determined as the method of choice for phase I of the ETS, in which no legally binding emission caps
would apply.
As Skjærseth et al. (2016, p.103-111) argue, the change of direction reflected in the 2002 proposal
is best explained by the discussion process ensuing the publication of the Green Paper in 2000. In fact,
this period was marked by strong dissent among the EU’s member states concerning several key areas
of the proposed ETS. Whereas there was widespread support for the general concept of implementing
a harmonized system relying on a common allocation method as well as on community-wide monitor-
ing, reporting and verification standards, there was no consensus on both the mandatory nature of the
system and on the allocation strategy. For instance, Germany and the UK as the two most influential
member states, whereas for diferent reasons, favored a voluntary ETS for at least the initial trading
phase. In Germany, the discussion process was influenced by industrial organizations such as the BDI
(Bundesverband Deutscher Industrie) or the VCI (Verband der Chemischen Industrie), which vehemently
opposed the country’s participation in a centralized European ETS. Accordingly, Germany’s negotiating
position, which can be interpreted as a compromise between the BMU (Bundesministerium für Umwelt,
Naturschutz und nukleare Sicherheit) arguing in favor of the proposed ETS Directive and the BMWA
(Bundesministerium für Wirtschaft und Arbeit) taking the opposite stance, was aimed at promoting a
voluntary structure including opt-out options on a national, sectoral or installation level. However, after
extended negotiations lasting until December 2002, Germany was ready to accept the European Union’s
terms and give in to a mandatory ETS. Among the concessions made by the EU in the process, the most
notable is certainly the pooling provision which entered the ETS Directive as Article 28 and enabled
"member states" to "allow operators of installations...to form a pool of installations from the same activ-
ity for...the first five-year period" (European Commission, 2003). The UK, in turn, originally opposed
a centralized European ETS in favor of their own, domestic system, which was initiated as planned in
2002, comprising 34 industrial companies on a voluntary basis. Intended to run for five years until 2007,
the UK ETS also differed from its European counterpart in the inclusion of six GHGs instead of one as
well as in the sectors covered. Eventually, the UK, which had aimed at modelling the EU ETS to its own
approach during the negotiations, had to give in and reconsider their position – however, not without
the EU conceding an opt-out clause for installations on a national level (Skjærseth & Wettestad, 2016,
p.111).
Apart from the aformentioned concessions to Germany and the UK, three other propositions by
member states were accepted by the Environment Council which assembled in December 2002:
• "The possibility of unilateral additions of certain activities and gases from 2008;
• Free of charge allocation of allowances for the first phase and at least 90% free of charge allocation
in the second phase, thereby making the use of auctioning possible for member states who choose
11
CHAPTER 3. THE EU ETS 3.2. THE EVOLUTION OF THE ETS
to do so;
• Penalties to operators of 40 Euros in the first phase and 100 Euros in the second phase for each
excess tonne of carbon dioxide (CO2) emitted and not covered by sufficient allowances"
(Council of the European Union, 2002, pp.6).
On 13 October 2003, the first version of the ETS Directive (European Commission, 2003) was put into
effect, forming the legal basis of the European emissions trading system to be initiated by the beginning
of 2005. It has been amended several times since its publication – to be specific, in 2004, 2008, 2009,
2013, 2014, 2015, 2017 and 2018. The Linking Directive (European Commission, 2004), in turn, was
published in October 2004 and aimed at establishing a "link between the EU Emissions Trading Scheme
and the other two flexible mechanisms born out of the Kyoto Protocol – the JI and the CDM1" (Skjærseth
& Wettestad, 2016, p.45). As Skjærseth et al. (2016, p.45-47) point out, the months before its release
were marked by protests led by both industry and environmental NGOs (ENGOs). Whereas the former
demanded unrestricted transferablility of CERs, the latter feared negative effects on third world countries
as well as a dilution of GHG emission targets. Hence, several of the restrictions proposed by the European
Commission are not reflected in the final version of the ETS Directive: First, the EU-wide cap on CDM
credits was omitted in favor of limits imposed on a national level. Second, the use of CERs became
independent from the start of the Kyoto protocol’s first commitment period launched in 2008, meaning
that external allowances could be employed from 2005 onwards and third, the once permanent exclusion
of nuclear projects was reduced to a temporary ban.
3.2 The Evolution of Emissions Trading in Europe
Eventually, on 1 January 2005, the European Union Emission trading scheme was officially initiated,
covering about 11.500 installations in 25 member countries (Ellerman et al., 2010, ch.1). According
to the ETS handbook (European Commission, 2015b, p.7), the first two years following the system’s
implementation in 2005 were intended as a "pilot phase", the primary objective of which was to create
the infrastructure required for its regular operation. While in its early stages, the ETS had been limited
to carbon dioxide emissions originating from the 25 member states of the EU, its scope was extended over
the years: According to the European Environment Agency (Cludius et al., 2019, pp.25), installations
from Bulgaria and Romania have been covered since the beginning of 2007, while Liechtenstein and1Both Joint Implementation and Clean Development Mechanism are instruments under the Kyoto protocol, which
enable participating countries to substitute domestic GHG abatement with investments in equivalent international projects.
Whereas JI is limited to countries with binding emission limits, CDM aims exclusively at projects in developing countries.
Both mechanisms issue credits – emission reduction units or ERUs for JI projects and certified emission reduction units or
CERs for CDM projects – which can be used in the EU ETS.
12
CHAPTER 3. THE EU ETS 3.2. THE EVOLUTION OF THE ETS
PHASE I2005-2007
PHASE II2008-2012
PHASE III2013-2020
PHASE IV2021-2030
Assist.-Prof.in Mag.a Dr.in Christine Blanka
GRANDFATHERING BENCHMARKING
NATIONAL ALLOCATION PLANS NATIONAL IMPLEMENTATION MEASURES
CARBON LEAKAGE POLICY
INDEPENDENT NATIONAL REGISTRIES COMMON UNION REGISTRY
EUTL
MRV - MONITORING, REPORTING, VERIFICATION
EUAA - AVIATION ALLOWANCES
MSR - MARKET STABILITY RESERVE
AUCTIONING
Figure 3.1: Development of the EU ETS from 2005 to 2030
Norway joined the ETS in 2008. Iceland and Croatia, in turn, were introduced by the beginning of phase
III in 2013. With regard to the greenhouse gases covered, N2O emissions, which had been included via an
opt-in process by several member states (AT, NL, NO, IT, UK) during phase II, were introduced EU-wide
in 2013. This implied the inclusion of several new activities, among them "the production of nitric and
apidic acid, glyoxal and glyoxilic acid" (2019, pp.25) as well as of perfluorcarbons or PFCs stemming
from the production of aluminium. Extending the scope of the ETS, the European Commission agreed
on including emissions from the aviation sector by the beginning of 2012 (European Commission, 2008).
Designed as a cap-and-trade system, the EU ETS imposes an upper limit on the total emissions
released by all installations. Whereas during phase I&II, said cap was fixed, the European Commission
agreed on a linear decrease of 1.74% p.a. from 2013 onwards. Starting with 2.35 billion tons of CO2 p.a.
in 2005, the cap was lowered to 2.1 billion tons in 2008. For 2013, a base value of 2.084 billion tons was
determined, yielding an annual reduction of 38.3 million tons of CO2. The aviation cap, in turn, has
been fixed to 210 million allowances p.a. during phase III. Whereas throughout phase I&II, the cap on
emissions had been set on a national level, the EU has been in control of its level since 2013 (European
Commission, 2015b, pp.22).
13
CHAPTER 3. THE EU ETS 3.2. THE EVOLUTION OF THE ETS
Acknowledging that the allocation and surrendering of allowances are among the core activities of a
cap-and-trade system, it is necessary to establish clear guidelines as to how and when these processes are
scheduled. For this purpose, the European Commission has agreed on a set of dates and deadlines which
are mandatory for installation holders participating in emissions trading and national registries alike:
1. Until February 28, allowances for the current trading period are allocated to both stationary in-
stallations and aircraft operators
2. Until 31 March, account holders are obligated to submit the verified emissions for the preceding
period to the national authorities for approval
3. Starting with 1 April, sanctions for accounts and account holders failing to submit their verified
emissions apply. With regard to the punishment of non-compliance, Art. 16 of the ETS Directive
(European Commission, 2020e) grants member states a high degree of autonomy. However, a
penalty of 100 EUR adjusted for inflation from 2013 onwards has been set for each ton of carbon
dioxide equivalent which is incorrectly declared. Discrepancies between the emissions reported for
each installation and the emissions verified by the national authorities are to be accounted for in
the following year.
4. Until 30 April, operators are obliged to surrender the number of allowances corresponding to the
verfied emissions of the last period
5. Starting with 1 May, data on verified emissions as well as on surrendered allowances and compliance
for the previous year is published via the EUTL website.
(Deutsche Emissionshandelsstelle (DEHSt) im Umweltbundesamt, 2017, p.8)
Since the allocation of allowances takes place two months before the deadline for surrendering, it
is possible to use these newly acquired allowances to fulfill the compliance obligation of the previous
period. However, this practice referred to as borrowinng is only possible within the limits of each trading
phase, meaning that allowances acquired in one phase of the ETS cannot be surrendered in the previous.
In a similar fashion, positive account balances may be transferred between trading periods. Since the
transition from phase I to phase II, allowances no longer 7expire, so that market participants may use
allowances from previous trading periods for surrendering in the current period. According to data
published by the EC in 2020 (European Commission, 2020d), the number of allowances banked from
phase II amounts to 1.75 billion. So far, there is no indication that the European Commission is going
to change its position concerning the banking of allowances in phase IV (European Commission, 2015b,
p.133).
14
CHAPTER 3. THE EU ETS 3.3. CORE COMPONENTS OF THE EU ETS
In addition to the specifics of the system already discussed in this section, a broad range of measures
have been implemented, adapted and improved during the first 15 years, the understanding of which is
crucial for the empirical part of my thesis. Fig. 3.1 gives an overview of the milestones in the evolution
of the EU ETS from 2005 to 2020 while providing an outlook on key changes to the system during phase
IV. In the following section, I give a detailed account of the development and the current state of several
key components of the ETS, ranging from allowance allocation and the central registry to the market
stability reserve and the monitoring, reporting and verification system.
3.3 Core Components of the EU ETS
3.3.1 The Allocation of Allowances
The allocation of allowances is one of the core elements of the European Union’s ETS. While throughout
the first years of its existence, free allocation, which is often referred to as grandfathering, had been
the primary means of issuing allowances, auctioning has become the standard allocation method since
the beginning of phase III and is going to gain further importance during the next phase starting in
2021. This development can be traced along the evolution of the ETS Directive, which has been updated
several times since its initial publication in 2003. Accordingly, the allocation policy employed in phase
I&II is based upon Art. 10 of the initial version of the EU ETS Directive (European Commission, 2003),
which states that "Member States shall allocate at least 95% of the allowances free of charge" during a
period from 2005 to 2007 and at least 90% during phase II from 2008 to 2012. According to the ETS
handbook (European Commission, 2015b, p.28), the remaining 5% in phase I and 10% in phase II were
available for auctioning. Nevertheless, this right was scarcely exercised, leading to only 4% of all auctions
being auctioned during phase II. In order to determine the number of allowances allocated to individual
installations, member states were required to submit National Allocation Plans or NAPs detailing both on
the quantity of allowances and on the allocation method employed during a certain period. In accordance
with Art. 9 of the ETS Directive (European Commission, 2003), the national NAPs, which were to be
established at least 18 months before the period they were applied in, were subsequently assessed by the
European Commission on the basis of a set of criteria listed in ANNEX III.
By the beginning of phase III, substantial changes were implemented, the main aim of which was to
reduce the amount of free allocation in favor of auctioning as the primary allocation method. According
to Art. 10a of the 2009 version of the ETS Directive (European Commission, 2009c), the share of free
allocation for installations in the power generation sector has been cut to 0% by 2013, with the exception
of modernization measures meeting the criteria listed in Art. 10c. These are targeted at member states,
which, according to Art. 10c(1), were either
15
CHAPTER 3. THE EU ETS 3.3. CORE COMPONENTS OF THE EU ETS
• "not directly or indirectly connected to the network interconnected system operated by the Union
for the Coordination of Transmission of Electricity (UCTE)" by 2007,
• connected only "through a single line with a capacity of less than 400 MW" or in which
• "more than 30% of electricity was produced from a single fossil fuel, and the GDP per capita at
market price did not exceed 50% of the average GDP per capita at market price of the Community"
in 2006.
With regard to all remaining industries which are not at risk of carbon leakage2, Art. 10a of the
ETS Directive (European Commission, 2009c) sets a benchmark for free allocation decreasing from 80%
in 2013 to 30% in 2020, which is also referred to as the carbon leakage factor, or CLEF. In fact, the
most radical change implemented by the beginning of phase III is the introduction of a benchmarking
approach to replace the previous allocation method based on grandfathering. According to the ETS
handbook (European Commission, 2015b, p.40), said strategy relies on product-related GHG emission
benchmarks based on the average CO2-efficiency of the top 10% of all installations of each sector rather
than using historical emissions data on an individual level. The authors conclude that unlike grandfather-
ing, benchmarking "allocates allowances based on their production performance instead of their historical
emissions", ensuring that efficient installations are granted a comparative advantage while creating an
incentive for inefficient installations to modernize their production process. Accordingly, a comprehensive
list of 52 products (European Commission, 2011a) covering about 75% of all industrial emissions subject
to the EU ETS has been published by the European Commission for phase III. As the ETS handbook
(European Commission, 2015b, p.103) states, a single installation may produce more than one of the
listed products, which requires the creation of sub-installations in order to calculate the total quantity
of allowances allocated to an applicant. In case a sub-installation is not covered by any of the product
benchmarks listed, three fall-back options have been defined, the first of which is a benchmark for measur-
able heat production using a transfer medium such as water or steam, which is determined as the relation
between emission intensity and net calorific value of natural gas and assumes 90% conversion efficiency in
heat production. The second benchmark, which is based on fuel consumption, also relies on the emission
efficiency of natural gas, whereas the third benchmark is targeted at so-called process emissions, which,
according to the guidelines of the European Commission (European Commission, 2011c, p.8), includes
both "non-CO2 greenhouse gas emissions" and "emissions from the combustion of incompletely oxidised
carbon". In total, the European Commission has published 8 reference guides related to free allocation,
a detailed assessment of which, however, exceeds the scope of this thesis.2Carbon leakage is the presumed tendency of European firms affected by the ETS to shift production to countries with
more lenient environmental standards. For further information, see section 3.3.2.
16
CHAPTER 3. THE EU ETS 3.3. CORE COMPONENTS OF THE EU ETS
Returning to the standard case, the following formula is used to calculate the amount of free allocation
×(Cross-Sectoral Correction Factor OR Linear Reduction Factor)(3.1)
(European Commission, 2015b, p.44)
The second of the eight guidance documents (European Commission, 2011c) contains detailed infor-
mation on all factors of the equation, including the HAL (historical activity level), the CLEF (carbon
leakage exposure factor), the CSCF (cross-sectoral correction factor) and the LRF (linear reduction fac-
tor). The HAL or historical activity level is defined as the median of an installation’s or sub-installation’s
annual activity levels during a baseline period either form 2005 to 2008 or from 2009 to 2010, whereas
the period with the highest activity has to be selected. According to guidance document no.9 (European
Commission, 2011d), which includes an exhaustive list of product types with specific information on
the applicability of free allocation, said activity levels are usually reported in metric tons of production
according to a set of criteria which is not consistent across industries. While installations deemed at
risk of carbon leakage are granted free allowance allocation of up to 100% of the product benchmark,
all sectors which are not included in the current list compiled by the European Commission (European
Commission, 2014a, (13)), receive allowances on the basis of the CLEF. The cross-sectoral correction
factor or CSCF, which is intended to prevent the number of allowances allocated for free from exceeding
the maximum value defined by Art. 10a(5) of the 2009 ETS Directive, applies to all installations, which
are "not identified as ’electricity generator’" (European Commission, 2011b, pp.18). Installations used
for the generation of electricity as well as new entrants, in turn, are subject to the linear reduction factor
or LRF. Both the CSCF and the LRF are compiled annualy by the European Commission on the basis
of the national implementation measures.
3.3.2 Carbon Leakage
The term carbon leakage refers to a firm’s tendency to shift its production to countries outside the EU
in reaction to GHG reduction measures, particularly the EU ETS. According to the ETS handbook
(European Commission, 2015b, p.60-62), these policies result in a competitive disadvantage for firms
in the European Union, which particularly affects energy-intensive industries. While the authors argue
that there is currently no empirical evidence supporting the existence of this phenomenon, the European
Union has been implementing various various measures to tackle carbon leakage and is determined to
sustain these in future evolutions of the ETS. This assessment of carbon leakage is in line with research
by Naegele&Zaklan (2019), who analyze the impact of the ETS on international trade flows to and from
European manufacturing sectors, finding no evidence to support the existence of this effect.
17
CHAPTER 3. THE EU ETS 3.3. CORE COMPONENTS OF THE EU ETS
Taking a closer look at the evolution of the ETS Directive starting from 2003, it becomes apparent
that the issue of carbon leakage has not been addressed until mid 2013 (European Commission, 2013b).
This is not surprising, considering the fact that the measures proposed to prevent this phenomenon
consist primarily of the prolongation of free allocation for certain industry sectors, which had already
been the preferred allocation method during the first two phases. In order to determine, which sectors
are eligible for these subsidiaries, a two-fold assessment strategy consisting of both a quantitative and
a qualitative analysis has been created. This system, which is going to be replaced by the beginning of
phase four, has been employed to create two lists of industries prone to carbon leakage, with the first one
being applied from 2013 to 2014 and the second one from 2015 to 2020. Currently, the list contains a
total of 245 industry sectors based on the NACE scheme as well as an additional 24 subsectors based on
the CPA- or PRODCOM-classification, which, according to Eurostat’s NACE handbook (Eurostat, 2008,
p.42), serve as extensions to the 4-digit NACE-code (European Commission, 2014a, (13)). Following the
set of criteria defined by the quantitative method, an industry sector is at risk of carbon leakage if:
• "direct and indirect costs induced by the implementation of the directive would increase production
cost, calculated as a proportion of the gross value added, by at least 5% and
• the sector’s trade intensity with non-EU countries (imports and exports) is above 10%" (European
Commission, 2020c).
• In addition, a sector or sub-sector is considered at risk if "the sum of direct and indirect additional
costs is at least 30%" or if
• the non-EU trade intensity is above 30% (European Commission, 2020c).
To complement the quantitative method, the EU has established a framework to assess the eligibility of
industries that do not meet the above requirements. According to Art. 10a(17) of the ETS Directive issued
in 2009 (European Commission, 2009c), the qualitative assessment is based on the following criteria:
• "The extent to which it is possible for installations in the sector to reduce their GHG emissions or
electricity consumption through additional investment
• The current and projected market characteristics of the sector, such as the market concentration,
homogeneity of the product, competitive position relative to non-EU producers and bargaining
power of the sector in the value chain
• Profit margins of the sector as an indicator for the ability to absorb costs and long-run investment
or relocation decisions."
18
CHAPTER 3. THE EU ETS 3.3. CORE COMPONENTS OF THE EU ETS
In order to complement the aforementioned provisions for carbon leakage and address an issue referred
to as indirect emission costs, Art. 10a(6) of the 2009 ETS Directive (European Commission, 2009c) defines
that member states are allowed to financially compensate electricity-intensive installations for increased
energy prices attributable to the EU ETS. As stated in the Commission’s Official guidelines on state aid
measures in the context of the ETS (European Commission, 2012b, Art. 3.1), these subsidiaries may be
granted directly via national state aid schemes to a limited number of industries. All sectors eligible for
financial compensation are mentioned in an exhaustive list, which has been compiled based on criteria
similar to those applied for measures concerning direct emission costs (European Commission, 2012b,
ANNEX II). In total, this includes 15 industry sectors ranging from "Aluminium Production" to "Mining
of Iron Ores".
Among other changes addressed in the previous chapters, the latest version of the ETS Directive
(European Commission, 2018c) includes a reformed assessment strategy for identifying industries prone
to carbon leakage, which is going to replace the procedures listed in Art. 10a of previous versions in
phase IV. The most notable alteration refers to the quantitative analysis, which is now based on a single
benchmark calculated by multiplying an industry’s "intensity of trade with third countries...by their
emission intensity measured in kg CO2, divided by their Gross Value Added", whereas the trade intensity
is "defined as the ratio between the total value of exports...plus the total value of imports from third
countries and the total market size of the European Economic Area", which, in turn, equals the "annual
turnover plus total imports from third countries" (European Commission, 2018c, Art. 10b(1)).
Risk of Carbon Leakage = Intensity of Trade × Emissions Intensity (kg CO2)Gross Value Added
Intensity of Trade = Total Value of Exports + Total Value of ImportsTotal Market Size of EEA (Annual Turnover+Imports)
In case the result of this calculation exceeds a threshold of 0.2, an industry is eligible for up to 100% free
allocation until 2030. According to Art. 10b(2), industries failing to meet this criterion while yielding
a value of at least 0.15 may qualify for the same subsidiaries based on a qualitative assessment. The
same goes for industries, for which the ratio between emission intensity and gross value added exceeds
1.5. In 2017, the European Commission has initiated a process to reevaluate, which industries are at risk
of carbon leakage (European Commission, 2019e). So far, a preliminary list has been compiled, which is
still pending for adoption (European Commission, 2019a) .
19
CHAPTER 3. THE EU ETS 3.3. CORE COMPONENTS OF THE EU ETS
3.3.3 The Auctioning of Allowances
While during the first two trading periods of the EU ETS, grandfathering had been the primary method of
allocating allowances, auctioning has been gaining importance since the beginnning of phase III. Starting
with a legal limit of 5% of alllowances to be auctioned in 2005, the cap was raised to 10% in 2008.
However, according to the ETS handbook (European Commission, 2015b, p.28), this option was hardly
ever used in phase I, whereas only 4% of allowances were auctioned between 2008-2012. As stated in
Art. 10 of the 2009 ETS Directive (European Commission, 2009c), "Member states shall auction all
allowances which are not allocated free of charge" from the beginning of phase III onwards in line with
a new Auctioning Regulation (European Commission, 2010), which was last amended in 2019 (European
Commission, 2019c) to reflect changes affecting the fourth trading period of the EU ETS. For the aviation
sector, in turn, a fixed upper limit of 15% of the total allocated volume has been in effect since 2012. For
the fourth trading period to be initiated in 2021, Art. 10(1) of the latest version of the ETS Directive
(European Commission, 2020e) sets a target of 57% of all allowances to be auctioned, whereas a further
2% are intended for the creation of a so-called modernisation fund supporting investment projects in
member states with a per-capita GDP amounting to less than 60% of the EU average in 2013. As to
the distribution of allowances intended for auctioning, Art. 10(2) specifies, that 90% of allowances for
auctioning are allocated to member states in accordance with the share of verified emissions said state
has reported either in 2005 or from 2005-2007. The remaining 10% are reserved for member states with
a comparatively low per-capita GDP, which are listed in ANNEX IIa of the ETS Directive.
However, for the current trading period, slightly different regulations referred to in earlier iterations
of the ETS Directive apply: From 2013 to 2020, 88% instead of 90% of allowances auctioned have been
distributed according to each member state’s share of verified emissions in either 2005 or from 2005 to
2007, whichever yields the highest value. In congruence with the revised directive, 10% are assigned to a
list of member states defined in ANNEX IIa. Further 2% are allocated to member states which reported
an emission reduction of at least 20% from their respective base period specified in the Kyoto protocol
and 2005, which are listed in ANNEX IIb of the 2013 ETS Directive (European Commission, 2013b).
The European Commission’s 2019 Report on the functioning of the European carbon market (2019g,
pp.22) provides an insight into the revenues generated by auctioning during phase III: From 2012 to 2019,
these amounted to EUR 42 billion, 14 billion of which were generated in 2018 alone. Furthermore, about
80% of these revenues have been used for "specified climate and energy related purposes" from 2013 to
2018 in accordance with Art. 10(3) of the ETS Directive, which demands a minimum share of 50% to be
employed for climate-related projects.
20
CHAPTER 3. THE EU ETS 3.3. CORE COMPONENTS OF THE EU ETS
According to Art. 17 of the Auctioning Regulation (European Commission, 2010), "auctions should be
carried out by means of a single-round, sealed-bid and uniform-price format". This implies that during a
bidding window of at least two hours, registered bidders are able to submit an unlimited number of bids on
lots of 500 allowances. Upon closure, the clearing price at which all allowances in an auction are allocated
is determined as the "price of the bid at which the sum of the volumes bid matches or exceeds the volume
of allowances auctioned". In case the clearing price is lower than the auction reserve price or if the volume
of bids is lower than the number of allowances or lots auctioned, the auction is automatically canceled
(European Commission, 2019c, Art. 7). According to the ETS handbook (European Commission, 2015b,
p.32), both the EEX and the ICE deliver auctioned allowances within one business day as either two-day
spot or five-day futures (European Commission, 2019c, Art. 4). According to Art. 18, access to auctions
is restricted to the following entities: "companies in possession of an operator holding account or aircraft
operating holder account, authorised investment firms, authorised credit institutions and public bodies or
state-owned entities in possession of an OHA or AOHA".
With regard to the practical implementation of the auction process, a common auctioning platform is
nominated for a period of up to five years based upon the guidelines specified in the Joint Procurement
Agreement (European Commission, 2011e). Currently, the European Energy Exchange (EEX) in Leipzig
fills this role for 28 states, with the remaining three – Germany, Poland and the UK – either appointing
the EEX as their opt-out platform, or in case of the UK, using the London-based ICE Futures Europe
instead (European Commission, 2020b). For all states covered by the joint procurement agreement,
weekly auctions are scheduled on Mondays, Tuesdays and Thursdays, whereas German auctions take
place on Fridays. Poland, in turn, holds auctions on a monthly basis, whereas the UK uses a two-week
interval with auctions on Wednesdays (European Commission, 2015b, p.33).
3.3.4 The Union Registry and the EUTL
In the course of a centralization process initiated by the 2009 ETS Directive (European Commission,
2009c), the national registries maintaining the operation of the EU ETS during the first two trading
periods were merged into one common registry under the responsibility of the European Commission.
The Union Registry was created as a centralized system aimed at managing all accounts held by both
natural persons, companies and member states within the ETS while recording transactions with both
european emissions allowances and international CERs and ERUs. In addition, it monitors the allocation
of allowances projected by the national allocation tables as well as the verified emissions both on an
installation and on a national level. As a second line of defense to ensure the integrity of the system,
the EUTL or European Union Transaction Log constantly checks and validates registry data in what is
referred to as the reconciliation process by Art. 103 of the Registry Regulation released in 2013 (European
21
CHAPTER 3. THE EU ETS 3.3. CORE COMPONENTS OF THE EU ETS
Commission, 2013a), which constitutes the legal basis of both the registry and the EUTL. In addition,
the EUTL serves as a public frontend to the registry. ANNEX XIV of the Registry Regulation details
on the legal requirements for reporting via the EUTL website by providing a comprehensive list of all
information to be made available to the public:
• First, all account data indicated as "displayed on the EUTL public website" in Table I-III of ANNEX
III, including account type, commitment period, account holder name, account holder address,
company registration number, account opening date and account closing date. However, specific
information such as account IDs or account identifiers are only publicised for operator holding
accounts in accordance with table VI-I of ANNEX VI.
• Second, all completed transactions registered by the EUTL with a delay of three years, updated on
a yearly basis on 1 May. This entails information on the transferring and on the acquiring account
involved in a given transaction, their national registries, time and date as well as an identification
code. Information on transactions involving Kyoto units, in turn, is limited.
• Third, data aggregated on a national and EU-wide level, for instance, the national allocation tables
as well as the international credit entitlement tables of all member states or the total number of EU
allowances, ERUs and CERs within the ETS. This also comprises transactions issued in compliance
with the Effort Sharing Decision3.
With the exception of the transaction log and certain other sources, all data is to be updated on a daily
basis.
3.3.5 Monitoring, Reporting, Verification
In order to ensure that operators meet their compliance obligation, the European Commission has created
a system which is referred to as MRV or Monitoring, Reporting and Verification. Based on experience
gained from phase I&II of the ETS, the EU Monitoring and Reporting Regulation or MRR (European
Commission, 2019d) establishes comprehensive guidelines for the verification of emissions. The MRR,
which entered into force by the beginning of phase III in 2013, requires both aircraft and installation op-
erators to submit to an annual compliance cycle involving three national authorities: First, the competent
authority – an agency appointed by each member state, which is not only responsible for the approval
of monitoring plans specifying the monitoring responsibilities of each operator and the issuance of GHG
permits, but also for the inspection and enforcement of the MRV process. In Austria, this function is3The Effort Sharing Decision or ESD (European Commission, 2009b), (European Commission, 2018e) establishes binding
standards for GHG abatement in each member state, including both industry sectors covered by and independent from the
EU ETS.
22
CHAPTER 3. THE EU ETS 3.3. CORE COMPONENTS OF THE EU ETS
assigned to the Ministry of Sustainability and Tourism. The national accreditation body, in turn, is part
of the Ministry for Digital and Economic Affairs and appoints the verifiers, which, in turn, are responsi-
ble for monitoring the annual emission reports (AER) submitted by operators until March 31 each year.
Currently, three institutions – TÜV SÜD Landesgesellschaft Österreich GmbH, TÜV Austria Services
GmbH and Lloyd’s Register EMEA – are accredited for this purpose in Austria (Bundesministerium für
Digitaliserung und Wirtschaftsstandort, 2019).
In order to participate in the EU ETS, an operator’s first step is to develop a monitoring plan estab-
lishing reproducible and transparent procedures to monitor an installation’s GHG emissions. According
to Chapter 3.3 of the EC’s Guidance document on the Monitoring and Reporting Regulation (European
Commission, 2017), this comprises the following aspects: data collection – whether emissions are calcu-
lated or recorded directly via a CEMS4, measuring procedures, including laboratory analyses, sampling
of materials and fuels as well as calibration of measuring equipment, control procedures, data storage and
constant evaluation of the procedures used. With regard to monitoring, standards differ by an installa-
tion’s average annual emissions. Art. 19(2) of the MRR lists three categories with rising requirements for
data quality: A for installations ≤ 50.000 metric tons of annual carbon dioxide emissions, B for values of
up to 500.000 tons and C for emitters with a total exceeding 500.000 tons. For operators, this classifica-
tion translates into a tier-based system, which defines standards for accuracy, precision and uncertainty.
According to Art. 47 of the MRR, a simplified approach applies for installations with an average emis-
sion of less than 25.000 metric tons per year: member states may supply these smaller emitters with
standardized monitoring plans in order to reduce the administrative burden. Also, lowered standards for
data collection, uncertainty assessment and verification apply.
As soon as the monitoring plan is accepted by the competent authority, a GHG permit is issued,
upon which the operator is obliged to request an Operator Holding Account within 20 working days
(European Commission, 2019d, Art. 17(1)). After the OHA has been set up, the monitoring cycle
starts in accordance with the MRR: By March 31 each year, operators are obliged to submit an annual
emission report (AER) for the preceding year to the competent authority. Prior to submission, the AERs
are verified by one of the aformentioned institutions appointed by the national accreditation body in
accordance with the Accreditation and Verification Regulation 2012/600 (European Commission, 2012a)
which was replaced by Regulation 2018/2067 (European Commission, 2018a) in 2018. In addition to
the annual emission reports, operators are be required to submit so-called improvement reports (IR) if
certain conditions detailed in Art. 69 of the MRR apply. Also, changes to the capacity, activity level and
operation of an installation are to be reported to the competent authority by December 31 in accordance
with Art. 24(1) of Commission Decision 2011/278 (European Commission, 2011a).4continuous emission measurement system
23
CHAPTER 3. THE EU ETS 3.3. CORE COMPONENTS OF THE EU ETS
3.3.6 The NER and the NER 300 Programme
According to Art. 10a(7) of the 2014 version of the ETS Directive (European Commission, 2009c), five
percent of the EU-wide cap on emissions from 2013 to 2020 has been reserved for free allocation to new
entrants – a term which refers to installations which have either "obtained a greenhouse gas emissions
permit for the first time after 30 June 2011" or to installation which have "had a significant extension after
30 June 2011, only in so far as this extension is concerned" (European Commission, 2009c, Art. 3(h)).
For the period from 2013 to 2020, the European Commission reports a total of 145.8 million allowances
allocated to 996 installations, 568 of which were already in operation before 2011. Another 23.9 million
allowances were still awaiting allocation as of 15 January 2020. These two values combined represent
about 35% of the total volume of 480.2 million reserved for allocation to new entrants. The remaining
65% or 310.5 million allowances will be made available through the Market Stability Reserve in phase IV
(European Commission, 2020a).
In accordance with Art. 10a(8) of the ETS Directive (European Commission, 2009c), another 300
million allowances were directed to a fund referred to as the NER 300 programme, the original objective
of which was "to help stimulate the construction and operation of up to 12 commercial demonstration
projects that aim at the environmentally safe capture and geological storage (CCS) of CO2 as well as
demonstration projects of innovative renewable energy technologies" until 31 December 2015. However,
as the final progress report on the implementation of the NER 300 funding programme (European Com-
mission, 2020f) states, both the programme’s deadline and its limitations were extended, so that in total,
2.1 billion € in funding was awarded to 39 project in 20 member states, 19 of which were still active in
December 2019. Of these 19, one had already been completed, further 9 were in operation and finally, 9
projects were not yet operational.
Funding was organised in two trenches, with the first one in 2012 covering EUR 1.1 billion or 200
million allowances, and the second one initiated in 2014 covering EUR 1 billion or 100 million allowances
plus all allowances not used in the first round (European Commission, 2020i). In order to qualify for
the NER 300 programme, applicants were required to submit to an assessment process administered by
the member states based on eligibility criteria stated in ch.5.1 of the Call for Proposals published by
the European Commission in 2013 (European Commission, 2013b). For instance, the paper presents a
comprehensive list of technology categories eligible for subsidies. Furthermore, different requirements for
renewable energy and CCS projects apply with regard to the innovative nature of the technology used
and the implementation of the project. Also, the NER 300 programme imposes capacity thresholds as
well as deadlines for a project’s entry into operation: Not only were applicants required to obtain permits
in advance, but also they were expected to commence commercial operation by 30 June 2018.
24
CHAPTER 3. THE EU ETS 3.4. THE FUTURE OF THE EU ETS
3.3.7 The Market Stabilty Reserve
In reaction to a surplus of allowances accumulating in connection to the economic crisis since 2009, the
European Commission has implemented measures to counter the steady decline of the allowance price
reaching its all-time low in 2013. By postponing the auctioning of 900 million allowances from 2014-2016
until the end of phase III, the EC managed to reduce the surplus from an initial 2 billion allowances
in 2012, followed by an even higher 2.1 billion in 2013, to about 1.78 billion in 2015, which equals a
reduction of 30% compared to the projected value without intervention (European Commission, 2020g).
The practice of back-loading allowances, which was legitimized in 2014 by an amendment to the Auctioning
Regulation (European Commission, 2014b), has eventually been replaced by a mechanism referred to as
the Market Stability Reserve in January 2019. The MSR, which was established on the basis of a decision
issued in 2015 (European Commission, 2015a), serves two main purposes – to manage and distribute
the existing surplus of allowances on the one hand and to increase the stability of ETS in the event of
economic crises by controlling the supply of allowances on the other hand.
The system operates as follows: If the number of allowances in circulation is higher than 833 million,
the surplus is added to the reserve. Whenever said number is lower than the threshold of 400 million,
allowances from the reserve are distributed. Accordingly, both the allowances withheld from 2014-2016
and all unallocated allowances from 2019 onwards are going to be transferred to the MSR. From 1
September 2020 to 31 August 2021, the total transfer volume will amount to 332,519,000 allowances
(European Commission, 2020d, p.5).
On the basis of Directive 2018/410 (European Commission, 2018b), the mechanism of the MSR for
phase IV was designed to extend this principle: Until 2023, "the percentage of the total number of
allowances in circulation determining the number of allowances put in the reserve if the threshold of 833
million allowances is exceeded is temporarily doubled from 12% to 24%", meaning that over a period of
12 months from 1 September 2020 onwards, at least 200 million allowances are going to be withheld from
auctions. From 2023 onwards, any number of allowances exceeding the previous year’s auction volume
will be invalidated (European Commission, 2020d, p.2).
3.4 The Future of the EU ETS
In order to meet the ambitious goals for mitigating climate change agreed upon in the Paris Agreement
in 2015, the European Union has adopted a long-term strategy to become climate-neutral by 2050. Part
of this strategy is the 2030 climate target plan, the outlines if which were defined in a communication
released in September 2020 (European Commission, 2020d): In relation to earlier attempts envisioning
a 40% cut in GHG emissions compared to 1990 levels by 2030, the EU has raised its ambitions to a
reduction of 55%. As an integral part of this strategy to lower emission levels by promoting renewable
25
CHAPTER 3. THE EU ETS 3.4. THE FUTURE OF THE EU ETS
energy and energy efficiency, the EU ETS is subject to major changes during the fourth trading period
from 2021 to 2030. In the following passage, I summarize the most significant adaptations to the ETS,
some of which have already been discussed in other chapters, in a comprehensive manner:
1. The annual reduction rate of the cap on emissions is going to be increased from the current value
of 1.74% to 2.2% starting with 2021.
2. Free allocation is going to be phased out for certain sectors from an initial 30% 2026 to 0% in 2030.
However, the current carbon leakage policy aimed at protecting vulnerable industries is going to be
maintained throughout phase IV, however under updated guidelines.
3. The market stability reserve, which was put into effect in 2019, is going to undergo further devel-
opment.
• From 2019 to 2023, the number of allowances withheld in the MSR will be doubled to 24% of
the allowances in circulation.
• Starting with 2023, the number of allowances contained in the MSR will be restricted to the
number of allowances auctioned during the previous period, whereas any surplus is automati-
cally invalidated.
• 200 million allowances from the MSR will be reserved for new entrants.
4. Two funds providing subsidies for innovation and modernisation in the power sector and other
energy-intensive industries will be created:
• The modernisation fund will aim at investments in energy efficiency by companies in the power
sector, with a special focus on low GDP member states.
• The innovation fund will succeed the existing NER 300 programme funding renewable energy
and CCS (Carbon Capture and Storage) technology. During phase IV, it is going to provide
the market value of 450 million allowances, which equals a 50% increase compared to its
predecessor (European Commission, 2020h).
26
Chapter 4
Data
4.1 Data Sources
The empirical part of this thesis relies on several independent data sources, not all of which are centered
around the EUTL. Whereas its primary goal is to gain new insight from transaction and account data,
it is nevertheless necessary to make use of auxiliary databases in order to draw a holistic image of the
EU ETS and establish a link between the inside and the outside perspective. In the course of this
section, I discuss the data sources involved in my analyses, detailing on their challenges, potential, and
downsides. Starting with the EUTL database as the most integral component of my research, I discuss
the specifics of aggregated data compiled by the European Environment Agency and end by expanding
on three complementary data sources which are vital in exploring core aspects of the ETS.
4.1.1 The EUTL
According to the ETS handbook (European Commission, 2015b, pp.76), the EUTL or European Union
Transaction Log is a system which "automatically checks, records, and authorises all transactions that take
place between accounts in the Union registry". Established as a successor to the Community Independent
Transaction Log or CIL, which had been in use during phase I and II, it was installed in 2013 and
contains data for all transactions since 2005. Via the publicly available EUTL website, a plethora of
datasets related to the ETS can be accessed, only a fraction of which bears relevance for the research
questions addressed in this thesis. In addition, it provides information on the emission targets specified
by the Effort Sharing Decision as well as on the compliance status and annual balance of each member
state from 2013 to 2020. In connection to the actual ETS, data on the following aspects can be accessed:
• Allowance allocation for phase I, II and III ranging from 2005 to 2020 on a national level including
CHAPTER 4. DATA 4.1. DATA SOURCES
both stationary installations and aircraft operators.
• Allowance allocation, verified emissions, surrendered units and compliance status as well as infor-
mation on the activity type and the holder of each installation covered by the ETS, however limited
to installations active in phase III.
• Data on international credit entitlement for CERs and ERUs originating from CDM or JI projects
on an installation level.
• All transactions within the EU ETS from 2005 to 2017, published with a delay of three years.
In the context of this thesis, I employ four of the datasets available providing information on an installation
level in analogy to research performed by Cludius (2016b)&(2016a):
1. The operator holding accounts or OHA dataset, which is a registry of all installations in the EU
ETS, containing data on operators’ sector, allocated and surrendered allowances, verified emissions
and compliance status. This dataset currently comprises about 16,972 accounts from 31 countries
and is limited to phase III of the EU ETS. Intallations from national registries established since
the beginning of phase I, in turn, are listed in a separate dataset. However, detailed information
on the latter is not publicly available on the EUTL website, so that the OHA dataset remains the
only useful data source.
2. The person holding accounts (PHA) as well as the trading accounts dataset containing information
on all registered accounts which are not related to a physical installation. These are held not
only by banks, brokers or energy trading companies, but also by installation operators using a
separate account for their trading activities. Whereas there are many similarities between the two
account types, they differ in terms of flexibility. Like OHAs, person holding accounts are limited to
interactions with so-called trusted accounts, whereas trading accounts offer unlimited access to the
market. This means that OHAs and PHas are required to submit a list of potential trading partners
to the national registry prior to issuing transactions. However, these distinctions are soon going
to be obsolete, for in accordance with Art. 84 of Regulation 2019/1122 (European Commission,
2019b), all PHAs are to be converted to trading accounts in 2021.
3. The transfer or transaction dataset, which records relevant data of all physical transactions per-
formed within the EU ETS, including information on the parties involved, the transaction date
and time, the transaction volume as well as on the transaction type. This includes international
certificates issued in accordance with the Kyoto protocol such as CERs and ERUs. With regard to
transaction types, the EUTL distinguishes between 9 different categories, each of which is assigned
a numeric identifier:
28
CHAPTER 4. DATA 4.1. DATA SOURCES
• 1: Issuance or the initial creation of a unit
• 2: Conversion or the transformation of a unit to create an ERU
• 3: External transfer of a unit between national registries, including non-EU countries
• 4: Cancellation or the internal transfer to a cancellation account
• 5: Retirement
• 6: Replacement
• 7: Carry-Over
• 8: Expiry date change
• 10: Internal transfer of a unit between operators
However, three of said transaction types — 6,7 and 8 — have not been employed to a relevant extent
since the creation of the EU ETS and may thus be considered as irrelevant for our analysis. The issuance
(1), the internal transfer (10) and the external transfer (3) of allowances, however, represent the most
commonly used transactions in the EU ETS.
In order to distinguish between account types, the transfer dataset uses three-digit codes referring
to different groups of account holders: According to the the ETS Registry system user guide (European
Commission, 2018d, pp.31), both operator holding accounts, aircraft operator holding accounts, person
holding accounts, trading accounts or accounts from external trading platforms are assigned 100, whereas
121 is reserved for PHAs in national registries which are limited to CERs or ERUs stemming from projects
under the Kyoto protocol. A statistical analysis of the transaction dataset reveals that, for transactions
completed after the 31st of December 2012, 94.4% of all transferring accounts and 95.8% of all acquiring
accounts belonged to the category 100, whereas only 1.4% and 2.5% are attributed to 121. Whereas on
the transferring side, only 4 different types – 100, 110, 120,121 – can be identified, the list of acquiring
accounts contains several types not mentioned in the manual – 210, 230, 250 and 300 – which represent
allowance deletion accounts of minor relevance. Accounts which have been transferred from national
registries until the beginning of phase III, follow a different naming scheme – while PHAs from this
period are denominated as 121, operator holding accounts carry the identification 120. Administrative
accounts, in turn, were already labeled 100 in phase I&II.
In addition to the OHA dataset, the EUTL website provides a plethora of account lists, the majority
of which provide only limited insight: In total, 51 datasets are available in the Accounts menu, ranging
from trading accounts to credit exchange accounts. However, in most cases, said datasets lack crucial
information such as account identifiers, which renders them useless in the context of data manipulation.
For instance, both the person holding accounts as well as the trading accounts list would prove excep-
tionally useful for categorizing individual transactions from the transaction log, if they followed the same
29
CHAPTER 4. DATA 4.1. DATA SOURCES
structure as the OHA dataset. Since the information available is limited to the account holders, however,
meaning that the only insight to be gained is whether or not a specific company owns PHAs or trading
accounts, I see no use in including these additional datasets in my analysis. Neither the EUTL website
nor literature give any indication as to why this crucial information has been omitted or if there are legal
concerns which might have prevented its publication.
4.1.2 Aggregated Data compiled by the European Environment Association
(EEA)
Second, I rely on aggregated data provided by the European Environment Agency to assess the devel-
opment of free allocation and reported emissions from 2005 to 2019. Available as a comprehensive CSV
file which is freely available online, the EEA dataset contains industry-level data for each member state,
meaning that both the actual emissions and the number of freely allocated allowances for each of the
activities defined by the ETS Directive can be monitored for all trading periods. Hence, data can be
aggregated by both industry, nation and year, enabling comprehensive analyses of allowance allocation
and emission levels over time. In terms of scope, the EEA dataset covers 29 industry sectors or activity
types, including two additional categories offering total values with and without Combustion of Fuels. In
absolute numbers, it contains 57,744 lines of data on all 31 participating countries. Given the complexity
of EUTL database and its issues with data quality, I employ the EEA dataset for areas which do not
require an installation- or transaction-level analysis. In addition, assuming that all data published by
the EEA has undergone a thorough verification process, it serves as a benchmark to gauge the integrity
of the EUTL database.
4.1.3 Auxiliary Datasets
The broad spectrum of subjects covered by my thesis makes it necessary to include additional datasets
which are independent from the EUTL. This involves indicators of economic performance, allowance price
data as well as data on allowance auctions and NACE codes, which I obtain from four independent sources:
First, I employ gross and per-capita GDP data from EUROSTAT for all nations involved with emissions
trading from 2005 to 2019. Second, I use historic spot price data for EU allowances obtained from Ember
Foundation (2020), which is available from 2008 to 2020 in daily intervals. Third, both market places
managing the auctioning of EU allowances – ICE London and EEX Leipzig – offer extensive data on
auction volumes, auction prices and the number of bids for auctions held during phase III. Finally, a
dataset by Jaraite et al. (2013) enables me to link physical installations to corresponding NACE codes.
30
CHAPTER 4. DATA 4.2. DATA PREPARATION
Figure 4.1: The EUTL Web Interface.
4.2 Data Preparation
Whereas both aggregated data from EEA and the complementary datasets require minimum effort for
preparation, the EUTL is less accessible, necessitating a more complex and time consuming process to
extract data. Given the fact that the EUTL web portal does not provide a viable means of exporting
large amounts of data — currently, a restriction of 3,000 lines per download applies — I have used
the commercial software Octoparse to extract HTML tables directly from all search results pages for a
specific query. This process, which is commonly referred to as Webscraping, allowed me to download
and export both the entire transaction log and the OHA dataset at a rate of approximately 400 lines
of data per minute, resulting in a set of CSV files which I subsequently merged and consolidated using
Microsoft Excel. While the extraction of the transaction dataset containing about 990.000 entries from
2005 to 2017 turned out as relatively straightforward, the OHA dataset required me to link separate
tables by programming the software to automatically perform several simulated clicks for each line of
data. Once the CSV files had been compiled, I imported both datasets to SPSS in order to prepare
them for the following statistical analysis. The main objective of this process being to establish a link
between the account identifiers found in the transaction log and their respective counterparts in the OHA
list, I initially attempted to automatically import the variable Main Activity Type into the transaction
dataset using the MERGE command. However, in accordance with (Cludius, 2016a, pp.9), I found that
only a perceived 70-80% of all transactions involving physical installations can be automatically linked
31
CHAPTER 4. DATA 4.2. DATA PREPARATION
to a corresponding activity type, meaning that it is necessary to manually check and correct the output
in order to yield accurate results. Common inconsistencies, which prevent the software from matching
identical account identifiers, include both differences in name as well as mere spelling mistakes.
Taking into account these complications, which necessitate extensive manual adjustments to the
dataset, establishing a link between account and transaction data for all 994,280 entries is a prohibitively
time-consuming task. Hence, limiting the scope to a reduced subset seems a viable option, which is why
I select all transactions involving accounts in the Austrian registry either on the acquiring or on the
transferring side, reducing the size of the dataset to about 21,600 entries, which equals 2.2% of the initial
volume. This extends to account data as well, limiting the number of installations to be linked to 296.
In addition to facilitating the aforementioned process, narrowing the focus enables me to perform a more
in-depth analysis of the sectoral distribution of installations.
Considering that an account’s activity type refers solely to the process by which greenhouse gases
are emitted and often fails to give an indication of the respective installation’s actual industry sector, I
deem it necessary to introduce a different classification scheme. Hence, in order to translate the activity
types specified in ANNEX 1 to the ETS Directive (European Commission, 2009a) to the more universal
NACE v.2 structure, I employ a dataset compiled by Jaraite et al. (2013) with the intention of linking
individual accounts with their parent companies. This dataset, the scope of which is limited to phase I
of the ETS, is publicly available as a XLS file and contains not only the account information from the
OHA list to which it can be matched, but also NACE v.2 codes and information on the current and
past global ultimate owner or GUO – a term, which is used by Bureau van Dijk’s company database
ORBIS to identify subsidiaries of multinational corporations. By means of the MERGE command, I was
able to automatically add NACE data to the OHA dataset using Installation Name as the key variable.
Given that the dataset compiled by Jaraite et al. is based on historical data which does not perfectly
correspond to the most recent version of the OHA list, several manual adjustments were required to
assign each installation the correct industry sector. In order to fill the blanks, I employed the German
website www.firmenwissen.com, which provides NACE codes apart from general company information.
Subsequently, I manually compiled a SPSS dataset matching NACE v.2 codes to written descriptions
based on the EUROSTAT NACE guide (Eurostat, 2008), which I used to assign each installation a label
in plain text. Finally, I automatically matched the NACE v.2 codes to all transactions involving Austian
installations using both the acquiring account identifier and the transferring account identifier as key
variables.
32
CHAPTER 4. DATA 4.2. DATA PREPARATION
4.2.1 Coping with the Limitations of the EUTL
Despite the unmistakable value of the EUTL as a repository of all transactions within the EU ETS,
its publicly available database suffers from several flaws and limitations which impair the quality and
validity of the data contained. One of the first steps in writing this thesis being several case studies
on individual installations, the purpose of which has been to gain a thorough understanding of the
underlying mechanisms of emissions trading, I was able to identify a variety of shortfalls which limit the
EUTL dataset’s usability as a research subject. This section intends to discuss these issues in detail.
Starting with transaction data, the core problem I had to face lies in its incompleteness – in total,
101,858 or 10,2% of 994,280 transactions recorded from January 2005 to April 2017, all of which were
performed between accounts of the same registry, lack information on at least one of the parties involved.
Fig. 4.2 gives an impression of the magnitude of this effect. In absolute numbers, 15.5% of transactions
completed during phase I&II exhibit missing values, which translates to 35.5% of the total transaction
volume. Of all insufficiently labeled transactions, 34.0% alone were issued by UK accounts, whereas
another 18.9% originated from Italy. The regional distribution of the remaining transactions, however, is
more in line with the average transaction volumes of each member state during phase I&II. For another
4,568 transactions, information on both the acquiring and the transferring accounts is missing. Interest-
ingly, the majority of these – 62.1% and 26.8% – originate from Austrian and Greek accounts. Further
3.5% were transferred from EU accounts, whereas the remaining national registries play only a minor role.
However, it is worth mentioning that all of said gaps in the dataset are limited to dates ranging from 2005
to 2012. The causes of these irregularities are subject to speculation – neither literature nor the EUTL
website give a clear indication as to why such a large proportion of the dataset is incomplete. Hence,
it remains unclear wether the loss of data has occurred during the transition process from a national
administrative structure to the current EUTL or if the data collected by the national agencies had been
incomplete in the first place. Apart from these corrupted entries, there are another 35,400 transactions
involving accounts outside the EU ETS as well as CDM accounts, which also lack information on one of
the parties involved. However, this does not constitute an irregularity, since the EUTL keeps no records
of market participants not registered by the system.
A detailed investigation of the individual accounts of several Austrian Companies – Calcit GmbH,
Energie AG, Stölzle Oberglas GmbH, Voestalpine AG and Wienerberger AG – gives further proof of the
necessity to reduce the scope of my investigation to phase III of the EU ETS: Not only is the naming
scheme of individual accounts inconsistent across datasets, but there are also instances in which accounts
are transferred to new owners while maintaining their balance, making it virtually impossible to monitor
individual accounts over an extended period of time. With the support of Dr. Bettina Dallinger and
Wolfgang Strasser from Energie AG, I was able to resolve complications arising from the transition
EEX total bidsEEX auctionedICE total bidsICE auctioned
Figure 5.35: Both EEX and ICE exhibit a strong correlation between the demand for allowances and the
number of allowances auctioned per week.
71
CHAPTER 5. DISCUSSION AND RESULTS 5.5. AUSTRIAN ACCOUNTS
5.5 Narrowing the Focus: Observations based on Transactions
to and from Austrian Accounts
5.5.1 An Anatomy of the Austrian Emissions Market
Given the imperfections of the EUTL, linking account and transaction data is a time consuming and often
unrewarding task which cannot be easily automatized. Nevertheless, said link provides some valuable
insights on certain aspects of the ETS which are impossible to deduce from transaction data alone.
Hence, I perform a more detailed analysis of the EUTL with a limited scope, focusing on transactions
originating from as well as directed to Austrian accounts. This enables me to shed light on aspects of
emissions trading left out or only partially explored in previous sections. For instance, I am able to expand
on the distribution of account types using transaction data, complementing the information provided in
section 5.2. Furthermore, narrowing the focus enables me to translate an account’s activity type to the
NACE code of the account holder, making it possible to identify the distribution of industry sectors in
the ETS as well as the development of trading volumes for each industry sector. Finally, I determine
the share of transactions to and from foreign accounts, investigating which countries Austrian firms have
been interacting with during phase III.
Year
20172016201520142013
num
ber o
f acc
ount
s ac
tive
per y
ear
300
200
100
0
119
296
OthersOHA
Figure 5.36: The number of both physical installations, PHAs and trading accounts, which have been
active at least once in a given year, is substantially lower than the theoretical maximum derived from
account data.
72
CHAPTER 5. DISCUSSION AND RESULTS 5.5. AUSTRIAN ACCOUNTS
Since the EU’s member states are extremely heterogeneous in terms of population size and economic
performance, I first establish, how the Austrian market is structured in relation to the EU average.
In terms of registered accounts, the OHA dataset currently lists 296 physical installations in Austria,
only part of which have been constantly active during phase III. In comparison to the EU average, the
Austrian registry holds a significantly lower number of PHAs and trading accounts: Whereas 65.1% of
26,079 accounts registered across all 31 countries participating in the ETS are OHAs, the share of physical
installations rises to 71.3% when limiting the scope to Austria. As to the remaining account types, the
equivalent numbers are 27.4% for PHAs and 7.5% for trading accounts in the European perspective. Of
415 account identifiers listed by the Austrian registry, in turn, only 20.2% are identified as PHAs and
further 8.4% as trading accounts. In terms of installation density, Austria scores below average, reaching
only 33.4 installations per million inhabitants as opposed to the EU average of 59.9.
Andere
Industrial plants for the production of (a) pulp from
timber or other fibrous m
aterials (b) paper and board
Installations for the production of cement
clinker in rotary kilns or lime in rotary kilns or in
other furnaces
Production of pulp
Production or processing of ferrous metals
Production of bulk chemicals
Production of pig iron or steel
Other activity opted-in pursuant to Article 24
of Directive 2003/87/EC
Manufacture of glass
Production of cement clinker
Installations for the manufacture of ceram
ic products by firing, in particular roofing tiles, bricks, refractory bricks, tiles, stonew
are or porcelain
Production of lime, or calcination of
dolomite/m
agnesite
Production of paper or cardboard
Manufacture of ceram
ics
Aircraft operator activities
Com
bustion installations with a rated therm
al input exceeding 20 M
W
Com
bustion of fuels
40.00
30.00
20.00
10.00
0.00
ATEU
sha
re in
%
Figure 5.37: Distribution of registered installations in Austria across activity types compared to the EU.
Backing the theoretical numbers with EUTL data, fig. 5.36 illustrates the results of a transaction-
level analysis monitoring the number of PHAs and OHAs which were involved in at least one transaction
in a given year. The considerable discrepancy observed between the number of registered accounts and
the empirical results from transaction data indicates that not all installations have been entitled to
free allocation or surrendered emission allowances in every period. Whereas I am unable to provide an
explanation for this phenomenon based on empirical data, it is probable that a considerable number of
installations have been put into operation, sold, modernized or closed down during phase III, resulting
73
CHAPTER 5. DISCUSSION AND RESULTS 5.5. AUSTRIAN ACCOUNTS
in periods of inactivity. Due to the absence of compliance obligations, this behavior seems even more
plausible for PHAs or trading accounts. Evidently, both account types exhibit a significant downwards
trend during phase III. In concrete terms, the number of active installations has decreased from 220, which
equals 74% of the theoretical maximum, to 188 or 63.5% from 2013 to 2017. PHAs and trading accounts,
on the other hand, exhibit even lower numbers, ranging from 38 accounts or 32% of the theoretical
maximum in 2013 to a peak of 46 or 38.7% in 2014 and the all-time low of 27 or 22.7% in 2017.
Production of amm
onia
Production of secondary aluminium
Production of nitric acid
Production of coke
Refining of m
ineral oil
Production or processing of gypsum or
plasterboard
Production or processing of non-ferrous metals
Industrial plants for the production of pulp from
timber or other fibrous m
aterials / paper
Installations for the manufacture of glass including
glass fibre
Mineral oil refineries
Other activity opted-in pursuant to Article 24 of
Directive 2003/87/EC
Production of bulk chemicals
Installations for the manufacture of ceram
ic products by firing, in particular roofing tiles, bricks,
Production or processing of ferrous metals
Installations for the production of cement clinker or
lime in rotary kilns
Manufacture of glass
Production of pulp
Production of pig iron or steel
Production of cement clinker
Production of lime, or calcination of
dolomite/m
agnesite
Production of paper or cardboard
Manufacture of ceram
ics
Aircraft operator activities
Com
bustion installations with a rated therm
al input exceeding 20 M
W
Com
bustion of fuels
0
100.00
80.00
60.00
40.00
20.00
0.00
Figure 5.38: Distribution of registered installations in Austria across activity types.
Fig. 5.37 and fig. 5.38 give an indication of the distribution of activity types and industry sectors
across Austrian installations. Due to the limited size of the market, only 23 of 38 possible categories
are actually represented in the Austrian registry. In comparison with EU data, Austria exhibits a lower
concentration of activity types, with combustion of fuels amounting to only 33.8% versus 43.0% of all
installations. Furthermore, 81.8% as opposed to 84.6% of installations belong to one of the eight largest
categories, whereas 30.4% versus 60.5% of activity types do not exceed the 1% mark. With regard to
the Hirschmann-Herfindahl index, Austria scores 1,599 compared to 2,249.7, despite the lower number of
categories in use.
In order to provide an alternative to the category scheme used by the EUTL, which differentiates
installations or sub-installations based on the way GHGs are emitted rather than by focusing on the
installation operators’ industry sectors, I establish a link between each individual account and the parent
company’s NACE code. However, contrary to my initial assumption, this process, which requires extensive
74
CHAPTER 5. DISCUSSION AND RESULTS 5.5. AUSTRIAN ACCOUNTS
manual adjustments and can thus not be extended to the whole transaction dataset, fails to create a more
even distribution. As fig. 5.39 indicates, the largest category – electricity, gas, steam and air conditioning
supply – comprises a similar share of Austrian installations as combustion of fuels (29.7% vs. 33.8%). In
addition, the HHI rises from 1,599 to 1,729, which, however, is still below the EU average. On the other
end of the spectrum, 8 or 38% of 21 NACE codes in the dataset score below the 1% mark, whereas the
8 largest categories make up 91.6% of all installations. However, the dominance of certain sectors can
be explained by both the limited size of the Austrian market and the country’s low supply of natural
resources, challenging the significance of direct comparisons with other countries or the EU as a whole.
NACE_Branche_l1
Herstellung von D
atenverarbeitungsgeräten, elektronischen und optischen Erzeugnissen
Samm
lung, Behandlung und Beseitigung von Abfällen; R
ückgewinnung
Herstellung von M
etallerzeugnissen
Sonstiger Fahrzeugbau
Getränkeherstellung
Gew
innung von Steinen und Erden, sonstiger Bergbau
Herstellung von G
umm
i-und Kunststoffwaren
Herstellung von pharm
azeutischen Erzeugnissen
Gew
innung von Erdöl und Erdgas
Herstellung von Kraftw
agen und Kraftwagenteilen
Herstellung von Textilien
Kokerei und Mineralölverarbeitung
Herstellung von N
ahrungs-und Futtermitteln
Herstellung von H
olz-, Flecht-, Korb-und Korkw
aren (ohne Möbel)
Herstellung von chem
ischen Erzeugnissen
Metallerzeugung und -bearbeitung
Herstellung von Papier, Pappe und W
aren daraus
Luftfahrt
Herstellung von G
las und Glasw
aren, Keramik,
Verarbeitung von Steinen und Erden
Energieversorgung
0
100.00
80.00
60.00
40.00
20.00
0.00
Figure 5.39: Distribution of registered installations in Austria across NACE codes.
5.5.2 The Sectoral Distribution of Market Activity
Whereas valuable insight on the distribution of activity types or industry sectors can be derived from
account data, transaction data is required to investigate the development of both transaction volumes,
transaction numbers and transaction sizes during phase III. As fig. 5.40 reveals, the distribution of
industry sectors in the OHA dataset does not match the actual transaction volumes derived from EUTL
data. Whereas a mere 7.1% of installations are related to the manufacture of basic metals, their share
of allowances traded amounts to 26.3%. On the other hand, electricity and gas supply, which, as the
dominant sector, covers 29.7% of installations, is responsible for only 20.9% of the overall transaction
volume. The same applies to manufacture of non-metallic, mineral products, which is associated with
75
CHAPTER 5. DISCUSSION AND RESULTS 5.5. AUSTRIAN ACCOUNTS
only 10.5% of allowances traded while covering 23.0% of installations. The annual number of transactions
represented by fig. 5.41, in turn, is more in line with account data. Of all transactions issued to and
from Austrian accounts between 2013 and 2016, 35.2% can be attributed to electricity and gas supply,
further 20.8% to manufacture of non-metallic, mineral products, and finally, 8.7% to manufacture of
basic metals. Said results are also apparent in fig. 5.42, which illustrates the average transaction sizes
aggregated by NACE code. From this perspective, manufacture of coke and refined petroleum products as
well as manufacture of basic metals exhibit exceptionally high values, suggesting that company sizes in
these industries are comparably large. In the case of Voestalpine AG and its subsidiaries, this statement
can undoubtedly be confirmed.
The significance of these observations, however, is impaired by both the low sample size of only
296 installations operated by 171 legal entities and the heterogeneous nature of the Austrian market.
As the OHA dataset reveals, the Austrian registry is characterized by a considerable number of SMEs
with relatively low annual turnover and trading activity, which is in stark contrast to global players like
Voestalpine, OMV or Wienerberger. Whereas this may explain why certain industry sectors featuring
low account numbers are overrepresented in terms of transaction volume, it still remains unclear whether
variables such as annual turnover or the number of employees have a causal effect on a company’s trading
activity. A thorough analysis of this relationship, however, would require the use of additional data
sources exceeding the scope of my thesis.
2016201520142013
mill
ion
allo
wan
ces
(tonn
es o
f CO
2 equ
ival
ent)
40.00
30.00
20.00
10.00
0.00
Air TransportManufacture of Paper and Paper ProductsManufacture of Chemicals and Chemical ProductsAverageManufacture of Coke and Refined Petroleum ProductsManufacture of Non-Metallic, Mineral ProductsElectricity and Gas SupplyManufacture of Basic Metals
Figure 5.40: Total volume of transactions involving Austrian accounts by NACE code.
76
CHAPTER 5. DISCUSSION AND RESULTS 5.5. AUSTRIAN ACCOUNTS
2016201520142013
num
ber o
f tra
nsac
tions
600
500
400
300
200
100
0
Manufacture of Food ProductsManufacture of Chemicals and Chemical ProductsAir TransportAverageManufacture of Paper and Paper ProductsManufacture of Basic MetalsManufacture of Non-Metallic Mineral ProductsElectricity and Gas Supply
Figure 5.41: Total number of transactions involving Austrian accounts by NACE code.
2016201520142013
allo
wan
ces
per t
rans
actio
n
1,000,000
800,000
600,000
400,000
200,000
0
Manufacture of Paper and Paper ProductsOther Mining and Quarrying Manufacture of Non-Metallic Mineral ProductsElectricity and gas supplyManufacture of Chemicals and Chemical ProductsAverageManufacture of Basic MetalsManufacture of Coke and Refined Petroleum Products
Figure 5.42: Average number of allowances per transaction involving Austrian accounts by NACE code.
77
CHAPTER 5. DISCUSSION AND RESULTS 5.5. AUSTRIAN ACCOUNTS
5.5.3 National and International Transactions
2016201520142013
mill
ion
allo
wan
ces
(tonn
es o
f CO
2 equ
ival
ent) 50.00
40.00
30.00
20.00
10.00
0.00
NationalInternational
Figure 5.43: Volume of national versus international transactions involving Austrian accounts. A trans-
action is identified as international if the transferring registry and the acquiring registry are not identical,
provided that the transaction is not administrative.
Apart from generating data on the distribution of industry sectors on the basis of NACE codes,
the focus on Austrian accounts provides valuable insight on another relevant aspect: By analyzing the
acquiring as well as the transferring registries of transactions in the Austrian dataset, I am able to
compare the share of transfers between domestic companies to those originating from or directed to
foreign accounts. In this context, I restrict my analysis to market transactions, meaning that both
administrative transactions involving the national registry accounts as well as intra-company transfers are
ignored. Relative to the total number of 8,007 transactions involving Austrian accounts recorded by the
EUTL from 2013 to 2016, the former category makes up 51%, which is on par with the European average
of 48.3%. In terms of transaction volumes, however, only 24.1% as opposed to 56.3% are attributed to
market transactions, which can be explained by a considerable number of internal transfers issued by the
Austrian Emissionshandelsstelle. After eliminating 15 of said transactions with volumes ranging from
1.25 to 41 million allowances, the share of market transactions rises to a more plausible 38.0%. As fig.
5.44 indicates, the numbers of both national and international transactions have been in decline since
2013. Whereas national transactions, have lost 76.1% during phase III, this value amounts to 56.8% for
78
CHAPTER 5. DISCUSSION AND RESULTS 5.5. AUSTRIAN ACCOUNTS
transactions involving foreign accounts.
As to the phase III average, the number of international transactions exceeds that of national transac-
tions by 93.1%. This substantial gap is also evident from fig. 5.44, amounting to an even greater 110.5%.
Concerning the development of transaction volumes, however, the downwards trend observed from 2013
to 2016 is less significant than in terms of transaction numbers, amounting to –21.9% for international
and –30.9% for national transactions. In accordance with these numbers, fig. 5.45 indicates a steep up-
wards movement in terms of transaction sizes, which amounts to +80.9% for international and +189.4%
for national transactions. The phase III average, in turn, is nearly identical for both variables, deviating
by only 3.9%.
2016201520142013
num
ber o
f tra
nsac
tions
600
400
200
0
NationalInternational
Figure 5.44: Number of national versus international transactions involving Austrian accounts.
As a final step, I differentiate transactions by their transferring and acquiring registries, enabling me
to determine, which countries Austrian companies are predominantly involved with in terms of emissions
trading. For this purpose, I investigate both transaction volumes and numbers for transactions originating
from and directed to Austrian accounts. In absolute terms, 1,704 international transactions were issued
between 2013 and 2016, 1,079 or 63.3% of which exhibit Austrian recipients. As to the national registries
involved, the transaction dataset lists 22 countries on the transferring as well as 19 on the receiving end.
However, only a small number of countries exhibit substantial volumes in all four trading periods observed:
Among transactions originating from Austrian accounts, the UK yields the highest average annual volume
with 4.34 million allowances, closely followed by Germany with 3.94 million allowances. These two
79
CHAPTER 5. DISCUSSION AND RESULTS 5.5. AUSTRIAN ACCOUNTS
2016201520142013
allo
wan
ces
per t
rans
actio
n150,000
100,000
50,000
0
NationalInternational
Figure 5.45: Number of allowances per transaction involving Austrian accounts.
outperform other nations such as Romania (1.07 million), France (0.5 million) and the Netherlands (0.33
million) by a considerable margin. On the receiving end, Germany reaches 8.14 million allowances p.a.,
followed by the UK with 2.96 million, France with 1.83 million, Romania with 1.46 million and the
Netherlands with 0.93 million. Regarding the dominance of Germany and the UK, the EUTL reveals
that a substantial share of transactions originating from these countries involve banks or brokers. This
may also entail the acquisition of allowances through auctions, in which no Austrian installation operators
have partaken so far. However, further research is needed to back this observation, which is based on an
exemplary analysis of a small sample, with concrete data.
The same reasoning applies to the development of transaction numbers and transaction volumes
illustrated in fig. 5.46 to fig. 5.46. Whereas the sharp decline in transactions involving CDM or Kyoto
credits at the beginning of phase III can be explained by legal restrictions put into effect in 2013, it is
difficult to formulate a coherent hypothesis as to the volatility of transaction volumes to and from foreign
accounts. Given the low number of international transactions in combination with the limited share of
Austrian accounts interacting with foreign trading partners, however, an in-depth analysis of this subject
should not yield significant results.
80
CHAPTER 5. DISCUSSION AND RESULTS 5.5. AUSTRIAN ACCOUNTS
2016201520142013
mill
ion
allo
wan
ces
(tonn
es o
f CO
2 equ
ival
ent)
10.00
8.00
6.00
4.00
2.00
0.00
Netherlands France Romania Germany United Kingdom
Figure 5.46: Annual volume of transnational transactions originating from Austrian accounts, aggregated
by nation.
2016201520142013
num
ber o
f tra
nsac
tions
80
60
40
20
0
Netherlands France United Kingdom Romania Germany
Figure 5.47: Annual number of transnational transactions originating from Austrian accounts, aggregated
by nation.
81
CHAPTER 5. DISCUSSION AND RESULTS 5.5. AUSTRIAN ACCOUNTS
201620152014201320122011201020092008
mill
ion
allo
wan
ces
(tonn
es o
f CO
2 equ
ival
ent) 12.00
10.00
8.00
6.00
4.00
2.00
0.00
Netherlands Romania France CDM United Kingdom Germany
Figure 5.48: Annual volume of transnational transactions received by Austrian accounts, aggregated by
nation.
201620152014201320122011201020092008
num
ber o
f tra
nsac
tions
200
150
100
50
0
Romania Netherlands France CDM United Kingdom Germany
Figure 5.49: Annual number of transnational transactions received by Austrian accounts, aggregated by
nation.
82
Chapter 6
Conclusion
The European Union’s emissions trading system has been in operation since 15 years and is on the verge
of entering into its fourth evolutionary phase. Over time, it has undergone substantial changes aimed
at adapting the system to the development of the market. This not only entails the centralization effort
resulting in the creation of a unified registry by the beginning of phase III, but also affects the way
allowances are allocated. In my thesis, I aim at shedding light on the underlying mechanisms of this
development by making use of a data source which has rarely been used by researchers to this date.
In the course of the establishment of the ETS registry, the EU has created a publicly available database
encompassing both account and transaction data for all trading periods from 2005 onwards with a delay
of three years. Whereas in theory, this comprehensive source of data grants a tremendous opportunity to
perform research on various aspects of the EU ETS and its participants on a transaction level, my personal
experience proves that there are still challenges to be overcome in order to realize the European Union
Transaction Log’s full potential. This not only entails questions of usability, but also encompasses certain
issues with data quality, which have not yet been alleviated. To begin with, I criticize the limitations of
the EUTL website, which, in addition to its unstructured design, lacks a useful export feature for both
account and transaction data. The existing implementation, in turn, restricts the number of data points
per download, necessitating the use of web scraping software in order to compile large datasets. Whereas
said shortfall constitutes only a minor hindrance to the ambitious researcher, deficiencies concerning
data quality are harder to overcome. As my research indicates, a considerable proportion of transactions
issued prior to 2013 exhibit missing account identifiers either on the transferring, the acquiring or on
both sides. The fact that these incompletely labeled transactions amount to 15.5% of the total number
and 35.5% of the total trading volume from 2005 to 2012, renders certain analyses based on transaction
types impracticable. Provided that these inconsistencies stem from the transition to the centralized
registry at the end of phase II, it is doubtful whether substantial improvements are to be expected in the
CHAPTER 6. CONCLUSION
future. Another flaw in the transaction dataset relates to the conversion of emission allowances under
the Kyoto protocol to EU allowances, which resulted in a considerable number of duplicate transactions
in the dataset. These, however, cumulate on a certain date and can thus be eliminated if necessary.
Overall, said insufficiencies of the database have not occurred since 2014, suggesting that data quality
has already seen major improvements during phase III and is going to improve further during phase
IV. Hence, I am positive that in future, researchers working with the EUTL database will be able to
focus more on its content rather than on coping with its deficiencies. Finally, one of the most significant
drawbacks of the EUTL, which should definitely be addressed by the European Commission, is the missing
integration of account and transaction data. Currently, neither the account identifiers nor the account
holders found in the transaction dataset are identical to those in the OHA dataset. Both the PHA and
the trading accounts dataset, in turn, don’t even name account identifiers, making it virtually impossible
to distinguish between these types on a transaction level. By eradicating this shortfall, the European
Commission would enable researchers to conduct a broad range of analyses without having to manually
link transaction and account data. This involves distinguishing between OHAs, PHAs and Trading
Accounts as well as matching installations with corresponding activity types or NACE codes. Ideally,
each individual account should be assigned a unique alphanumeric identifier containing information on
national registry, type and account holder in order to facilitate the preparation and processing of data.
Compared to the emissions trading system as a whole, however, evaluating the EUTL and its challenges
is an infinitely less complex subject. In fact, the development of the European Union’s ETS during the
first 15 years of its existence can be analyzed from a multitude of perspectives. For this purpose, I
employ several independent sources of data, ranging from an official emissions dataset compiled by the
European Environment Agency to GDP statistics by EUROSTAT. The main objective of my thesis being
to provide new insight into emissions trading using EUTL data, I predominantly base my conclusions on
the European Union Transaction Log’s publicy available database, striving to explore aspects of the ETS
which have not yet been covered in literature.
Starting with general data on the scope and size of the European Emissions Trading System, the
EEA’s aggregated dataset makes it possible to trace the development of emission levels and allowance
allocation over time. As my analysis reveals, the supply of allowances has been exceeding the number of
surrendered units from 2009 to 2013, leading to the accumulation of a substantial surplus lowering the
allowance price by more than 80% compared to its initial value. In an attempt to alleviate this systemic
flaw, the European Commission cut the supply of allowances to be auctioned during phase III, eventually
implementing an instrument called the Market Stability Reserve. The MSR, which manages allocation
based on the number of allowances in circulation, will have to prove its effectiveness in stabilizing the
allowance price during phase IV. In recent publications, this subject has been discussed controversially,
with debate centered mainly around an aspect which is referred to as the waterbed effect – a situation, in
84
CHAPTER 6. CONCLUSION
which complementary environmental policies mitigate the overall GHG reduction achieved by the ETS.
Hence, government subsidies targeting CO2-efficiency in a certain sector are expected to grant beneficiaries
of such policies a surplus of allowances, leading to a decrease in prices, which, in turn, disincentivices
emissions abatement (Mulder, 2021). Following this reasoning, I conclude that the overallocation of
allowances to certain industry sectors, which has remained a major factor throughout phase III, may
result in a similar effect, impairing overall abatement effciency while shifting allowances from subsidized
to non-subsidized industry sectors. As a side effect, this leads to companies from emission-intensive
industries such as steel or cement production realizing substantial windfall profits – a practice, which,
although justified by the European Commission as a means of preventing carbon leakage, contradicts the
fundamental principles of a cap-and-trade system. This is especially relevant since recent studies find no
evidence for the alleged exodus of companies due to the burden of the EU ETS.
According to Perino (2018), the cancellation of allowances via the MSR is going to temporarily reverse
the waterbed effect, whereas the impact of this reversal decreases the later emission abatement takes place.
Eventually, as soon as the number of banked allowances falls below the threshold, which may occur as soon
as 2023, the waterbed effect is expected to return. Rosendahl (2019), in turn, takes a more pessimistic
stance, arguing that even the prospect of future reduction policies complementary to the ETS may impair
emission abatement, since account holders anticipating these changes tend to bank less allowance, leading
to a lower number of cancelled allowances. Flachsland et al. (2020) make a case for the installation of a
price floor for EU allowances, which may not only serve as a complementary measure, but even replace
the existing MSR. In fact, an auction reserve price already exists, albeit with minor practical impact. In
the current auctioning regulation, the price floor is tied to the secondary market price for EU allowances,
so that in reality, the European Commission is unable to exert control over the market. In addition,
bid-to-cover ratios have been stable during phase III, so that only two auctions have been canceled for
this reason so far.
Continuing with the auctioning of allowances, a transaction-level analysis reveals that, despite the
relevance that this allocation method has gained by the beginning of phase III, only 122 different account
holders have received transactions from one of the two market places involved so far. Given the fact that 64
of these are either banks, brokers or trading companies, it is evident that installation operators, which, in
theory, represent the target audience of allowance auctions, hardly ever partake in the system. However,
despite being able to trace transaction flows from from the EU to the bidders’ accounts, exemplary
analyses of Citigroup and Deutsche Bank fail to give a clear indication as to whether PHAs involved in
auctioning have been acting as brokers on behalf of smaller installation operators unwilling or unable to
take part in allowance auctions. In order to explore this question and identify patterns in the dataset,
which go beyond identical transaction volumes appearing in different transactions, further research is
needed. This, however, requires the use of sophisticated algorithms exceeding the scope of this thesis.
85
CHAPTER 6. CONCLUSION
Comparing emission values between member countries, major national economies such as Germany or
the UK are dominant in absolute numbers. When adjusting for GDP, however, the ranking is inversed,
suggesting that less performant countries predominantly in Eastern Europe are affected by the ETS to
a higher degree than their economically advanced counterparts in Western Europe. In fact, I identify a
statistically significant correlation between a country’s per capita GDP and its adjusted emission levels,
which further corroborates this hypothesis. The same trend can be observed in terms of installation
density, meaning that poorer countries exhibit higher installation numbers per billion EUR of GDP.
This unequal distribution of the burden that the ETS imposes on national economies, has already been
addressed by granting exemptions and subsidies to the countries affected. Given the exceptional GDP
growth rates in countries like Bulgaria, however, I expect the vast economic disparities which still prevails
among the EU’s member countries to narrow in the long run.
With regard to the state and development of the emissions market, several significant insights can be
derived from both transaction and account data offered by the EUTL. In terms of annual transaction
volumes and transaction numbers, both variables have been in decline during phase III after reaching a
peak in 2008/2009, suggesting that the constant reduction of the cap on emissions has a negative effect
on trading activity. Differentiating by three transaction types – market transactions, intra-company
transfers and administrative transactions – I establish a category scheme which aids in understanding
the dynamics of the EU ETS. Relative to the total volume, market transactions make up about 43.5%,
compared to 47.2% for administrative transactions, indicating that administrative processes such as the
allocation and surrendering of allowances or transfers between national registries constitute a slightly
larger volume than actual emissions trading between accounts. Intra-company transfers or transactions
between accounts operated by the same company, in turn, are substantially less relevant both in terms
of transaction volume and transaction numbers, amounting to 9.3% of the total volume. However, my
analysis disregards the affiliation of registered account holders to large corporations, which may constitute
a source of error leading to unrealistically low numbers of intra-company transfers. This flaw in my model
may indeed pose a problem when evaluating market activity, since it is hard to judge whether transactions
between legal entities within a corporation serve the original purpose of emissions trading. Referring to a
strategy used by Jaraite et al. (2013) on phase I data, I recommend linking account holders to a company
database in order to resolve this issue.
Taking an in-depth look at administrative transactions, it becomes apparent that a subset of this
category responsible for only 0.65% of administrative transactions makes up 21.5% of the total transaction
volume between 2013 and 2016. Their low number being contrasted by exceptional transaction sizes, this
category can be identified as the main cause of spikes in the dataset, which, due to their sheer dimension,
have a distorting effect on the overall market activity. In order to identify and analyze these irregularities,
I shift perspective, calculating the average transaction volumes, transaction numbers and transaction sizes
86
CHAPTER 6. CONCLUSION
for each day of the year. On the one hand, this allows me to examine periodic events such as allowance
allocation and surrendering, while on the other hand, it enables me to isolate spikes both in terms of
transaction volumes and transaction numbers, laying the foundation for a more thorough analysis. Indeed,
several cases of abnormally high transaction volumes or numbers limited to short periods of time can be
identified in the dataset, all of which are attributed to registry-internal processes without relevance for
emissions trading. As a transaction-level analysis reveals, these irregularities can be attributed to singular
events such as the obligation to retire Kyoto credits from previous phases in 2015 or the cancellation of
expired EU allowances in 2013. With regard to recurring patterns in transaction data, I identify three
periods of increased activity related to the allocation of allowances until February 28 – the surrendering of
allowances until April 30 as well as to the delivery of EUA forwards and futures in December. Both these
periodic events and the spikes in transaction data can also be observed when aggregating transaction
volumes and numbers by month.
Aggregating transaction numbers by hour and weekday, in turn, adds another insightful perspective
to my analysis of emissions trading, revealing that transactions issued by regular market participants are
restricted to a narrow timeframe. Whereas market transactions are accepted by the system only within
office hours and require a minimum 26 hours until completion, administrative transactions such as the
allocation of allowances are automatically issued and thus independent of time or weekday. Accordingly,
while the former category is evenly distributed across the permitted timeframe, the latter exhibits a peak
around 1 am.
Finally, by analyzing transaction data limited to Austrian accounts, I investigate certain aspects of
the ETS which require a more thorough analysis of EUTL data, necessitating manual adjustments which
could hardly be upscaled to the whole ETS in a reasonable amount of time. This entails linking account
data with the transaction dataset in order to determine the corresponding account types and industry
sectors for each party involved in a transaction directed to or originating from Austrian accounts. Starting
with a comparison between the number of registered accounts and the number of active accounts, the
limited perspective enables a more thorough analysis compared to using the whole dataset. Across all
participating countries, the number of active accounts has decreased by 14.9% during phase III, ending an
upwards trend which culminated in a peak of 216% of the 2005 value in 2013. This negative development
prevails among most nations, with only Italy and Luxembourg exhibiting neutral to slightly positive
growth rates. However, my analysis based on the entire ETS treats both OHAs, PHAs and trading
accounts indisccriminately, meaning that changes in installation numbers are not addressed individually.
Using data from Austrian accounts enables me to overcome this limitation on the basis of OHA data.
During phase III, the number of Austrian installations participating in emissions trading has decreased
by 14.5% from its initial value of 220, which equals 74% of all registered OHAs. This downwards trend
has been more drastic with regard to PHAs and trading accounts, which exhibit a decrease of 28.9%
87
CHAPTER 6. CONCLUSION
between 2013 and 2017, from a mere 32% of the theoretical maximum by the beginning of phase III to
only 22.7% within 4 years. Unsurprisingly, participation is considerably higher with physical installations
than with other account types, which can be explained by the formal requirements involved in opening
an Operator Holding Account.
Given the limitations of the activity types used to identify industry sectors, I attempt an analysis based
on the more common NACE scheme by manually assigning each installation operator a corresponding
NACE code, revealing that certain categories such as steel production exhibit disproportionally high
market activity. However, I concede that the conclusions drawn from this observation are questionable
due to the limited sample size of the Austrian market. Taking into account this shortfall, extending the
scope to a larger fraction of the dataset would be advisable. In addition, linking transaction data with a
company database would make it possible to account for differences in annual turnover while analyzing
the impact of company size on market activity.
At last, I analyze the distribution of transactions across Austrian and foreign accounts in order to gain
insight both on the frequency of what I refer to as international transactions and on the countries involved.
Evidently, the share of transactions involving foreign accounts exceeds that of national transactions by
a considerable margin, amounting to 65.9% compared to 34.1% of the total number between 2013 and
2016. Whereas both categories exhibit declining transaction numbers during phase III, international
transactions are affected by this downwards movement to a greater extent. In terms of transaction
sizes, however, both variables are nearly on par, differing by only 3.9% on average. Proceeding to the
distribtion of countries involved in trading with Austrian accounts, Germany and the UK are dominant
both on the transferring and on the acquiring side, followed by Romania, France and the Netherlands. In
total, Austrian companies have interacted with trading partners from 22 countries. Whereas transactions
with smaller countries tend to involve account holders associated with Austrian companies, my analysis
reveals an exceptional share of banks and energy trading companies among German and UK accounts.
Considering this observation, I assume that at least part of said transactions are related to allowance
auctions. However, further research is needed to corroborate this hypothesis based on emipirical data.
88
Bibliography
Bundesministerium für Digitaliserung und Wirtschaftsstandort. (2019). Accredited Verifiers under the
EU ETS regulation 600/2012.
Carbon Market Watch. (2016). Industry windfall profits from Europe’s carbon market: How energy-
intensive companies cashed in on their pollution at taxpayers’ expense. https://carbonmarketwatch.