On the C[O.sub.2] Emissions Determinants During the EU ETS Phases I and II: A Plant-level Analysis Merging the EUTL and Platts Power Data.

• Author: Cheze, Benoit

1. INTRODUCTION

   Has the European Union Emissions Trading Scheme (EU ETS) effectively reduced C[O.sub.2] emissions? At the current price of 22.82 Euro per ton of C[O.sub.2] for futures traded on the ICE ECX, the aim of this scheme, which was set up in 2005, is to reduce C[O.sub.2] emissions in Europe by setting emissions caps for over 11,700 installations (1) which are required to return a volume of allowances that corresponds to their verified C[O.sub.2] emissions for each annual compliance assessment. The EU ETS is in force in 31 countries (2) and covers over 45% of their overall greenhouse gas (GHG) emissions. Phase II lasted from 2008 to 2012. This article aims at studying the determinants of the C[O.sub.2] emissions during 2005-2012 (first and second phases) of the power plants regulated by the EU ETS. Although Phase III of the EU-ETS has been practiced for several years until 2020, this article focuses exclusively on ex-post econometric analysis, hence on Phases I-II only that have been completed.

   The European Commission stated in its report on the operation of the EU ETS in November 2012 that "the EU ETS is facing a challenge in the form of an increasing allowance surplus, primarily because the economic downturn has reduced emissions by more than was expected" (3) It is indeed likely that the slowdown in economic activity within the European Union did have an impact on the fall in C[O.sub.2] emissions, but can we argue that the downturn was the main reason or even the only reason for that fall?

   Economists typically weigh several criteria for gauging the performance of an emissions trading system. Around the notion of economic efficiency, environmental effectiveness refers to the reduction of emissions that is attributable to the instrument, whereas cost-effectiveness refers to the marginal abatement cost curve. Regarding transaction costs, public finance and administrative issues are largely ignored in empirical applications.

   Besides the 2008 economic recession, the environmental effectiveness of the EU ETS may be endangered by overlapping climate policies. We distinguish two kinds of overlaps. On the one hand, several environmental regulation tools coexist with emissions trading: i) the EU Commission Climate-Energy Package, ii) renewable energy deployment objectives and iii) specific sectors regulations. On the other hand, this policy mix is implemented at both the regional and national levels.

   Indeed, factors other than the economic crisis could also have played a role, especially the actual efforts made to decarbonize the economy, and increasing renewable energy's share in the energy mix. Indeed, the commitments made at the European level, which resulted in the so-called "20-20-20" targets, (4) were implemented via a series of directives, including the directives on renewable energy and energy efficiency, which were combined with national policies. These commitments were reflected by the notable development of renewable energy in most States. (5) In which case, can we estimate to what extent these efforts contributed to reducing C[O.sub.2] emissions? Likewise, we need to ask whether changes in the price of energy affected C[O.sub.2] emissions or whether the allowance system and specifically the carbon "price signal" that it reflects, effectively played a role by encouraging fuel-switching in energies and investments in technologies that emit less carbon.

   The power sector is also exposed to different kinds of energy or environmental policies that also impact fossil fuel power plant emissions levels. On top of the carbon price that was established in 2005, national policies to develop renewable energy are widespread in the European Union. Since 2009, national targets are consolidated in a directive at the European level, and the Member States established action plans to reach the desired development in renewable energy. (6) According to them, electricity from renewable sources will reach 33% of the total final electricity consumption at the European level in 2020, when it was only 15% in 2005. To reach their objectives, many Member States have put in place deployment policies such as feed-in tariffs or "green" certificates (Ringel, 2006) that were successful in channeling investments in renewable energy production without any connection to the C[O.sub.2] price level. Other environmental command and control policies are also applied in the European power sector, like the Large Combustion Plant Directive (LCPD) that limits the use of some power plants since 2008. We thus can take advantage of the data provided in the EU Transaction Log on power plants participating in the EU ETS, to evaluate the impact regarding C[O.sub.2] emissions of the carbon price, but other complementary policies that affect emissions levels.

   We choose to focus our analysis on the power sector for various reasons. It is the largest sector in the EU ETS regarding C[O.sub.2] emissions. Half of the allowances were allocated to power or Combined Heat and Power (CHP) plants (see Figure 5 in the Appendix). It differentiated from the other sectors also because, since 2005, it is the only industry that as a whole was short in European Union Allowances (EUAs), i.e., its free allocation of EU allowances was lower than the amount of C[O.sub.2] it emitted. The Member States have anticipated this. It can be explained by the perception that cheaper abatement options exist in the power sector than in other industrial sectors, and the low risk of carbon leakage in power production (Ellerman and Buchner, 2008; Monjon and Quirion, 2011). It has led power producers to include the carbon price in their operating decisions.

   Technical power plants' characteristics but also economic and energy market conditions should influence the C[O.sub.2] emissions of power plants. However, the magnitude of the influence of these C[O.sub.2] emissions determinants also seems to depend on the power plant under consideration, which varies widely among the EU ETS. To take into account the heterogeneity of installations, the role played by these variables on the C[O.sub.2] emissions of power plants concerned by the EU ETS is estimated using panel-data econometrics. The focus of this article is to provide quantitative answers to these questions, based on panel data econometrics for the EU 27. Based on an original merging methodology of the EUTL and Platts power data, we attempt to link C[O.sub.2] emissions with a series of explanatory variables that have an impact on emission trends and to gauge their relative contributions. Cross-sectional units of the panel-data sample correspond to the 1,453 electricity generation plants running on fossil fuels in Europe. We study the different factors reducing C[O.sub.2] emissions and evaluate the relative contribution of the EU ETS (carbon price), renewable energy deployment, command and control directives on local pollutants, and the economic crisis on the abatement in the European power sector.

   The gist of the results may be summarized as follows. Dynamic panel data regressions investigate the driving factors of C[O.sub.2] emission reductions of power plants regulated under the EU ETS. From the list of potential explanatory variables--including economic activity, energy prices, power plant technology, and climate and energy policies--the 2008/09 recession stands out as the primary driver of emission reductions. In contrast, the EU ETS carbon price appears to have no consistent effect. Based on these findings, we may cautiously suggest that the EU ETS has been inefficient regarding inducing emission reductions in the period 2008-2012.

   The remainder of the paper is organized as follows. Section 2 presents the literature review. Section 3 details the methodology. Section 4 contains empirical results and policy implications. Section 5 concludes.

2. BACKGROUND

   As seminal papers in this strand of literature, Ellerman and Buchner (2007, 2008) make use of preliminary verified emissions and allowance allocation data to diagnose the extent of "over-allocation" during the 2005-06 period. Such an early empirical evaluation of the explanatory factors of C[O.sub.2] emissions in the power sector has become a highly cited paper in the environmental economics literature. (7)

   Anderson and Di Maria (2011) provide another ex-post evaluation of phase I by resorting to dynamic panel data modeling. Based on emissions data from the Community Independent Transactions Log (CITL) and Industrial C[O.sub.2] emissions from Eurostat (at the aggregated country level), their analysis hardly establishes that abatement occurred over 2005-2007. This view is shared by most studies which reflect on this trial period of the EU ETS, gearing towards the conclusion that no abatement would be indicated (Convery and Redmond, 2007). Martin et al. (2016) have further extended this argument to 2008-2012; based on a literature review on aggregated industrial data, the evidence of the effect of the EU ETS on emissions of participating firms is not conclusive.

   Plant-level data would provide more insights into the emissions reductions achieved under the EU ETS. Abrell et al. (2011) perform such a task using robust regression, merging CITL and Amadeus data to determine plant-level performance. Overall, they capture 59 % of the total verified emissions between 2005 and 2008. After controlling for various characteristics of firms' performance, their results imply some degree of emissions abatement during phase II. This somewhat indirect finding stems from the fact that emissions reductions were not only achieved by reductions in the economic activity of the firms during the sample period. The authors admit, however, a few caveats in their data treatment procedure, and conclude that their results should be interpreted with caution.

   Additional works exist based on plant-level data. Zaklan (2013) also resorts to CITL and Amadeus matched data to establish plant-level evidence...