Figure 4 shows a considerable variation of the EUA price over time. To take rational investment decisions and, thus, to reflect long-term MAC, market participants need to be able to forecast these price developments. Many factors impact the EUA price including international energy prices, factor cost, as well as regulatory changes in the EU ETS itself and in complementary policies affecting allowances supply and demand. These include, e.g., renewable promotion and energy efficiency policies (see e.g. Hintermann et al., 2016). Thus, there are large uncertainties regarding the predictability of the carbon price signal. Also, the causes of past price changes cannot be firmly determined. Yet, recent changes, such as the MSR or the more stringent reduction path for the fourth trading period seem to have strengthened the confidence in thesystem. Nevertheless, experts emphasize that there are still large uncertainties about future developments of the EUA price, resulting in a wide spread set of predictions, which heavily depend on the underlying assumptions on policy developments and regulations as well as on the pace and extent of emissions reductions due to the uptake of renewable energies andthe phase out of electricity from coal. The range of price forecasts from market analysts is therefore quite large (e.g. varying between 22 € and 65 € for 2020 and 27 € and 60 € for 2030). 9
lower VAT or a subsidy to public transport. They find that the reform is regressive, but point out that regional differences are more important than differences in income. Bach et al. (2001) carry out a broad-based analysis of theGerman environmental fiscal reform of 1999, which increased taxes on fossil fuels andelectricityand in turn lowered social security contributions (SSC). As part of a larger fiscal reform, income tax and child benefits were also adjusted. Overall, they find only moderate effects. When taken in isolation, the introduction of the environmental tax is regressive, looking at the whole reform package, most households are better off than before the reform. Interestingly, there exist a number of studies for European countries (Labandeira and Labeaga 1999 for Spain; Tiezzi 2005 and Martini 2009 for Italy; Symons et al. 2002 for five European countries), which find that carbon taxes in those countries are not necessarily regressive, even before revenue is returned to households.
to other reforms.
In any case, our empirical results have to be taken with care since our carbon pricing policy designs are very stylized. In practice, it would be very hard to monitor how much emissions may be shifted from the ETS to the non-ETS sector or retired due to the price floor. Moreover, there exist huge sectoral differences in abatement costs within the ETS and non-ETS sectors (recall Figure 3, Appendix C). Thus, potential efficiency gains very much depend on whether the additional tax is levied on coal-fired power plants in theelectricity sector or on rubber production plants in the chemical sector. Policymakers may increase abatement targets in ETS sectors that face high marginal abatement costs but relax targets in non-ETS sectors that face low marginal abatement costs which may even result in efficiency losses. Analogously, regarding the efficiency analysis of policy 2 it also depends on which sectors are taxed and thus, how many emission allowances will be retired.
In December 2007, the UN climate conference took place in Bali, where the participating states discussed about further actions after the expiration of the Kyoto Protocol in 2012. They agreed upon the Bali Action Plan, which included that the successor agreement for the Kyoto Protocol would be adopted at the UN climate conference in Copenhagen in 2009 (Skjærseth and Wettestad 2010a: 117). Against this background the Commission‟s proposal for a revision of the EU ETS for its third phase was influenced by the forthcoming conference in three ways (Skjærseth and Wettestad 2010a: 118). First, the EU aimed to use its leadership position to push an ambitious agreement forward (Oberthür and Pallemaerts 2010: 45). To show the world its seriousness, the EU had set the target of reducing greenhouse gas emissions by 20 % until 2020 compared to 1990 levels with the option to increase this to 30 % if other countries would join in (Skjærseth and Wettestad 2010a: 117). Secondly, the Commission aimed to motivate other countries to cooperate by announcing to reduce external credits if an ambitious global deal could not be reached. However, if a global deal could be reached the EU would increase its target to 30 % and would allow half of the additional reduction efforts to be financed by external credits (Skjærseth and Wettestad 2010a: 118). Thirdly, regarding the potential risk of carbon leakage, the Commission postponed the decision about the threatened sectors to the time after the conference in Copenhagen (Skjærseth and Wettestad 2010a: 118). In January 2009, the EU issued its position paper for the climate conference in Copenhagen, which contained the main aim to restrict global warming “to less than 2°C above the pre- industrial level” (European Commission 2009). An important item was the effort to establish an OECD-wide carbon market by 2015 and to reform the CDM market (European Commission 2009).
6 | 16 Another possible cause is the strong production of green energy in Europe. The commis- sion as well as single EU members have formulated goals for renewable energy and have introduced suitable development systems having a regulatory function that overlaps with the emission trade. For example, the fixed energy feed-in tariffs for renewable energy technologies in Germany andthe production of CO 2 -free green energy made possible by those tariffs both lead to a reduced demand for emission allowances in theGerman energy sector. The price for allowances goes down, so that market participants of other sectors or countries can secure the allowances more cost effectively than they could in a situation in which less green energy was produced. Because of the coexistence with emission trading, the promotion of renewable energies merely relocates emissions, it does not avoid them (BMWA 2004).
The economic analysis is based on welfare theory applied to an electricity spot market with short-term marginal cost pricing. The welfare term is defined as the consumer and producer surplus as well as network congestion rents. Themarket price is determined by the intersection of the inverse demand function andthe supply curve defined by the marginal costs of the suppliers. Exchange capacity between different price zones is implicitly auctioned into themarket dispatch to maximize system welfare. This setting represents the prevailing market design in Central and Western Europe. It also indicates that themarket dispatch of the entire system is optimized without considering implications on the national level. As soon as limited inter- zone capacity becomes binding, electricity prices deviate across different national price zones. While additional cross-border capacity is required for the on-going integration of national electricity markets, these investments also affect the national welfare level. Thus, national regulators and TSOs might have second thoughts on investments which provide this additional exchange capacity. The two deviating objectives of integration and national welfare are illustrated in Figure 5.1. The problem formulation separates the decision on transmission investments in the upper level (leader) from themarket dispatch in the lower level (follower) into a bi-level optimization problem. The central planner with the objective of welfare optimization is a special case of investment planning with a single objective. As the leader andthe follower have the same objective value the bi-level model can be simplified to a common linear optimization problem (Kirschen and Strbac, 2004).
When it was launched in 2005, theEuropean Union emissionstradingsystem (EU ETS) was projected to have prices of around €30/ton CO2 and to be a cornerstone of the EU’s climate policy. The reality was a cascade of falling prices, a ballooning privately held emissions bank, and a decade of low prices providing inadequate incentive to drive investment in the technologies and innovation necessary to achieve long-term climate goals. TheEuropean Commission responded with administrative measures, including postponing the introduction of allowances (backloading) and using a quantity-based criterion for regulating future allowance sales (themarket stability reserve); although prices are beginning to recover, it is far from clear whether these measures will adequately support the price into the future. In the meantime, governments have been turning away from carbon pricing and adopting overlapping regulatory measures that reinforce low prices and further undermine the confidence in market-based approaches to addressing climate change. The solution in other carbon markets has been the introduction of a reserve price that would set a minimum price in allowance auctions. Opponents of an auction reserve price in the EU ETS have expressed concern that a minimum auction price would interfere with economic operations in themarket or would be tantamount to a tax, which would trigger a decision rule requiring unanimity among EU Member States. This Article reviews the economic and legal arguments for and against an auction reserve price. Our economic analysis concludes that an auction reserve price is necessary to accommodate overlapping policies and for the allowance market to operate efficiently. Our legal analysis concludes that an auction reserve price is not a “provision primarily of a fiscal nature,” nor would it “significantly affect a Member State’s choice between different energy sources.” We describe pathways through which a reserve price could be introduced.
Thus far, we have implicitly assumed that National Allocation Plans in the Eu- ropean EmissionsTrading Scheme are determined by a central planner, since we only investigated an individual country scenario. However, in the first two trading periods, each member state individually decided on its own allocation plan andtheEuropean Commission only has a veto right if an allocation plan contradicts EU jurisdiction (see Section 2.2). For future periods, thesystem will change. Ac- cording to the recently decided EU climate package, future allocation plans will solely be determined by the EU. 1 At this juncture, the question arises as to whether central planning by theEuropean Commission is really preferable. Kruger et al. (2007) argue that a decentralized simultaneous allocation would not lead to cost efficiency in general, since member states are probably not aware of other states’ abatement possibilities. Another justification for a central planner could be that decentrally determined allocation plans by national regulators provide incentives to behave strategically. Thereby, different incentives for strategic behavior are imaginable. First, member states can try to influence the emission permit market. A state whose Tr-sector is a net buyer of permits on the EU permit market is, of course, interested in a lower permit price; whereas a permit selling state prefers a higher permit price. Second, since all kinds of trade policies are abolished on the integrated Europeanmarket, member states could try to substitute absent trade policies through a strategic allocation of emission permits.
2 1 Introduction
Currently, about 17 percent of global greenhouse gas (GHG) emissions are covered by emissionstrading systems that have either already been implemented or are scheduled for implementation. The EU EmissionsTradingSystem (EU ETS) is still the largest of its kind worldwide. It covers roughly 40 percent of the EU27’s GHG emissionsand is considered to be the EU’s most important climate policy instrument. Due to an annual linear reduction factor (LRF), no further allowances will enter themarket beyond a certain point in time. Calculations based on the current LRF of 2.2% p.a. indicate the year 2057, or shortly after, as the expected cut-off date. In the likely case of a more ambitious EU climate target for 2030, the LRF would increase accordingly, so that the trajectory of newly issued allowances would reach the zero line some years earlier (European Commission 2018a). Given the recently agreed EU target of reaching net zero GHG emissions by 2050, there are even expectations that the installations covered by the EU ETS will generate net negative carbon dioxide (CO 2 ) emissions from 2045 onwards (European Commission 2018b).
The main market focus of course was on the price. In the early months, carbon prices rose steadily, tracking the rising gas price that determined the cost of switching away from coal in power sector generation. As gas prices continued to soar, the CO 2 price broke free from this marker and oscillated in the range EUR 20–25/tCO 2 for much of the year (Figure 1). From several perspectives, 2006 was the defining year for the EU ETS. It started with prices for phase I (2005–07) emission allowances reaching levels higher than anyone predicted, peaking at EUR 30/tCO 2 , whilst governments confidently issued draft National Allocation Plans (NAPs) for how they intended to allocate allowances for phase II, the Kyoto period of 2008–12. The year ended with phase I prices sinking close to zero, and several countries threatening to take legal action to overturn theEuropean Commission’s rejection of almost all the submitted NAPs as inadequate. It was certainly a year of vast learning – as befits the middle of the first, learning, period of a major new system.
Table 2: Percent of Firms that trade at the exchange with their share of emissions in theGerman EU ETS andthe corresponding share of traded permits. Costs are displayed on the right hand side with total costs in million EUR and average costs per firm in thousand EUR.
Table 2 shows that rising fixed costs and fees have almost no effect on the number of firms that trade at the exchange andthe share of traded emissions at the exchange. This is because the threshold trading volume ¯ s in Table 1 is unsensitive for changes in fixed costs m and fees f when reproducing the empirical pattern. Table 2 also shows that the larger share of traded permits is traded at the exchange (86% of about 154 million permits) and that these large volumes are traded by a small number of firms (about 5% out of n=819). Total costs and average costs per firm are sensitive to changes in fixed costs m and fees f . Since the fees were derived so that the empirical trading patterns in the small sample could be mimicked (Table 1) for given fixed costs m, fixed costs are the main driver of total and average costs in Table 2. Since a trading account for carbon permits, for example at theEuropean Energy Exchange (EEX), comes at costs of about EUR 20,000 and additional internal costs (personal costs, hardware, training of staff etc) apply, fixed costs of m = 50, 000 seem to be a good guess for actual fixed costs for exchange trading. At m = 50, 000, total costs amount to EUR 6 million and average costs per firm are EUR 7,300 with a quiet large standard deviation of EUR 13,000 8 .
Procurement of balancing reserves is done by various mechanisms, including legal supply obligations, bilateral negotiations, or more formalized balancing markets for periods ranging from days to years . German balancing reserves are procured through Regelleistung.net, an online platform now used by various European TSOs. The design of balancing market auctions is prescribed by the regulator. In the past few years, the duration of contracts was shortened from months to weeks and days, and minimum bid sizes were reduced, leading to a strong increase in participation and competition. Both capacity and energy are remunerated on a pay-as-bid basis, andthe award criterion was changed in 2018 to also include the energy price. Marginal pricing and free bids are scheduled to be introduced in 2020.
brokers also affects transaction volumes: From 2013 to 2016, 55.7% of allowances in 67.6% of transactions involving the EEX Auction Delivery Account were directed towards these bidders. However, the average transaction volume of 347,500 allowances was considerably lower than that of installation holders, which reached 575,500 allowances. With regard to the average volume per account holder, in turn, banks and brokers are ahead by a considerable margin, reaching 434,000 versus 368,700 allowances. Considering these results, two hypotheses can be formulated as to the role banks play in the auctioning of allowances. First, it is probable that banks predominantly act as brokers or intermediaries by placing bids on behalf of installation operators lacking the skill or infrastructure necessary to participate in allowance auctions. This would explain the exceptionally low number of installation operators receiving transactions directly from ICE or EEX. Second, banks other market participants not directly involved in the ETS may be acting on their own account, acquiring allowances at auctions with the purpose of trading. According to Art. 18 of the Auctioning Regulation (European Commission, 2010), both alternatives are legally viable. However, a transaction-level analysis of accounts operated by two major players – Deutsche Bank and Citigroup Global Markets – fails to provide useful insight on this issue. In concrete terms, this involved searching for transactions with identical volumes in temporal proximity to the transfers issued by either ICE or EEX. Unfortunately, I was unable to identify conclusive patterns using this relatively straightforward approach. Hence, further research using a more sophisticated, algorithm based method is needed to shed light on the business practice of banks or financial institutions involved in the auctioning of allowances.
First, a promotion of RES increases uncertainty. The substitution of fossil by renew- able electricity generation aects the certicate price and thus the potential of the ETS to reduce CO 2 emissions. In order to set an optimal objective for emission reduction,
the regulator requires to perfectly foresee and consider the development of renewable electricity generation. A non-optimal emission cap causes additional costs. However, this argumentation is not totally valid anymore in case of the EU after a mechanism to withdraw excessive certicates from themarket was installed (European Parliament and Council of the EU, 2018). Moreover, Schäfer (2018a) suggests the introduction of a unilateral exible cap which allows to decouple the promotion of renewable energy from the ETS. Then a substitution of fossil electricity generation by subsidized RES does not aect the cap anymore.
Further, Perino (2015) differentiates between sector-specific and general climate cam- paigns. The former are “campaigns directly targeting specific products discouraging their consumption” (p. 473), so “the purpose of such campaigns is to reduce emissions in the targeted sectors, that is, those that experience a reduction in demand” (p. 476). Therefore, sector-specific campaigns aim at encouraging individuals to reduce their demand for products and services that are either emission-intensive in production or provision or dur- ing or after the use phase, e. g. electricity, meat, petrol or flights. Instead of aiming at reducing demand of emission-intensive products, it is also possible that campaigns are designed to promote alternatives to these and hence rather aim at increasing the demand for less emission-intensive products, e. g. “green” electricity tariffs, plant-based food, bicycles or train journeys. However, since both options aim at changing demand in the same direction and might induce similar results, both are referred to as “demand reduc- ing” in the following. Next to these, general climate campaigns “aim at increasing the intrinsic motivation of consumers to reduce their carbon footprint generally and leave it to consumers how they want to achieve this” (p. 473). In this case, consumers have to rely on information about the carbon footprint of products and services which is often calculated via an environmental impact assessment or a life-cycle analysis using stand- ards like ISO 14040, ISO 14044 or ISO 14067 (International Organization for Standardization, 2018, 2020a, 2020b). However, both types of campaign recommenda- tions are given and carbon footprint specifications are calculated without taking environ- mental policy instruments or regulations like cap-and-trade schemes or carbon taxes into account (Perino, 2015). Above, Rosendahl (2019a) differentiates between a temporary and a permanent abatement or demand reduction, whereas temporary means the demand change is only in one year and permanent stands for a demand change that lasts from the year when the first demand change is made onwards. According to this two differentia- tions, the focus in the following will be on sector-specific climate campaigns with a tem- porary effect, since Perino (2015) focused on them as well andthe goal is to update his findings. However, possible effects of general campaigns and of campaigns that lead to permanent behavior changes will be discussed later on.
Furthermore, some ‘dirty’ ﬁrms may have faced such high environmental costs from the EU ETS that they had to exit themarket. Similarly, multinational corporations with production units could delocalize their production outside the EU. In the end, only ﬁrms who were competitive in a clean environment could have kept their business running. In the aggregate, this would result in a more productive and cleaner business environment. This may explain ﬁnding no negative competitiveness eﬀects on ﬁrms that have stayed in themarket. If ﬁrms or subsidiaries exit themarket in order to relocate to places where environmental regulation is less restrictive ( ‘pol- lution havens ’), this is called the Pollution Haven Hypothesis (PHH; e.g. see Cole, 2004; Eskeland & Harrison, 2003; Kozluk & Timiliotis, 2016; Millimet & Roy, 2011; Wagner & Timmins, 2008; Yoon & Heshmati, 2017). In the case of the EU ETS, the PHH is supported if signi ﬁcant evidence of carbon leakage attributable to the EU ETS is established. If ﬁrms were to start emitting more once relocated, we could observe an increase in total emissions worldwide, making the EU ETS an ine ﬀective mechanism. No empirical evidence of carbon leakage or ﬁrm closures attributable due to the EU ETS has been documented so far. Preliminary results using emissions data by Dechezleprêtre, Gennaioli, Martin, and Muûls (2014) and Wagner et al. (2014) ﬁnd no supportive evidence for carbon leakage within companies which have non-treated plants during Phase II. Besides emissions, indirect measures of carbon leakage may be used. While short-term leakage is usually detected through increased imports, long-term production relocation may be analysed through outbound foreign direct investments (FDI, see Koch & Basse Mama, 2016). Several trade ﬂow analyses show that the carbon price level did not lead to any signi ﬁcant carbon leakage in theEuropean primary aluminium sector (Sartor, 2012), in the cement and steel sectors (Boutabba & Lardic, 2017; Branger, Quirion, & Chevallier, 2017) or in manufacturing sectors 11 (Naegele & Zaklan, 2017). Focusing on Germanand Italian multinationals, respect- ively, Koch and Basse Mama (2016) and Borghesi, Franco, and Marin (2016) show that the EU ETS did not lead to relocation through outbound FDI for the average ﬁrm. However, both studies reveal that particular sub-groups of enterprises did signi ﬁcantly react to the EU ETS stringency. Still, overall, the scarce evidence so far shows no evidence of carbon leakage for the average ﬁrm, and thus contradicts the PHH.
In this section, we analyze the effects of the EU-ETS on household welfare if the revenue obtained by theGerman government through the auctioning of emissions allowances is returned to households via lump-sum rebates and, alternatively, as a reduction in social security contributions (SSC). At a carbon price of €25, the estimated yearly revenue is €7.5 billion, equivalent to the amount of permits theelectricity sector requires times a carbon price of €25. With this amount the social security contributions rate could be reduced by 0.8 percentage points or provide a lump-sum transfer of €94 per person in the household. 10 As a point of comparison, as a result of the environmental fiscal reform, theGerman government generated revenues of roughly €17 billion per year (€9 billion from households directly and another €8 billion from industry) and was able to reduce SSC by 1.6 percentage points (Deutscher Bundestag 1999).
A few very recent studies aim to contribute empirical evidence to the academic and public debates by investigating the causal effects of the EU ETS on theemissionsandthe economic performance of regulated firms. These studies exploit treatment variation that results from the inclusion criteria of the EU ETS. Since only large emitters are regulated, there are regulated and unregulated firms within narrowly defined industries that can be compared. The empirical evidence on the impact of the EU ETS on firm-level emissionsand emission intensity is mixed. Studies using data from Germany, France, and Norway suggest that the EU ETS significantly reduced greenhouse gas emission (Petrick and Wagner, 2014; Wagner, Muˆ uls, Martin, and Colmer, 2014; Klemetsen, Rosendahl, and Jakobsen (2016)). 5 Jarait˙e and Di Maria (2016) do not find that the EU ETS significantly decreased firm-level emissions in Lithuania, but they find a significant negative effect on emission intensity. So far, there is no evidence that the EU ETS had a significant negative effect on indicators of economic performance. In contrast, Petrick and Wagner (2014) find a positive effect of the EU ETS on the revenues of regulated firms in Germany. Klemetsen, Rosendahl, and Jakobsen, 2016 find a positive effect on value added and labor productivity of regulated firms in Norway. Calel and Dechezleprˆ etre (2016) investigate the effect of the EU ETS on patenting. Their findings support that the EU ETS increased the number of low carbon patents developed by regulated firms. Bushnell, Chong, and Mansur (2013) examine how the EU ETS affects daily stock returns of European firms. They show that low allowance prices are associated with low stock prices for firms in both carbon andelectricity intensive industries. Their results
Supponen ( 2011 ), on the other hand, argues in favor of splitting Europe in further bidding zones better reflecting congestion in the network within countries in order to improve investment signals for (interconnec- tor) transmission capacity. Using a six-node demonstration network, Oggioni and Smeers ( 2013 ) show that the configuration of bidding zones and especially the determination of NTCs between the zones are crucial for the efficiency of a zonal pricing design, like theEuropeanmarket coupling. Burstedde ( 2012 ) analyzes potential bidding zones for theEuropeanelectricitymarket. The paper clusters nodes in the network by locational marginal prices using cluster analysis. Dispatch, re-dispatch, and total system costs are calculated for different zone configurations. A nodal pricing model serves as benchmark. Results show that a zonal market configuration only leads to a small increase of total system cost compared to nodal pricing. With the right choice of NTCs to represent scarcity signals for transmission a better ex-ante market dispatch is reached and less requirements for re-dispatch occur. Breuer et al. ( 2013 ) and Breuer and Moser ( 2014 ) use a similar methodology for the delimitation of bidding zones. Clustering nodes with similar prices to a different number of zones they find that about 14 zones for Europe would be optimal regarding the trade-off between network security andmarket efficiency, on the one hand, andthe stability of bidding zone delimitation, on the other hand. Due to the ever changing nature of theelectricitysystem with the ongoing commissioning and decommissioning of plants and lines, the delimitation of zones should change frequently, which does not favor market participants.
This paper presents and reflects the discussion surrounding the current performance of the EU ETS and specifically the persistently low price of EUAs. It analyses the performance of the EU ETS in terms of environmental effectiveness and cost-effectiveness, and examines the reform options additionally in terms of political feasibility. Our analysis concludes that the environmental effectiveness of the EU ETS is given (in fact the emission target has been overachieved), but that the EU ETS lacks dynamic efficiency. The EU Commission has suggested addressing this problem by introducing a reform that manipulates the supply of EUAs, i.e. theMarket Stability Reserve. However, we show that this reform proposal does not address the problem of dynamic efficiency, mainly because the interplay between the magnitude of the EUA surplus andthe EUA price formation seems incomprehensible from an inter-temporal perspective. It also fails to address the problem of overlapping policies arising from the existence of supplementary policy instruments at the Member State level that could undermine the overall performance of the EU ETS. By contrast, our analysis clearly shows that instead of a narrow reform of the EU ETS focusing on the EUA surplus, a comprehensive reform addressing a series of aspects of carbon pricing is required. This includes (i) setting a price collar within the EU ETS, (ii) expanding the EU ETS to other sectors (e.g. transport, buildings) (iii) addressing additional market failures through policy instruments in addition to carbon pricing and (iv) addressing the possible problem of carbon leakage by expanding the group of countries that adopt comparable GHG prices.