Abatement and Allocation in the Pilot Phase of the EU ETS

Environ Resource Econ (2011) 48:83–103DOI 10.1007/s10640-010-9399-9

Abatement and Allocation in the Pilot Phaseof the EU ETS

Barry Anderson · Corrado Di Maria

Accepted: 2 August 2010 / Published online: 20 August 2010© Springer Science+Business Media B.V. 2010

Abstract We use historical industrial emissions data to assess the level of abatement andover-allocation that took place across European countries during the pilot phase (2005–2007)of the European Union Emission Trading Scheme. Using a dynamic panel data model, weestimate the counter factual (business-as-usual) emissions scenario for EU member states.Comparing this baseline to allocated and verified emissions, we find that both over-allocationand abatement occurred, along with under-allocation and emissions inflation. Over the threetrading years of the pilot phase we find over-allocation of approximately 280 million EUAsand total abatement of 247 Mt CO2. However, we calculate that emissions inflation of approx-imately 73 Mt CO2 also occurred, possibly due to uncertainty about future policy designfeatures.

Keywords Emissions trading scheme · Climate policy · Dynamic panel data analysis

JEL Classification C23 · O13 · Q54 · Q58

1 Introduction

Since January, 2005 over 10,000 firms involved primarily in electricity generation and heavyindustry in the European Union have monitored, reported and verified their CO2 emissionsas participants in the European Union Emissions Trading Scheme (EU ETS), the largestgreenhouse gas emissions trading program in the world. As the cornerstone of European

B. AndersonSchool of Geography, Planning & Environmental Policy, University College Dublin, Planning Building,Richview, Clonskeagh, Dublin 14, Irelande-mail: [email protected]

C. Di Maria (B)Queen’s University Management School, Queen’s University Belfast, 25 University Square,Belfast BT7 1NN, UKe-mail: [email protected]

123

84 B. Anderson, C. Di Maria

climate policy, the EU ETS is motivated by the economic theory that market based policytools encourage the development and adoption of pollution abatement technology or behaviorto enable emissions reductions more efficiently than command and control style regulation(Milliman and Prince 1989; Jung et al. 1996; Requate and Unold 2003). The actions requiredto reduce emissions are theoretically mobilized by the price on CO2 emissions, and emissionsare reduced where the costs of doing so are less than the costs of buying the permits. Pilotphase (2005–2007) European Union emissions allowances (EUAs) were grandfathered freelyto participating installations based on burden sharing obligations under the Kyoto Protocol,past emissions, and economic projections for the pilot phase trading period. According toMontgomery (1972), grandfathering should not interfere with the efficiency and performanceof the system as well-defined property rights, and not the method of delivering those rights,should ensure an efficient outcome.

The spot price of EUAs varied significantly over the pilot phase, reaching a high of approx-imately e30 in early 2006, and hovering around e15 for most of 2006 before commencinga steady decline to a negligible price for the remainder of the pilot phase. While it is obviousthat the eventual price collapse was caused by the supply of permits exceeding demand, themore pressing question is what lead to this “excess supply” of EUAs?

At opposite ends of the spectrum of possibility, either firms continued with businessas usual and too many permits were freely grandfathered (over-allocation), or emissionsreductions (abatement) were much easier and cheaper to realize than previously expectedand deviations from expected emission levels decreased the demand for EUAs sufficiently tocause a price collapse. In theory, either of these two phenomenon could have caused the EUAprice collapse, but in practice it is likely a combination of the two due to the combination ofvoluntary (and largely unverified) firm level data used to construct historical emissions, andoptimistic economic growth forecasts which both were instrumental in determining the capfor the entire trading system.

In this study we make use of data for the pilot phase and panel data econometric techniquesto estimate what emissions would have been in the absence of the EU ETS, referred to here-after as the counter factual or BAU emissions. The estimation of a credible BAU emissionsscenario is necessary determine whether over-allocation or “excessive” abatement lead tothe excess supply of EUAs in the market. Similarly, the comparison of BAU emissions toallocated and verified emissions data reveals the occurrence and location of under-allocationand also emissions inflation by member state. Estimating the counter factual is not a deter-ministic exercise, and Ellerman and Buchner (2008) succinctly explain that, “the counterfactual is not observed and never will be. It can only be estimated, but there are better andworse estimates and much can be done to narrow the range of uncertainty, particularly whenthe evaluation is done ex post when the levels of economic activity, weather, energy pricesand other factors affecting the demand for allowances are known.”

The rest of the paper is organized as follows. Section 2 gives a brief overview of the EUETS pilot phase and reviews the existing literature, Sect. 3 describes that data used, andpresents our econometric specification. Estimation results and EU ETS pilot phase BAUestimates are shown in Sect. 4, along with tests for structural integrity and robustness checks,while Sect. 5 concludes.

2 The EU ETS Pilot Phase

It is possible to broadly summarize developments during the pilot phase of the EU ETS byreferring to Table 1, which displays allocated and verified EUAs and the final differences

123

Abatement and Allocation in the Pilot Phase of the EU ETS 85

Table 1 Final positions for pilot phase (Mt CO2)

Member State Allocation Emissions Gross Short Gross Long Net Position Net as %

AT 97.7 97.5 10.2 10.4 0.2 0.2

BE 178.7 162.9 25.1 40.9 15.8 8.8

CY 17.0 15.7 1.3 2.5 1.2 7.4

CZ 290.8 253.7 5.1 42.2 37.1 12.8

DE 1,486.3 1,440.0 90.7 137.0 46.3 3.1

DK 93.1 90.0 13.9 16.9 3.1 3.3

EE 56.3 40.1 0.4 16.6 16.2 28.8

ES 498.1 547.1 120.7 71.7 −49.0 −9.8

FI 133.9 119.9 6.8 20.8 14.0 10.4

FR 450.2 384.8 17.7 83.0 65.3 14.5

GR 213.5 213.9 13.3 12.8 −0.5 −0.2

HU 90.7 78.8 4.2 16.0 11.9 13.1

IE 57.7 64.8 12.4 5.4 −7.1 −12.2

IT 624.5 679.8 113.6 58.3 −55.3 −8.9

LT 34.4 19.1 0.5 15.8 15.3 44.4

LU 9.7 7.9 0.0 1.8 1.8 18.6

LV 12.2 8.6 0.4 3.9 3.5 28.9

MT 6.5 4.0 0.1 2.7 2.6 39.5

NL 259.3 236.9 20.2 42.6 22.4 8.7

PL 712.5 621.8 12.2 102.9 90.7 12.7

PT 110.7 100.7 3.9 14.0 10.0 9.1

SE 67.6 54.6 10.0 23.0 13.0 19.2

SI 26.1 26.6 1.8 1.2 −0.5 −2.1

SK 91.4 75.3 0.8 17.0 16.2 17.7

UK 627.8 744.3 165.2 48.8 −116.5 −18.5

EU25 6,246.6 6,088.9 650.5 808.2 157.7 2.5

Source: CITL

between the two for each member state. The gross long totals are the sum of the differencesbetween allocated and verified emissions for installations where the number of allocated per-mits was greater than verified emissions. The gross short totals are calculated the same wayfor installations where the allocation was insufficient to cover verified emissions and the netpositions is the difference between the two. The high emitting countries with excess permitsat verification were the Czech Republic, France, Germany and Poland, while Spain, Italy andthe UK installations needed to buy permits to cover their emissions levels. In 2005, 2006 and2007 the entire market was net long by 82.4 (3.9%), 36.1 (1.7%) and 26.9 (1.3%) millionEUAs,1 respectively, according to the Community Independent Transaction Log (CITL).

2.1 Abatement and Allocation

The literature on various aspects of the EU ETS is growing with much attention being paidto allocation issues (Ellerman and Buchner 2007; Neuhoff et al. 2006; Ahman et al. 2007),

1 The EU25 was net long 2.5% for the entire pilot phase.

123


competitiveness concerns (Demailly and Quirion 2006, 2008; Oberndorfer and Rennings2007; Grubb and Neuhoff 2006; Reinaud 2004; Ponssard and Walker 2008), and the driversof CO2 prices (Mansanet-Bataller et al. 2007; Alberola et al. 2008). There are few publishedstudies analyzing the levels of abatement which might have occurred as a result of the emis-sions constraint and the resulting CO2 price signal. In this study we define abatement asthe difference between observed emissions and estimated BAU levels. Conversely, we defineemissions inflation as the situation where observed emissions are greater than BAU estimates.

Given the circumstances surrounding the pilot phase, there is a real possibility of emissionsinflation due to the methodology used for pilot phase EUA allocation and uncertainty aboutfuture (2008–2012) trading period allocation methodologies. Most governments allocatedEUAs relative to ex ante BAU projections, with past emissions strongly influencing the moredetailed distribution of allowances. EU ETS participants may have learned that reportinginflated (historical) emissions leads to more generous future allocations and if the prospectof future allowance distribution being contingent upon recent emissions (‘updating’) is likely,there is a direct incentive to inflate actual emissions (See the discussion in Grubb et al. 2005).

We take a similar approach in identifying over and under-allocation (collectively referredto as misallocation) as the difference between BAU emissions and allocated EUAs. Naturally,this definition is imperfect as the EU ETS was designed to reduce emissions, not maintainthe BAU trends. If regulators of the pilot phase had set a community wide emissions cap ofa certain percentage below BAU levels, then the definition would be adjusted accordingly.However, the community wide cap of the EU ETS was constructed in a ‘bottom up’ fashionwith varying levels of ambition across member states making it difficult to judge what wasthe desired level of emissions reductions, relative to BAU for the entire trading community.

Ellerman and Buchner (2008) provide the only ex post analysis of the entire EU ETS andestimate levels of over-allocation and abatement that occurred only during 2005–2006. An“allocation ratio” is constructed to signify which countries were likely over-allocated basedon the aggregated relative net/gross final positions of their installations. The ratio of net shortor long to gross short or long is calculated by member state and the authors conclude thatfor 2005 and 2006 the maximum over-allocation is in the order of 6%, or 125 million EUAs,assuming that none of the length observed in these countries can be attributed to abatementor unexpected transient conditions that created length in these years. The authors use GDPgrowth trends, national energy and CO2 intensity trends, and assumptions about structuralbreaks in these trends to calculate BAU emissions and, as a result, ranges for levels of abate-ment. They conclude that between 130–200 Mt of CO2 were abated in 2005, and 140–220 Mtin 2006 for all member states.

In an ex-post analysis of the largest sector in the EU ETS by share of emissions2 Delarueet al. (2008a) focus specifically on the European power sector possibilities for short term CO2

abatement through fuel switching. Using both a non-calibrated and a historically calibratedsimulation model the authors’ estimates of abatement are between 34.4 and 63.6 Mt in 2005,and 19.2 and 35 Mt in 2006 in the power sector alone. These abatement estimates are basedsolely on the substitution effect of fuel switching and do not take into account the incomeeffect of emissions reductions resulting from industrial electricity consumers becoming moreefficient due to the rise in electricity prices resulting from the pass through of CO2 costs.3

In general the views about levels of abatement, or the difference between BAU and verifiedemissions, which likely occurred during the pilot phase can be summed up by contrasting

2 According to the CITL, ‘Combustion’ installations were allocated approximately 70% of pilot phase EUAs.3 Sijm et al. (2006) find the EUA cost pass-through rate was between 60–100% in Germany and Netherlands.

123


two insights. Kettner et al. (2008) conclude abatement played a minor role in determiningthe final net position of countries:

...it is rather unlikely that the EU ETS has already created incentives for abatementinvestments in the first trading years. Given the rather low carbon prices, it is alsoextremely unlikely that industries with a heavy CO2 cost component, such as cementand lime, have reduced their production levels because of the stringency of allowances.In a few installations the option for a fuel shift may have been used. Most probablythe only reduction option that was widely used was the improved operation of existingequipment. The reduction potential of this option is, however, rather limited.

In contrast, Ellerman and Buchner (2008) explain why the conclusion of significant earlyabatement and a significant gap between BAU emissions and observed emissions is logicaland plausible:

...the refutable presumption must be that the EU ETS succeeded in abating CO2 emis-sions in its first two years based on three observations.

1. A significantly positive EUA price. A significant price was paid for CO2 in2005–2006, which would have the effect of reducing emissions as firms adjustto this new economic reality.

2. Rising real output. Real output in the EU has been rising at the same time that therate of improvement in CO2 intensity has been declining, which has led to risingCO2 emissions before 2005.

3. Historical emissions data that indicate a reduction of emissions even after allowingfor plausible bias.

There seem to be logical arguments in favor of either concluding that pilot phase verifiedemissions were very similar to BAU emissions levels and the excess supply of permits wascaused by over-allocation, or that abatement created a significant difference between thecounter factual and verified emissions. In the rest of the paper we attempt to clarify this issueby estimating BAU emissions, and assess the previous claims. We begin by describing thedata and the econometric methodology used for our estimates of BAU emissions for the pilotphase.

3 Data and Econometric Specification

3.1 Data

In order to calculate an emissions counter factual for the EU ETS pilot phase we use his-torical data on European industrial emissions, industrial economic activity levels, weathereffects, and energy prices. The historical and pilot phase period data for factors influencinglevels of CO2 emissions (our dependent variable) are obtained from public data sources.However, the only data for EU ETS participants for the years prior to the trading period arethe historical baseline emissions as supplied by firms which are aggregated and reported inNational Allocation Plans (NAPs). This “baseline” period was generally around 2002 whiledifferent countries allowed the use of different methodologies, such as averaging or droppingcertain years, for establishing a ‘reasonable’ historical baseline (Ellerman and Buchner 2007).A single historical point does not allow for an econometric analysis and therefore we obtain

123


Table 2 EU ETS and Eurostat emissions matching

Eurostat emissions sources Eurostat emissions categories EU ETS sectors (EC, 2003)

Fuel combustion Public electricity and heat Combustion inst. > 20 MW

production

Fuel combustion Petroleum refining Mineral oil refineries

Fuel combustion Manufacture of solid fuels Coke ovens

and other energy industries Combustion inst. > 20 MW

Fuel combustion Iron and steel Pig iron or steel

Fuel combustion Non-Ferrous metals Metal ore (including sulphide ore)

roasting or sintering

Fuel combustion Chemicals Cement, clinker and lime ion

Glass including glass fibre

Ceramic products

Fuel combustion Pulp, paper and print Pulp from timber,

paper and board > 20 t/d

Fuel combustion Food processing, beverages Combustion inst. > 20 MW

and tobacco

Industrial processes Metal production Pig iron or steel

Metal ore (including sulphide ore)

roasting or sintering

Industrial processes Mineral production Cement, clinker and lime

Glass and ceramics

Source: Eurostat and European Commission (2003)

historical emissions data from Eurostat, and match historical emissions classified by NACEcodes to the sectors participating in the EU ETS.

3.1.1 Industrial CO2 Emissions

Eurostat aggregates emissions according to the NACE classification of the emitter. Based onthese classifications, we have selected a subset of total national emissions that are comparableto those sectors participating in the EU ETS. Table 2 outlines the process of matching histor-ical emissions according to NACE classifications in Eurostat to the sectors participating inthe EU ETS as outlined in Annex 1 of the EU ETS Directive (European Commission 2003).Just as the EU ETS does not cover the entire economy, we have chosen a subset of total CO2

emissions that best represents emissions from those industries participating in the EU ETS.Hereafter, the chosen subset of emissions is referred to as “Eurostat emissions”.

Eurostat emissions can be considered a reasonable proxy for historical EU ETS emissionsonly if the matching process produces satisfactory data in terms of the absolute magnitudeof emissions by country, and if the ratio between Eurostat emissions and the (unobserved)historical emissions of EU ETS firms is relatively constant over time. We use verified emis-sions data from 2005 to 2006 to analyze the quality of our chosen subset of total emis-sions from Eurostat. Table 3 displays Eurostat emissions, verified EU ETS emissions andthe ratio of the two. A ratio close to one is indicative of a good matching between NACEclassifications in Eurostat and EU ETS sectors. For most countries in both 2005 and 2006,

123


Table 3 EU ETS verified emissions and Eurostat emissions (Kt CO2)

Country 2005 2006 Historicalratio

EurostatA

VerifiedEUAs B

Ratio = B/A EurostatC

VerifiedEUAs D

Ratio = D/C

AT 35,598 33,373 0.937 34,962 32,383 0.926 1.009

BE 56,770 55,363 0.975 55,253 54,775 0.991 1.081

CY 4,416 5,079 1.150 4,608 5,259 1.141 1.021

CZ 83,411 82,455 0.989 86,744 83,625 0.964 1.261

DE 441,661 474,991 1.075 444,580 478,017 1.075 1.152

DK 26,253 26,476 1.009 33,547 34,200 1.019 1.002

EE 13,219 12,622 0.955 12,573 12,109 0.963 0.985

ES 183,219 183,627 1.002 174,270 179,697 1.031 1.054

FI 34,081 33,100 0.971 45,296 44,621 0.985 0.974

FR 137,844 131,264 0.952 131,409 126,979 0.966 1.066

GR 69,472 71,268 1.026 66,364 69,965 1.054 1.036

HU 29,612 26,162 0.883 29,653 25,846 0.872 0.928

IE 21,298 22,441 1.054 20,305 21,705 1.069 0.966

IT 224,965 225,989 1.005 225,186 227,439 1.010 1.065

LT 6,727 6,604 0.982 6,410 6,517 1.017 0.989

LU 2,338 2,603 1.113 2,491 2,713 1.089 1.213

LV 2,861 2,854 0.997 2,912 2,941 1.010 1.102

MT 1,960 1,971 1.006 1,976 1,986 1.005 1.014

NL 92,344 80,351 0.870 87,555 76,701 0.876 0.888

PL 213,054 203,150 0.954 223,631 209,616 0.937 1.004

PT 33,204 36,426 1.097 30,292 33,084 1.092 1.096

SE 21,325 19,382 0.909 20,876 19,889 0.953 0.972

SI 8,451 8,721 1.032 8,424 8,842 1.050 1.170

SK 24,640 25,232 1.024 24,867 25,543 1.027 1.083

UK 241,016 242,513 1.006 245,920 251,160 1.021 1.155

EU25 2,009,739 2,014,017 1.002 2,020,104 2,035,612 1.008 1.051

Source: Eurostat, DEHSt (2005), CITL and own calculations

the only years for which verified EU ETS data and Eurostat emissions data exist, the chosensubset of emissions is a good proxy for EU ETS firms as the ratios are close to one at boththe national and aggregate EU25 levels.

The last column in Table 3 contains the ratio of baseline emissions as reported in the NAPsof each member state, and summarized by the German Emissions Trading Authority (DEHSt2005), and Eurostat emissions. In countries where averaging was used to determine baselineemissions, averages of Eurostat emissions were also used for the relevant years to ensureconsistency. These baseline figures are not verified data and, according to Ellerman andBuchner (2007), the data collection effort was largely a voluntary submission by the indus-tries involved, conducted under severe time pressures with a lack of desirable verification.While industries cooperated in submitting the relevant emissions data, it is clear that therewas an incentive to attempt to influence allocation by reporting inflated historical emissions.

123


This point is highlighted by comparing the ratios for 2005 and 2006, to the baseline ratios.For example, Austrian EU ETS baseline emissions were very close to Eurostat emissionsin magnitude with a ratio of 1.009, but in the first two years of trading, verified emissionswere approximately 93–94% of Eurostat emissions. Either Austrian baseline emissions datawere inflated, or the EU ETS’ share of Eurostat emissions is shrinking due to other economicfactors. For most countries (18 out of 25), the historical baseline ratio is higher than the ver-ified emissions ratios for 2005 and 2006. It is not unreasonable to assume historical inflatedemissions reporting, given the clear incentives to do so for industry and we are comfortablethat the assumption of a relatively constant share of ETS industry of the chosen industrialemissions does not invalidate the use of Eurostat emissions as a proxy for historic EU ETSemissions. In order to make projections of emissions for 2007 we assume that the EU ETSshare of emissions in 2007 is the same as in 2006 since at the time of writing Eurostatemissions data for 2007 is not available.

3.1.2 Industrial Economic Activity Levels

Eurostat provides data on industrial production levels by NACE classification. Eurostat’sannual indices of production (working days adjusted) for Mining and Quarrying (NACE C),Manufacturing (NACE D), Energy (NACE E), Total Industry excluding construction (NACEC_D_E) and “Total industry excluding construction and the energy sector” are consideredas potential indicators of the economic activity levels for the EU ETS sectors. Based on theactivities of EU ETS firms as outlined in Annex 1 of the EU ETS Directive (European Com-mission 2003) both indices of economic activity for the manufacturing and energy sectorsare used in the analysis. As a robustness check, alternative indices are also used in a separateanalysis, whose results are discussed in Sect. 4.2. All indices are normalized such that the2005 level equals 100 for each member state.

3.1.3 Weather Factors

Eurostat contains variables for annual heating degree days for each member state from 1990onwards. Heating degree days are calculated as the difference between the daily mean temper-ature and 18 degrees Celsius, the ‘heating threshold’. Each day with an average temperaturebelow 18 degrees contributes to the annual total of heating degree days. A variable for cool-ing degree days is created using data from the European Climate and Assessment Database4

(ECA&D) where daily mean temperature above 18 degrees contribute to an annual measureof cooling degree days. Precipitation data was also extracted from the ECA&D and used tocreate an annual precipitation variable for each country as higher precipitation levels allowthe possibility of producing emissions free hydroelectricity. As a consequence, a reduction ofreal emissions could take place (Mansanet-Bataller et al. 2007). Wherever data are missingin the ECA&D, observations from www.tutiempo.net are used to fill the gaps.

3.1.4 Energy Prices

Fossil fuel prices play an important role in determining CO2 emissions in both the shortrun—due to the substitution effect of fuel switching from emissions intensive fuels to lessintensive alternatives (Delarue et al. 2008b; Delarue and D’haeseleer 2007)—and in the longrun, through the income effect and the process of directed technical change. The fossil fuel

4 See http://eca.knmi.nl/.

123

www.tutiempo.net

http://eca.knmi.nl/


prices considered relevant in determining national emissions are those of coal, oil and naturalgas. These fuels vary greatly in carbon content and therefore we expect their relative pricesto impact the CO2 intensity of electricity produced in the power generation sector. The pricesfor steam coal, natural gas and heavy fuel oil are obtained from the International EnergyAgency’s Energy Prices and Taxes database. Where data for different fuels is missing, theaverage price for European OECD countries is used as a proxy. While relative prices willdetermine the fuel mix of the utility sector, electricity prices are the relevant influence for theconsumption of electrical energy in the rest of the economy. Therefore, we estimate two spec-ifications, one including variables for coal, oil and natural gas, and another with a variablefor the price of electrical energy for industrial customers in EU member states as providedby Eurostat. Using this variable is imperfect as it does not explicitly capture the effects onemissions from fuel switching, but it does offer some insight into the effect of energy priceson induced technical change and CO2 emissions.

3.1.5 EU ETS Data

The Community Independent Transaction Log (CITL) is the definitive EU ETS resource cre-ated by the European Commission and contains all records of issuance, transfer, cancellation,retirement and banking of allowances that take place in the registry. The data in the regis-try was viewed and extracted using the CITL Viewer courtesy the European EnvironmentalAgency.5

3.2 Econometric Specification

We use dynamic panel data estimation techniques to estimate an econometric equation forindustrial CO2 emissions (Eurostat emissions). Dynamic models have been commonly usedin energy economics to study the demand for electricity, natural gas or other sources ofenergy. Such studies often use a version of the flow adjustment model from Houthakker et al.(1974) where the dependent variable is the quantity of the variable of interest, and a laggeddependent variable, price, and other explanatory variables are on the right hand side of theequation. While the purpose of our study is different from those demand studies, the methodsemployed are similar. The main difference between our study and energy demand studiesis the use of price variables for fossil fuels and for electrical energy for industrial users asexplanatory variables instead of a price for CO2 emissions.

The constructed panel data set is unbalanced as some countries are missing data for theearly part of our study period. The missing observations primarily occur in eastern Europeancountries that would eventually join the European Union. The existence of systematic missingobservations and the N × T dimensions of the constructed data set lead us to the bias cor-rected least squared dummy variable (LSDVC) estimation techniques as prescribed in (Bruno2005b). Instrumental variable (IV) and General Method of Moments (GMM) dynamic panelestimators as proposed in Anderson and Hsiao 1982; Arellano and Bond 1991 and Blundelland Bond 1998 are also considered, but LSDVC techniques are preferred as IV and GMMestimator properties hold for data sets with large N , but can be severely biased and imprecisein panel data with a small number of cross-sectional units (Bruno 2005a). Judson and Owen(1999) and Bun and Kiviet (2003) also prove the virtues of least square dummy variable(LSDV) estimation in small N panels that are common in macroeconomic studies, but onlyfor the case of balanced panel data sets. Bruno’s (2005b) LSDVC estimator is adapted for the

5 The CITL viewer is available at http://dataservice.eea.europa.eu/atlas/viewdata/viewpub.asp?id=3529.

123

http://dataservice.eea.europa.eu/atlas/viewdata/viewpub.asp?id=3529


case of unbalanced panels and is deemed the most appropriate estimation technique giventhe characteristics of our data set. LSDVC estimators are calculated using the algorithmoutlined in Bruno (2005a) where consistent estimators are initially calculated using eitherGMM (Arellano and Bond 1991; Blundell and Bond 1998) or IV (Anderson and Hsiao 1982)methods as a first step, and these estimators are then used to initiate a second step wherebias in the consistent IV or GMM estimators is corrected to surrender the final bias-correctedparameter estimates.

We use an unbalanced panel data set consisting of 25 countries6 to estimate the followingequations:

C O2i t = β1C O2i t−1 + β2econ_Dit + β3econ_Eit + β4hddit + β5cddit + β6rainit

+β7electi t + ui + eit

C O2i t = β1C O2i t−1 + β2econ_Dit + β3econ_Eit + β4hddit + β5cddit + β6rainit

+β7natgasit + β8coalit + β9oili t + ui + eit

where CO2 is Eurostat emissions, econ_D and econ_E are Eurostat’s indices of productionfor the Manufacturing and Energy sectors, respectively, elect is the price of electrical energyfor industrial customers and oil, coal, and natgas are the prices for oil, coal and naturalgas, hdd and cdd are heating and cooling degree days, rain is an annual measure of pre-cipitation, u is the country fixed effect, and e is the idiosyncratic error term. We calculateArellano and Bond (1991) estimators in the first step of the LSDVC algorithm and includetime dummies based on the requirement that ‘first step’ estimators must be consistent in orderto initialize the bias correcting second step. We initially included time period dummies basedon the recommendation of Roodman (2006) when calculating Arellano and Bond (1991)estimators but joint insignificance of the time dummies along with stronger model perfor-mance (as a predictor) without dummies, led us to repeat the analysis with time dummiesomitted. All variables are in natural logarithms and i and t subscripts denote country andyear, respectively.

4 Results

4.1 Estimation Results

The dynamic panel estimation is done using the techniques explained in Bruno (2005b) whereArellano and Bond (1991) (AB hereafter) estimators are generated in the first step to initiatebias correction in the second step of the algorithm. Table 4 gives the estimation results forfour specifications: equations (1) and (2) are estimated using data from 1990 to 2004 and con-tain different variables for energy prices. Equations (3) and (4) contain interaction variablesthat allow us to perform a Chow test for the structural integrity of certain parameters overthe study period when using both fossil prices and electrical energy prices as explanatoryvariables. The results of the Chow test are discussed further in Sect. 4.2

The significance of the lagged dependent variable is as expected across specifications,and is of similar magnitude to the lagged dependent variable coefficients in Kamerschenand Porter (2004) but slightly smaller than those for the industrial energy demand equation

6 Romania and Bulgaria are excluded from the analysis as they joined the scheme in 2007, and there are issuesconcerning their EU ETS and historical data.

123


Table 4 Estimation results

(1) (2) (3) (4)

CO2t−1 0.663*** 0.666*** 0.652*** 0.647***

(0.056) (0.055) (0.072) (0.048)

NACE_D 0.013 −0.065 0.014 −0.048

(0.051) (0.048) (0.052) (0.046)

NACE_E 0.282*** 0.321*** 0.265*** 0.301***

(0.100) (0.069) (0.086) (0.079)

hdd 0.055 0.077 0.084 0.096

(0.091) (0.079) (0.085) (0.077)

cdd −0.000 0.004 0.010 0.009

(0.012) (0.010) (0.011) (0.009)

rain −0.072** −0.066** −0.044* −0.046***

(0.031) (0.033) (0.024) (0.017)

elect 0.011 −0.006

(0.053) (0.058)

natgas −0.055 −0.029

(0.050) (0.045)

coal 0.033 0.017

(0.051) (0.048)

oil 0.055 0.042

(0.035) (0.030)

euets 0.160 0.943

(0.566) (0.743)

NACE_D_euets −0.012 0.090

(0.104) (0.092)

NACE_E_euets −0.029 −0.140

(0.114) (0.118)

elect_euets 0.001

(0.066)

natgas_euets 0.072

(0.061)

coal_euets −0.005

(0.089)

oil_euets −0.225***

(0.085)

Chow test: Prob > χ2 − − 0.2349 0.0147

Period 1990–2004 1990–2004 1990–2007 1990–2007

Observations 201 244 251 294

*** p<0.01, ** p<0.05, * p<0.1

in Considine (2000). The signs and statistical significance of the economic variables for theenergy sector and manufacturing sectors are also not surprising as a major source of emissionsin manufacturing is from electrical energy consumption.

123


This is reflected in the statistically insignificant coefficient on N AC E_D coupled withstrong significance of the N AC E_E parameter estimate across specifications. The coeffi-cient estimate on the energy price variables lack statistical significance, likely due to the factthat substitution between (energy and non-energy) factors of production according to relativeprices is not an immediate process.

The weather variables for heating and cooling degree days are not statistically significantand these results are similar to those found in earlier industrial energy demand studies byPolemis (2007) and Kamerschen and Porter (2004). This study focuses only on industrialemissions, and not residential or total national emissions. If the focus were shifted to residen-tial or commercial emissions, or emissions arising solely from the electricity sector, weathervariables would be expected to grow both in magnitude and statistical significance, as foundin Considine (2000) and Pardo et al. (2002). The coefficient on the precipitation variableis statistically significant and of the expected sign; increased precipitation levels allow forgreater production of CO2 free hydro electricity.

4.2 Parameter Stability and Robustness Checks

It possible that the parameter estimates displayed in equation (1) and (2) of Table 4 are notstable over the study period, due to a structural break caused by the the start of the EU ETS in2005. If this were the case then it would be inappropriate to apply the parameters estimatedusing data from 1990 to 2004 to the data for the pilot phase in order calculate expectedvalues. We perform a Chow test for structural change of parameters across time as explainedin Woolridge (2009) to test if a structural break occurred as a result of the presence of theEU ETS. We create a dummy variable (called ‘euets’) that takes a value of 1 for 2005–2007and 0 for the previous years for all countries. The dummy is interacted with the economicactivity variables for manufacturing and energy sectors and the energy price variables, andthe joint significance of the interaction terms is tested.

Given the p values of the test statistics for the joint significance of euets and the interac-tion terms for both the specification using electricity prices (Prob > χ2 = 0.2349), we cannotreject the null hypothesis of no structural break for the electricity price parameter estimate,and conclude that the existence of the EU ETS did not significantly change the relationshipbetween the level of CO2 emissions, and the potentially endogenous explanatory variables forthe estimation using electricity prices. The opposite is true when the same test is performedfor the specification using fossil fuel prices (Prob > χ2 = 0.0147).

Finding that the levels of economic activity in power generation and manufacturing werenot significantly altered by the establishment of the EU ETS is in line with the literature on theEU ETS and EU industrial “competitiveness.” In an ex-ante study Reinaud (2004) concludesthat the cost impacts of the EU ETS on some key industrial sectors are not likely to leadto major negative impacts on the competitiveness in the near term.7 Anger and Oberndorfer(2008) use German firm level data to empirically support the same conclusion of no economicdisadvantage to EU ETS firms in the pilot phase by econometrically examining EU ETS firms’revenues and employment levels. Most of the competitiveness concerns associated with theEU ETS were focused on sectors that participated in global markets with competitors thatdid not face emissions constraints, such as iron/steel and the cement industry. The iron andsteel industry experienced weak production losses and profitability losses—as measured by

7 Reinaud (2004) defines competitiveness losses as a loss of output through CO2 leakage while consideringthe effects of CO2 costs pass through, allocation and market structure.

123


Table 5 Calculating EU ETS projected emissions for Austria, 2005 (Kt CO2)

Year Allocated Verified Eurostat CO2 Fitted value Ratioa EU ETS BAUA B C D E = B/C F = E∗D

2005 32,413 33,373 35,598 35,818 0.937 33,579

Source: Own Calculations on NAPs, CITL and Eurostat dataa As shown in Table 3, figures may not add due to rounding

changes in EBITDA8—are likely to be positive given the level of free allocation in the pilotphase (Demailly and Quirion 2008). Demailly and Quirion (2006) conclude that given thepilot phase allocations, neither the production level nor the EBITDA of the European cementindustry is significantly impacted, even for a very high CO2 price. In an ex-ante study, Smaleet al. (2006) have similar findings where most UK participating sectors would be expected toprofit in general, although with a modest loss of market share in the case of steel and cement,and closures in the case of aluminum.

While it is imperative that the econometric estimation give results that are in line with apriori expectations in terms of coefficient sign and significance, the purpose of our estimationis not to draw inference from parameter estimates, but to calculate expected values for EUETS CO2 emissions. Accordingly, we compare fitted values from the above specification toobserved emissions data in order to analyze the model’s performance as a tool for predictingnational emissions. The high level of correlation (0.9987) between past emissions and fittedvalues from our model9 gives us confidence that the specification using electricity prices isappropriate for calculating EU ETS BAU emissions.

To verify the robustness of our results, we repeat the analysis using different indices ofindustrial production as indicators of economic activity, and different dynamic panel dataestimators to initialize bias correction and compute LSDVC estimators. The different indicesare those explained in Sect. 3, while AB, Anderson and Hsiao (1982) and Blundell and Bond(1998) estimators are used as first step estimators for each economic index. We judge theperformance of the alternative specifications based on the correlation and variance betweenfitted values from the model and observed values. We conclude that the specification using ABestimators to initialize bias correction, and N AC E_D and N AC E_E as economic variablesperforms strongest as a prediction tool.10

4.3 Pilot Phase Results

Pilot phase BAU emissions are calculated by taking the projections for industrial CO2 emis-sions and multiplying that by the percentage of Eurostat emissions that are attributable to theETS sectors as shown in Table 3. For example, Austria’s EU ETS BAU emissions for 2005are calculated as follows:

EU ETS BAU emissions are compared to allocated and verified EUAs to determine the lev-els of misallocation, abatement and emissions inflation in each member state for each year.For example, in 2005 Austria experienced emissions abatement of 206 Kt CO2 (33,579–33,373) , and zero tonnes of emissions inflation since verified EUAs are less than the EUETS BAU estimate. Continuing with allocation, Austrian firms were under-allocated 1,166

8 Earnings before interest taxes depreciation and amortization.9 We also regress the fitted values on observed values and that OLS regression has an R2 of 0.9970.10 The full results of these robustness tests are available from the authors upon request.

123


Table 6 Pilot phase outcomes (MtCo2)

Country Misallocation Abatement & inflation

Levels % of allocation Levels % of allocation(A) (B) (C) (D)

AT −4 −4.1 −4.2 −4.3

BE −1.1 −0.6 −16.9 −10.4

CY −0.2 −1.1 −1.4 −9.1

CZ −2.5 −0.8 −39.4 −15.5

DE 46.6 3.1 0.3 0.0

DK −0.2 −0.2 −3.2 −3.6

EE 13.7 24.4 −2.5 −6.3

ES −47.7 −9.6 3.8 0.7

FI 5.4 4.0 −8.2 −6.8

FR 44.7 9.9 −20.5 −5.3

GR −11.3 −5.3 −10.8 −5.1

HU 8.6 9.4 −3.3 −4.2

IE −15 −26.0 −7.3 −11.2

IT −112.4 −18.0 −57.1 −8.4

LT 14.6 42.3 −0.7 −3.7

LU 2.5 26.0 0.7 9.1

LV 3.4 28.2 −0.1 −1.1

MT 0.6 9.0 0 0.6

NL 15.3 5.9 −7.1 −3.0

PL 92.4 13.0 2.1 0.3

PT −1.4 −1.3 −11.5 −11.4

SE 6.1 9.0 −6.9 −12.6

SI −0.9 −3.6 −0.4 −1.5

SK 15 16.4 −1.2 −1.6

UK −100 −15.9 22.3 3.0

EU25 −27.9 0.4 −173.6 −2.8

Source: Own calculations

thousand (33,579–32,413) EUAs with no over-allocation occurring since the amount of allo-cated EUAs is less than the BAU level (Table 5). Calculations for each member state byyear for the levels of misallocation, abatement and emissions inflation are displayed in theAppendix.11

Tables 6 summarizes the results for the pilot phase by member state in absolute lev-els (columns A&C) and expressed as percentages (columns B&D) of each member state’sallowance allocation. Negative values for misallocation indicate under-allocation while posi-tive number signify over-allocation. The same is true with inflation (negative values) and

11 As with any econometric exercise, there is a certain level of uncertainty in our parameter and BAU esti-mates. However, errors operate in both directions and we have no reason to believe that our parameter or BAUestimates contain significant systematic bias. While yearly BAU estimates might be stronger for some memberstates than others, it appears the aggregate of the estimates is a credible estimate for EU25 BAU emissionsgiven the high correlation of fitted values and observed emissions in the past for the entire data set.

123


Table 7 Summary of observed vs. BAU emissions for EU25 (Mt CO2)

Country Year Alloc. EUA Verif. EUA BAU Under-alloc. Over-alloc. Infl. Abate.

EU25 2005 2,096 2,014 2,098 79 77 8 92

EU25 2006 2,072 2,036 2,097 118 92 26 88

EU25 2007 2,079 2,052 2,079 111 111 39 67

EU25 2005–2007 6,247 6,101 6,275 308 280 73 247

Source: Own calculations on NAPs, CITL and Eurostat data

abatement (positive values). Column A reveals the majority of over-allocation occurredGermany, France and Poland by volume. Conversely, Spain, Italy and the UK were themost under-allocated countries by volume. Abatement was common across most memberstates with Italy and the Czech Republic realizing the most emissions reductions in absoluteterms. The UK is found to have the highest levels of emissions inflation over the pilot phase.

However, the previously mentioned countries are also some of the largest economies inthe trading community and therefore, it might be more useful to consider the magnitude ofmisallocation, abatement and inflation in relative terms. Columns B&D display the preva-lence of over-allocation in new member states contrasted by widespread under-allocation innon-Eastern European countries with Ireland, Italy, Spain and the U.K. having the highestrelative levels of under-allocation. Most member states achieved modest levels of abatementwith Belgium, the Czech Republic, Ireland, Portugal and Sweden all realizing double digitpercentage reductions. Emissions inflation is found to be the highest in Luxembourg (9%)with the U.K also having emissions higher than BAU levels (3.1%).

Table 7 provides an overview of pilot phase outcomes by year for the Eu25 as a whole.The fifth column in the table contains our estimate of BAU emissions for EU25 memberstates over the pilot phase period. In 2005, 2006 and 2007 there was net abatement12 of 84.2,61.7 and 27.6 Mt CO2, respectively, which is the expected trend due to the declining priceof EUAs. In 2005, 2006 and 2007 the entire system was net under-allocated by 1.8, 25.6 and0.5 million EUA’s, respectively. Over the pilot phase, we find that the community wide capwas set only 27.9 million EUAs (0.45%) lower than BAU levels across the entire tradingcommunity.

5 Discussion

The goal of this paper is to perform an ex-post evaluation of the pilot trading period andestimate a credible BAU emissions scenario for EU member states. Comparing the BAU toallocated and verified EUAs allows us to identify any misallocation, abatement and emis-sions inflation that may have occurred. The quantitative findings of this study, i.e., 2.8% netemissions abatement and 0.45% net under-allocation at the EU25 level, refer to the pilotphase of the EU ETS and cannot thus be generalized to other emissions trading programs.The details regarding allocation methodologies and uncertainty about the future of Europeanclimate policy are unique and our quantitative results are only intended to give insights intoof the EU ETS in the pilot phase.

The results of our analysis highlight the importance of reliable and accurate datawhen designing an emissions trading scheme to create incentives to reduce emissions.

12 The difference between emissions inflation and abatement.

123


Table 8 NAP projections and verified EUAs (Mt CO2)

Country Ave. emissions NAP BAU NAP abatement (%) EU ETS BAU Abatement (%)

AT 32.5 34.7 6.8 33.9 4.3

BE 54.3 64 17.8 59.9 10.4

CY 5.2 5.7 8.7 5.7 9.1

CZ 84.6 103.7 22.5 97.7 15.5

DE 480 499 4.0 479.9 0.0

DK 30 39.3 30.9 31.1 3.6

EE 13.4 14 4.8 14.2 6.3

ES 183.2 181.6 −0.9 181.9 −0.7

FI 40.1 46.6 16.2 42.8 6.8

FR 128.3 163.8 27.7 135.1 5.3

GR 71.3 76 6.6 74.9 5.1

HU 26.3 31.3 19.1 27.4 4.2

IE 21.8 23 5.5 24.2 11.2

IT 226.6 244.5 7.9 245.6 8.4

LT 6.4 14 119.7 6.6 3.7

LU 2.6 3.7 40.8 2.4 −9.1

LV 2.9 4.4 52.7 2.9 1.1

MT 1.3 2.9 119.9 2 50.4

NL 79 98.6 24.8 81.3 3.0

PL 207.5 263 26.8 206.8 −0.3

PT 33.6 38.9 15.9 37.4 11.4

SE 18.2 26.6 46.1 20.5 12.6

SI 8.9 9.5 7.1 9 1.5

SK 25.1 36.2 44.2 25.5 1.6

UK 250.1 267.3 6.9 242.6 −3.0

EU25 2033 2292.3 12.7 2091.6 2.8

Source: Own calculations on NAPs, CITL and DEHSt (2005)

By observing in Table 3 that 18 of 25 member states have higher ratios of Eurostat emissionsto EU ETS emissions in the baseline period than in the pilot phase, we are forced to questionthe integrity of the historical baseline data. Table 8 displays the average verified emissionsby member state, the BAU projections from the NAPs as reported by the German EmissionsTrading Authority DEHSt (2005) and our EU ETS BAU emissions estimates and the amountof abatement that occurred relative to the NAP BAU estimates and our BAU estimates. Insummary, if we consider the ex ante BAU emissions from the NAPs accurate and trustwor-thy then the conclusion is that net abatement of 12.7% was realized across the EU25. Ourfindings of net abatement of 2.8% force us to conclude that the ex-ante projections of pilotphase BAU emissions were of questionable quality.

When questionable baseline data is combined with overly optimistic ex-ante economicprojections, the likely result is inflated BAU emissions projections and almost certainly poorpermit allocation. Misallocation in both directions undoubtedly leads to equity issues, whileover-allocation increases likelihood of market distortions and perverse incentives to reduceemissions due to the fear of ‘updating’ discussed in Sect. 2.1.

123


It appears the European Commission learned its lesson during the pilot phase and amend-ments to the original EU ETS Directive were adopted by the European Commission inJanuary, 2008.13 The main changes in the EU ETS design are a clearly announced movetowards increased auctioning of allowances, dismissal of NAPs, and a tightening of the over-all cap from 2013 en route to a target of 21% reduction below 2005 verified EU ETS levels.Auctioning, combined with top down cap formation is necessary so that distortions arisingfrom free-riding incentives for both industries and governments in the generation of projec-tions do not repeat themselves. From 2013 onwards, it appears that any strategic behaviourcarried out in the early years of trading will have a muted effect as only firms that are exposedto international competition will be awarded partial free allocations.

At the time of writing, it seems unlikely that any rewards will be gained by strategicactivities in the second phase (2008–2012) of trading. While pilot phase verified EU ETSemissions are below our BAU estimates, we also find gross over-allocation in 19 of the25 member states (See Appendix) over the three initial years of trading. This supports ourhypothesis that at least some strategic behavior was rewarded early on. As the recession inEurope continues to deepen and industrial activity falls there is much downward pressureon EUA prices. The dramatic difference between the projected economic path at the timeof 2008–2012 allowance allocation, and the current economic circumstances has altered theemissions trading landscape in the short term, and allocations that may have been somewhatof a strategic reward, are likely to have little value in the current economic environment.

Various countries of the world are in the process of designing climate change policy withemissions trading as a key feature. The most prominent is the United States where cap andtrade is already occurring at the regional level with the likelihood of federal legislation grow-ing constantly as support spreads from environmental groups, to some companies and othernon-traditional sources such as religious groups (Stavins 2008). As one of the largest emittersin the world,14 U.S. climate change policy is important due to the magnitude of its emissions,but also in political terms. Credible climate policy leading to a reasonable CO2 price in theU.S. is a necessary step towards the sort of global agreement that is likely required to dealwith the global nature of climate change.

We would like to conclude summarizing the main insights of our study for trading schemesbeyond the EU ETS. Our analysis points at three key elements for the success of an emissionstrading scheme: first, we highlight the necessity of reliable emissions data to establish a soundfooting for the construction of the BAU scenario; second, our discussion of the incentivesfor member states and industry, implies that a system of centralized cap setting (top-down) isless prone to produce lax caps and to lead to overallocation; finally, auctioning allowances,by limiting the possibility of windfall gains for the participants in the scheme is likely toreduce strategic behavior, and rent seeking.

Acknowledgements We would like to thank Frank Convery, Jurate Jaraite, Emiliya Lazarova, and oneanonymous referee for useful comments and suggestions. Barry Anderson gratefully acknowledges the finan-cial support of the Environmental Protection Agency of Ireland under the STRIVE program.

13 See adopted text at http://dataservice.eea.europa.eu/atlas/viewdata/viewpub.asp?id=3529.14 China passed the US in 2006 to become the world’s largest emitter.

123

http://dataservice.eea.europa.eu/atlas/viewdata/viewpub.asp?id=3529


Appendix

Table A.1 Summary of observed vs. BAU emissions by country, 2005–2007 (Kt CO2)

Country Year Allocated Verified BAU Under-all. Over-all. Inflation Abatement

AT 2005 32,413 33,373 33,579 1,166 – – 206

BE 2005 58,310 55,363 60,294 1,984 – – 4,931

CY 2005 5,471 5,079 5,730 259 – – 651

CZ 2005 96,920 82,455 95,859 – 1,061 – 13,404

DE 2005 493,482 474,991 487,700 – 5,782 – 12,709

DK 2005 37,304 26,476 30,981 – 6,323 – 4,505

EE 2005 16,747 12,622 14,724 – 2,023 – 2,102

ES 2005 172,161 183,627 178,197 6,036 – 5,430 –

FI 2005 44,666 33,100 43,718 – 948 – 10,618

FR 2005 150,412 131,264 137,449 – 12,963 – 6,185

GR 2005 71,162 71,268 74,223 3,061 – – 2,955

HU 2005 30,236 26,162 27,536 – 2,700 – 1,374

IE 2005 19,237 22,441 23,729 4,492 – – 1,288

IT 2005 216,150 225,989 243,535 27,385 – – 17,546

LT 2005 13,499 6,604 6,597 – 6,902 7 –

LU 2005 3,229 2,603 2,477 – 752 126 –

LV 2005 4,070 2,854 2,870 – 1,200 – 16

MT 2005 2,086 1,971 1,999 – 87 – 28

NL 2005 86,452 80,351 83,454 – 2,998 – 3,103

PL 2005 237,558 203,150 210,781 – 26,777 – 7,631

PT 2005 36,909 36,426 36,700 – 209 – 274

SE 2005 22,289 19,382 21,103 – 1,186 – 1,721

SI 2005 9,138 8,721 8,903 – 235 – 182

SK 2005 30,471 25,232 25,827 – 4,644 – 595

UK 2005 206,072 242,513 240,314 34,242 – 2,199 –

EU25 2005 2,096,444 2,014,017 2,098,281 78,626 76,789 7,762 92,026

AT 2006 32,649 32,383 34,450 1,801 – – 2,067

BE 2006 59,952 54,775 60,934 982 – – 6,159

CY 2006 5,612 5,259 5,768 156 – – 509

CZ 2006 96,920 83,625 97,087 167 – – 13,462

DE 2006 495,488 478,017 481,498 – 13,990 – 3,481

DK 2006 27,908 34,200 29,444 1,536 – 4,756 –

EE 2006 18,200 12,109 14,070 – 4,130 – 1,961

ES 2006 166,186 179,697 186,437 20,251 – – 6,740

FI 2006 44,618 44,621 39,083 – 5,535 5,538 –

FR 2006 149,967 126,979 136,473 – 13,494 – 9,494

GR 2006 71,162 69,965 75,666 4,504 – – 5,701

HU 2006 30,236 25,846 27,068 – 3,168 – 1,222

123


Table A.1 continued

Country Year Allocated Verified BAU Under-all. Over-all. Inflation Abatement

IE 2006 19,238 21,705 24,950 5,712 – – 3,245

IT 2006 205,050 227,439 248,685 43,635 – – 21,246

LT 2006 10,577 6,517 6,944 – 3,633 – 427

LU 2006 3,229 2,713 2,324 – 905 389 –

LV 2006 4,058 2,941 2,937 – 1,121 4 –

MT 2006 2,167 1,986 1,991 – 176 – 5

NL 2006 86,388 76,701 82,076 – 4,312 – 5,375

PL 2006 237,558 209,616 202,741 – 34,817 6,875 –

PT 2006 36,909 33,084 38,880 1,971 – – 5,796

SE 2006 22,484 19,889 20,245 – 2,239 – 356

SI 2006 8,692 8,842 9,164 472 – – 322

SK 2006 30,487 25,543 25,636 – 4,851 – 93

UK 2006 206,005 251,160 242,747 36,742 – 8,413 –

EU25 2006 2,071,740 2,035,612 2,097,296 117,927 92,371 25,975 87,659

AT 2007 32,649 31,751 33,648 999 – – 1,897

BE 2007 60,429 52,795 58,604 – 1,825 – 5,809

CY 2007 5,899 5,396 5,674 – 225 – 278

CZ 2007 96,920 87,738 100,283 3,363 – – 12,545

DE 2007 497,302 487,004 470,502 – 26,800 16,502 –

DK 2007 27,903 29,407 32,891 4,988 – – 3,484

EE 2007 21,344 15,330 13,778 – 7,566 1,552 –

ES 2007 159,717 186,184 181,119 21,402 – 5,065 –

FI 2007 44,620 42,541 45,683 1,063 – – 3,142

FR 2007 149,776 126,635 131,501 – 18,275 – 4,866

GR 2007 71,162 72,717 74,884 3,722 – – 2,167

HU 2007 30,236 26,835 27,542 – 2,694 – 707

IE 2007 19,240 21,246 24,039 4,799 – – 2,793

IT 2007 203,255 226,369 244,657 41,402 – – 18,288

LT 2007 10,318 5,999 6,287 – 4,031 – 288

LU 2007 3,229 2,567 2,367 – 862 200 –

LV 2007 4,035 2,849 2,928 – 1,107 – 79

MT 2007 2,286 2,027 1,959 – 327 69 –

NL 2007 86,477 79,875 78,496 – 7,981 1,379 –

PL 2007 237,543 209,602 206,745 – 30,798 2,857 –

PT 2007 36,909 31,183 36,590 – 319 – 5,407

SE 2007 22,846 15,348 20,156 – 2,690 – 4,808

SI 2007 8,246 9,049 8,952 706 – 97 –

SK 2007 30,487 24,517 25,030 – 5,457 – 513

UK 2007 215,875 256,581 244,854 28,979 – 11,727 –

EU25 2007 2,078,703 2,051,545 2,079,169 111,423 110,957 39,448 67,071

123


References

Ahman M, Burtraw D, Krueger J, Zetterberg L (2007) A ten-year rule to guide the allocation of EU emissionallowances. Energy Policy 35(3):1718–1730

Alberola E, Chevallier J, Chèze B (2008) Price drivers and structural breaks in European carbon prices 2005–2007. Energy Policy 36(2):787–797

Anderson TW, Hsiao C (1982) Formulation and estimation of dynamic models using panel data. J Econ18(1):47–82

Anger N, Oberndorfer U (2008) Firm performance and employment in the EU emissions trading scheme: anempirical assessment for Germany. Energy Policy 36(1):12–22

Arellano M, Bond S (1991) Some tests of specification for panel data: Monte Carlo evidence and an applicationto employment equations. Rev Econ Stud 58(2):277–297

Blundell R, Bond S (1998) Initial conditions and moment restrictions in dynamic panel data models. J Econ87(1):115–143

Bruno GSF (2005a) Approximating the bias of the LSDV estimator for dynamic unbalanced panel data models.Econ Lett 87(3):361–366

Bruno GSF (2005b) Estimation and inference in dynamic unbalanced panel-data models with a small numberof individuals. Stata J 5(4):473–500

Bun MJG, Kiviet JF (2003) On the diminishing returns of higher-order terms in asymptotic expansions ofbias. Econ Lett 79(2):145–152

Considine TJ (2000) The impacts of weather variations on energy demand and carbon emissions. ResourEnergy Econ 22(4):295–314

DEHSt (2005) Implementation of emissions trading in the EU: national allocation plans of all EU states.Discussion paper, German Emissions Trading Authority

Delarue E, D’haeseleer WD (2007) Price determination of ETS allowances through the switching level of coaland gas in the power sector. Int J Energy Res 31(11):1001–1015

Delarue E, Ellerman A, D’haeseleer WD (2008a) Short term CO2 Abatement for the European Power Sector.Discussion paper, Katholieke Universiteit Leuven

Delarue E, Voorspools K, D’haeseleer WD (2008b) Fuel switching in the electricity sector under the EU ETS:review and prospective. J Energy Eng 134(2):40–46

Demailly D, Quirion P (2006) CO2 abatement, competitiveness and leakage in the European cement industryunder the EU ETS: grandfathering versus output-based allocation. Clim Policy 6:93–113

Demailly D, Quirion P (2008) European emission trading scheme and competitiveness: a case study on theiron and steel industry. Energy Econ 30(4):2009–2027

Ellerman AD, Buchner BK (2007) The European Union emissions trading scheme: origins, allocation, andearly results. Rev Environ Econ Policy 1(1):66–87

Ellerman AD, Buchner BK (2008) Over-allocation or abatement? A preliminary analysis of the EU ETS basedon the 2005-06 emissions data. Environ Resour Econ 41(2):267–287

European Commission (2003) Directive 2003/87/EC of the European Parliament and of the Council of 13 Octo-ber 2003 Establishing a scheme for greenhouse gas emission allowance trading within the communityand Amending Council Directive 96/61/EC

Grubb M, Azar C, Persson MU (2005) Allowance allocation in the European emissions trading system: acommentary. Clim Policy 5:127–136

Grubb M, Neuhoff K (2006) Allocation and competitiveness in the EU emissions trading scheme: policyoverview. Clim Policy 6:7–30

Houthakker HS, Verleger PK Jr, Sheehan DP (1974) Dynamic demand analyses for gasoline and residentialelectricity. Am J Agric Econ 56(2):412–418

Judson RA, Owen AL (1999) Estimating dynamic panel data models: a guide for macroeconomists. Econ Lett65(1):9–15

Jung C, Krutilla K, Boyd R (1996) Incentives for advanced pollution abatement technology at the industrylevel: an evaluation of policy alternatives. J Environ Econ Manag 30(1):95–111

Kamerschen DR, Porter DV (2004) The demand for residential, industrial and total electricity, 1973–1998.Energy Econ 26(1):87–100

Kettner C, Köppl A, Schleicher S, Thenius G (2008) Stringency and distribution in the EU emissions tradingscheme: first evidence. Clim Policy 8:41–61

Mansanet-Bataller M, Pardo A, Valor E (2007) CO2 prices, energy and weather. Energy J 28(3):73–92Milliman SR, Prince R (1989) Firm incentives to promote technological change in pollution control. J Environ

Econ Manag 17(3):247–265Montgomery W (1972) Markets in licenses and efficient pollution control programs. J Econ Theory 5:395–418

123


Neuhoff K, Martinez KK, Sato M (2006) Allocation, incentives and distortions: the impact of EU ETS emis-sions allowance allocations to the electricity sector. Clim Policy 6:73–91

Oberndorfer U, Rennings K (2007) Costs and competitiveness effects of the European Union emissions tradingscheme. Eur Environ 17(1):1–17

Pardo A, Meneu V, Valor E (2002) Temperature and seasonality influences on Spanish electricity load. EnergyEcon 24:55–70

Polemis ML (2007) Modeling industrial energy demand in Greece using cointegration techniques. EnergyPolicy 35(8):4039–4050

Ponssard J-P, Walker N (2008) EU emissions trading and the cement sector: a spatial competition analysis.Clim Policy 8:467–493

Reinaud J (2004) Industrial competitivieness under the european union emissions trading scheme. Discussionpaper, International Energy Agency

Requate T, Unold W (2003) Environmental policy incentives to adopt advanced abatement technology: willthe true ranking please stand up?. Eur Econ Rev 47(1):125–146

Roodman D (2006) How to do Xtabond2: an introduction to difference and system GMM in stata. Discussionpaper, SSRN

Sijm J, Neuhoff K, Chen Y (2006) CO2 cost pass-through and windfall profits in the power sector. Clim Policy6:49–72

Smale R, Hartley M, Hepburn C, Ward J, Grubb M (2006) The impact of CO2 emissions trading on firmprofits and market prices. Clim Policy 6:31–48

Stavins RN (2008) Addressing climate change with a comprehensive US cap-and-trade system. Oxf Rev EconPolicy 24(2):298–321

Woolridge JM (2009) Introductory econometrics: a modern approach, 4th edn. Southwestern CollegePublishing, Cincinatti, OH

123

Documents

Abatement and Allocation in the Pilot Phase of the EU ETS