GA No.308481 Report on the empirical assessment of monitoring and enforcement of EU ETS regulation Antoine Dechezleprêtre London School of Economics, LSE
Executive Summary This report presents the first empirical evidence to date on compliance and enforcement in the EU Emissions Trading Scheme (EU ETS), based on data from the European Union Transaction Log (EUTL). Despite a gross over-allocation of permits supplied within the EU ETS, we find evidence of substantial non-compliance: between 2005 and 2012, there were 4,496 instances of non-compliance among installation-years, accounting for 5.2% of total installation-years and 3.6% of overall emissions. Compliance has improved from 91% in Phase 1 to 97% in Phase 2. Compliance rates vary substantially across member-states, with some achieving near-perfect compliance (e.g. Austria, Netherlands and Poland) while compliance is well below 90% elsewhere (e.g. Bulgaria, Czech Republic, France). We observe less variation across types of activity. We offer potential explanations for non-compliance behaviour. 2
Table of Contents Executive Summary... 2 1. Introduction... 4 2. Summary of Work Performed... 5 3. Results and Conclusion... 5 4. References... 9 5. List of Abbreviations... 9 3
1. Introduction With the ratification of the Kyoto Protocol in 2002, the members of the European Union (EU) committed themselves to reducing their collective greenhouse gas emissions by 8% below 1990 levels over the period 2008-2012. The EU has since committed to reduce its emissions by 20% before by 2020.1 To achieve these targets cost-effectively, the EU launched the EU Emissions Trading System (EU ETS) in 2005, a cap-and-trade program that covers over 11,000 of the most polluting power plants and industrial installation in the EU. The EU ETS sets a yearly cap on how much emissions to allow, and then divides it up and and allocates a number emissions allowances to each installation. Installations are permitted to buy and sell allowances on the market throughout the year, but by April 30th installations must surrender to the regulator enough allowances to cover their verified emissions during the preceding calendar year. As the program progresses, installations are allocated fewer allowances and they must reduce their collective emissions. The efficacy of this program, or any regulation, ultimately depends on the extent to which installations comply, and on European governments ability to demand compliance when is not forthcoming. But while substantial efforts have been devoted to understanding the movements of the allowance price, the trade flows of allowances, and the EU ETSs impact on emissions, performance, and innovation, little work has been done to understand the most elementary aspect of the programme s implementation: compliance and enforcement. Most analyses of the EU ETS to date have simply assumed universal compliance. Our paper is the first to empirically analyse the compliance behaviour of EU ETS installations based on EUTL data. The objective of this report is to produce descriptive statistics on monitoring and enforcement issues in the EU ETS. Using the EUTL dataset, we identify noncompliant installations during the first and the second phase of the EU ETS and produce statistics by country and by sector on the importance of non-compliance behaviour in terms of the number of installations and of emissions. We also analyse the link between the different monitoring, reporting and verification regulations, compliance behaviour and firm performance. Under Directive 2003/87/EC, penalties for surrendering fewer allowances than verified emissions (known as non-compliance ) include the following: 1) installations must buy allowances to make up the shortfall and surrender them the following year; 2) they are named and shamed ; and 3) they must pay a dissuasive fine for each excess tonne, which was 40/tonne CO 2 in Phase 1 and 100 in Phase 2. With such severe penalties for non-compliance, we expected to find very little evidence of noncompliance. However, this ultimately depends to what extent national governments enforce these penalties. 4
2. Summary of Work Performed We downloaded and cleaned all the data available from the EUTL website on surrendered allowances and verified emissions for all EU ETS-covered installations in phases I and II of the programme and analysed this data. The data collection effort was important as the data is not available in a reader-friendly format from the EUTL website, which is not targeted at academic researchers. The collection and analysis of this dataset was the primary objective of the report so that the target set was achieved. In addition we downloaded, processed and analysed data from the Member country reports also available on the European Commission website. This includes 145 reports for 28 EU countries (83 reports for a given country-year are missing from the website). Collection and analysis of this additional dataset was not an objective of the report, but proved indispensable for a good understanding of the subject. With these two sources of data at hand we started writing a paper on enforcement and compliance in the EU ETS, which is still in progress. 3. Results and Conclusion Our data comprises covers 13489 installations over 8 years (Phases I and II of the programme). However, 20615 data points are missing on either verified emissions or total allowances surrendered for a given year. We are thus left with 87297 companyyear observations. There are 4495 company-years where verified emissions are lower than allowances surrendered. However, before assessing non-compliance we need to take into account two additional facts: (i) (ii) there are also 5179 company-years where verified emissions are higher than allowances surrendered, in other words where installations overcomply. In some cases installations gave in fewer permits than their actual emissions because they had given in too many permits in the past; installations that do not comply in one year are supposed to surrender more permits the next year in order to make up for the permits that they would have had to surrender the year before. If installations do not surrender permits to make up for the shortfall, then under-compliance carries over to the next year. Taking this into account, we conclude that there are 4565 instances of noncompliance, accounting for 5.2% of total installation-years on which data is available and 835 million permits (each permit corresponding to a tonne of CO2 emissions), which represent 5.1% of overall distributed allowances. Therefore we find evidence of substantial non-compliance in the EU ETS. It is surprising to find such high levels of non-compliance despite the gross over-allocation of permits supplied within the EU ETS and the low price of carbon that has prevailed on the market. We observe 91% compliance during Phase 1 and 97% compliance during Phase 2 (see Figure 1). However, the figure in Phase 1 is explained mostly by noncompliance during the first year: in 2005, only 78% of installations were compliant. The results suggest compliance increased since the beginning of the EU ETS. These results may be expected: first, at the beginning installations may have been 5
unfamiliar with a newly imposed carbon constraint as well as the administrative procedures involved in monitoring, verification and reporting; second, a range of revisions in the regulatory system of the EU ETS have been implemented over the years, facilitating monitoring and compliance. The increased severity in Phase 2 is another likely explanation for the improved compliance. Figure 1. Compliance and Non-Compliance during Phases 1-2 of the EU ETS. Note: Total number of installations reduced in 2012 due to data availability. Compliance levels vary across EU member-states (see Table 1). Eight countries Austria, Latvia, Greece, Poland, the Netherlands, Luxembourg, Slovenia and Belgium all have higher than 99% compliance in Phases 1 and 2. On the other hand, there are several member-states that exhibit significant non-compliance. Italy, France, Czech Republic, Bulgaria, Cyprus, Malta and Slovakia all have compliance rates below 90%. One potential explanation for the observed variation in compliance is differential enforcement. Crucially, enforcement (e.g. imposing penalties) is implemented at the member-state level. It is likely that countries with low levels of enforcement similarly have low compliance levels. However, at present, data on enforcement is lacking; our current research efforts are directed at quantifying enforcement in order to provide further insights into this issue. 6
Table 1. Compliance and Non-Compliance across Member-States during Phases 1-2 of the EU ETS. Note: Data for Iceland unavailable. Country % Compliant % Non-Compliant Installation-Years Austria 99.54 0.46 1,728 Belgium 99.04 0.96 2,404 Bulgaria 66.08 33.92 802 Cyprus 28.57 71.43 91 Czech Republic 88.78 11.22 3,146 Denmark 91.29 8.71 3,018 Estonia 95.20 4.80 375 Finland 97.22 2.78 4,751 France 88.86 11.14 8,242 Germany 95.18 4.82 13,656 Greece 99.38 0.63 1,120 Hungary 93.56 6.44 1,787 Ireland 97.63 2.37 844 Italy 89.52 10.48 7,948 Latvia 99.43 0.57 696 Liechtenstein 90.00 10.00 10 Lithuania 97.82 2.18 825 Luxembourg 99.11 0.89 112 Malta 25.00 75.00 16 Netherlands 99.16 0.84 2,491 Norway 96.34 3.66 574 Poland 99.17 0.83 6,522 Portugal 97.94 2.06 1,750 Romania 97.60 2.40 1,333 Slovakia 87.70 12.30 1,293 Slovenia 99.06 0.94 743 Spain 96.84 3.16 7,976 Sweden 98.61 1.39 5,884 United Kingdom 98.46 1.54 7,185 Total 94.85 5.15 87,322 7
Interestingly, there is little variation across sectors in terms of compliance rates (see Table 2). This is surprising as some sectors have higher potential for low-cost abatement than others. Table 2. Compliance and Non-Compliance across Sectors during Phases 1-2 of the EU ETS. Sector % Compliant % Non- Compliant Cement Clinker 94.04 5.96 4346 Ceramic Products 94.01 5.99 8460 Chemicals 93.75 6.25 16 Coke Ovens 96.39 3.61 166 Combustion Installations 95.30 4.70 57373 Glass 94.48 5.52 3334 Iron or Steel 94.07 5.93 1873 Metal Ore 96.37 3.63 193 Mineral Oil Refineries 93.39 6.61 1226 Non-ferrous Metals 85.71 14.29 7 Other 94.70 5.30 3909 Pulp and Paper 93.22 6.78 6419 Installations Covered We find non-compliant installations to be relatively larger than compliant installations in terms of verified emissions or allowances distributed. To the extent that these measures accurately represent installation size, the results seem counter-intuitive, as larger installations should be better able to deal with the fixed administrative costs associated with reporting and verification. What does the observed non-compliance imply in terms of penalties? The official fines for non-compliance were 40 per tonne per year in Phase I and 100 per tonne in Phase II. Table 3 shows that installations should have had to pay around 56 billion euros in fines for non-compliance if the law had been enforced. If we excluded all installation-year observations where data on compliance is missing in the preceding year, so that the installation might well have surrendered the missing permits that year (a very generous assumption), this leaves us with 3200 instances of noncompliance, representing 500 million permits and a total fine of 26 billion euros. Note that of these 26 billion euros, 15 billion correspond to the first year of the system, where some tolerance might have been in place. 8
Table 3. Theoretical fines implied by observed non-compliance Year Number of noncompliant installations Number of permits not surrendered Theoretical fine (euros) 2005 2,291 380759618 15,230,426,874 2006 323 31508650 1,260,343,726 2007 240 44020080 1,760,806,560 2008 335 90214830 9,021,471,945 2009 350 60361350 6,036,148,300 2010 320 66445760 6,644,584,640 2011 335 85847435 8,584,735,125 2012 371 75811624 7,581,156,093 TOTAL 4,565 834,969,347 56,119,673,263 Member states are supposed to report in their annual reports to the Commission the fines that have been collected in this respect. From these reports we were able to establish that a maximum of 45 installations were fined for infringements on national provisions (which include excess emissions). We can thus confidently state that these theoretical fines have not been collected so far. This might explain why we still observe some non-compliance in the system at the end of Phase II, 8 years after the launch of the EU ETS. Our results suggest that more can be done to step up enforcement. Thus, they help achieve the goals of the Work Package by pointing to a shortcoming of the EU ETS that should be addressed in the upcoming EU ETS reforms. 4. References 5. List of Abbreviations EU ETS European Union Emissions Trading Scheme EUTL: European Union Transaction Log 9