The EU sustainable finance taxonomy and its contribution to climate neutrality

ABSTRACT The EU Taxonomy is the first standardised and comprehensive classification system for sustainable economic activities. It covers activities responsible for up to 80% of EU greenhouse gas emissions and may play an important role in channelling investments into low-carbon technologies by helping investors to make informed decisions. However, especially in transition sectors much depends on the stringency of the technical performance thresholds that the Taxonomy applies to economic activities that are not yet ‘green’. This paper shows that for several sectors, the thresholds are not yet on track to support the transition towards climate neutrality. To this end, we analyse a large-scale public consultation with detailed responses to the specific thresholds from a variety of stakeholders. Especially for emission-intensive sectors, two distinct use cases of the Taxonomy can be distinguished: For new investments, criteria should be stricter than for current activities of companies. We also argue that for the sectors not covered by the Taxonomy, there is a need to differentiate between low-emissions activities and high-emission activities, which are incompatible with a low-carbon future.


Background and introduction
The past years have shown an increase in volume of sustainable investment funds and sustainability indices (FNG 2020).Additionally, public actors are setting climate targets and governments are issuing green sovereign bonds.This has led to a multitude of classification systems for sustainability by public and private actors, resulting in a lack of transparency and comparability.The EU Taxonomy for sustainable activities is the first comprehensive science-based classification system designed to help understand whether an economic activity is sustainable.It provides a common definition for around 80 sustainable activities by setting performance thresholds (or technical screening criteria) for these activities.One goal is to make Taxonomy-related disclosure mandatory for large companies in the European Union.Ultimately, this is expected to help channel investments into low-carbon technologies, supporting the EU's transition towards climate neutrality by 2050 (IRENA 2017;European Commission 2018a, 2020a;McCollum et al. 2018;Sweatman and Hessenius 2020).
At the heart of the Taxonomy is the expectation that increased transparency and the disclosure of non-financial information will provide incentives for the private sector to shift towards the production of low-carbon goods and services.So far, carbon emissions are disclosed either due to voluntary initiatives or regulatory standards (Hahn, Reimsbach, and Schiemann 2015).In theory, disclosure of environmental information as mandated by the use of the Taxonomy can improve the information available to investors and consumers.Several authors have looked at the effect of (mandatory or voluntary) disclosure of nonfinancial environmental information such as greenhouse gas emissions on outcomes like firm value, the cost of capital, or environmental performance.Downar et al. (2020) present novel evidence that mandatory emissions disclosure in the UK led to an economically significant reduction of greenhouse gas emissions, without negatively affecting financial operating performance.Tomar (2019) reports similar findings for the US.For voluntary disclosure of carbon emissions and correcting for self-selection bias, Matsumura, Prakash, and Vera-Muñoz (2014) show that higher carbon emissions decrease firm value.Beyond carbon emissions, Chen, Hung, and Wang (2018) demonstrate for Chinese firms that mandatory corporate social responsibility (CSR) disclosure leads to lower levels of local pollution, albeit at the expense of profitability.Plumlee et al. (2015) show that CSR disclosure is linked to firm value both through expected future cash flows and the cost of equity capital, and that the quality of the disclosure affects this association.
More broadly, there is evidence in the literature that environmental ('green') performance is rewarded, since investors and analysts take into account environmental risk factors when making investment decisions and recommendations (Heinkel, Kraus, and Zechner 2001;Sharfman and Fernando 2008).Traditionally, many authors have assessed the relation of nonfinancial environmental, social, and governance (ESG) sustainability performance and stock returns (Konar and Cohen 2001;Derwall et al. 2005;Kempf and Osthoff 2007;Statman and Glushkov 2009), as well as ESG and corporate financial performance (see the literature review by Friede, Busch, and Bassen 2015).Although no clear consensus has been reached, most studies find a positive association of environmental performance and financial performance, which was traditionally explained by a more efficient use of resources by 'greener' firms (Sharfman and Fernando 2008).A separate strand of literature looks more explicitly at the cost of equity and debt capital, evaluating whether the market incorporates sustainability information when pricing risk.Many authors find that firms with a better environmental performance face lower costs of equity, or, conversely, that investors demand a premium for investments in firms with environmental concerns (El Ghoul et al. 2011, 2018;Chava 2014;Ng and Rezaee 2015;Gupta 2018).However, findings for equity do not necessarily translate to the debt market (bank loans, corporate bonds), where findings have been more mixed (Zerbib 2019).One explanation for these mixed findings is that credibility seems to be a key determinant for the existence of a 'green premium' in bond markets (Kapraun and Scheins 2019).
This paper contributes by assessing in which ways the EU Taxonomy may support a transition towards climate neutrality.In a first step, we discuss how the Taxonomy needs to be applied in order to contribute to the goal of a climate neutral economy, as well as different mechanisms how the Taxonomy might lead to a reduction of greenhouse gas emissions.We differentiate between two types of applications (or use cases) of the Taxonomy: On the one hand, the Taxonomy may be used to evaluate the climate performance of a specific investment, such as a new manufacturing plant (project level).On the other hand, the Taxonomy may be used to evaluate a company or a portfolio of assets, for example when providing a loan or equity to a company (firm level).
In a second step, we analyse whether some of the technical screening criteria (thresholds) proposed by the Taxonomy are compatible with a pathway towards climate neutrality and discuss how to complement them with additional instruments.The screening criteria need to be judged in light of the two distinct use cases we discuss: While a threshold may be sufficiently stringent to assess the climate performance of existing assets, the use of this same threshold may not be in line with climate neutrality when evaluating a new investment.For our discussion of the screening criteria, we first describe the economic activities covered by the Taxonomy, and identify gaps of emissions-intensive sectors that are currently not covered by the Taxonomy.Second, we classify economic activities into three types, namely activities that are eligible without a threshold ('green' or 'enabling'), activities that have a threshold with a pathway towards climate neutrality, and activities with a threshold but without a pathway towards climate neutrality.Our assessment is based on an in-depth quantitative and qualitative evaluation of the official public consultation on the interim report of the EU Taxonomy (TEG 2019), as well as a broader literature review.Finally, we discuss options to complement the Taxonomy, in case the use of technical thresholds alone does not incentivise investments into technologies compatible with climate neutrality.
The paper is structured as follows.Section 2 describes the data and methodology.Section 3 discusses two distinct use cases for the Taxonomy, three channels through which the Taxonomy may support the decarbonisation of the economy, as well as its sectoral coverage.Section 4 introduces a classification system for economic activities based on the Taxonomy and discusses performance thresholds for a selection of emissionsintensive sectors.Sector 5 discusses the results and concludes.

Data and methodology
Our analysis builds on an extensive evaluation of the official EU consultation on the TEG interim report released in June 2019. 1In this consultation, stakeholders were asked to comment in detail on each of the activities listed in the Taxonomy.A total of 642 stakeholders took part in the consultation, around half of which are private individuals.We only consider answers from public or private organisations and companies in our evaluation, i.e. a total number of 355 such stakeholders (see Figure 1).
We follow a two-step procedure in our evaluation of the EU consultation.First, we code the responses on the thresholds proposed for the different activities in the Taxonomy, as well as the answers on the choice of metric.This allows us to assess quantitatively the distribution of responses for each activity in the Taxonomy.In a second step, we perform a qualitative analysis of the answers for selected activities, in order to judge the compatibility of the proposed thresholds with the goal of climate neutrality (Section 4).Following this procedure, we carry out an in-depth assessment of a total number of 1672 responses on the different economic activities in the TEG report. 2  The overview of respondents shown in Figure 1 indicates a central challenge of our analysis: The distribution of organisations taking part in the EU consultation is not a random draw from all institutions affected by the EU Taxonomy.Instead, depending on the resources available and the interest in the subject, organisations self-select into participating in the consultation.As can be seen from Figure 1, around 70% of the respondents are either industry associations or private companies, whereas 12% are non-governmental organisations (NGOs).This compares to 50% 'In-house lobbyists and trade/business/professional associations' and about 25% NGOs among the 12,052 registrants in the EU Transparency Register. 3Organisations like NGOs were therefore underrepresented in the public consultation.In order to alleviate this challenge, we scrutinise arguments brought forward in the consultation independent of the number of times they were raised in the qualitative part of our assessment, using also additional literature for this purpose.

Applications and coverage of the Taxonomy
The focus of the Taxonomy is on greenhouse gas emissions.The currently proposed Taxonomy, developed by a technical expert group (TEG) and put into law by EU delegated acts, focuses on activities that contribute to climate change mitigation and adaptation (European Commission 2020b;TEG 2020).In order to be included in the Taxonomy, economic activities need to 'substantially contribute' to climate change mitigation or adaptation, without doing significant harm to people ('minimum social safeguards') and other environmental objectives ('do no significant harm' or DNSH criteria).The activities covered by the Taxonomy are typically characterised by high direct emissions (scope 1 emissions) and thus a large potential for climate change mitigation.Economic activities that are incompatible with net-zero emissions and where technological alternatives exist (e.g. the carbon-intensive burning of fossil fuels for the generation of electricity), are not included in the taxonomy.
The Taxonomy is the basis for several related initiatives under the EU Commission's action plan on sustainable finance (European Commission 2018b).Large companies reporting under the non-financial reporting directive (NFRD) will need to disclose the share of their Taxonomy-aligned activities by 2022.According to the Sustainable Finance Disclosure Regulation (SFDR), financial market participants that offer sustainable financial products need to start disclosing their Taxonomy-aligned activities or investments for different market segments by December 2021.Furthermore, the Taxonomy is to be used as a common definition for green financial products via the EU Ecolabel (Hessenius et al. 2020), as well as the EU Green Bond Standard.With respect to public investments, the climate mitigation criteria of the Taxonomy will be used as a reference for a 37% climate share within the EU Recovery and Resilience Facility, and the Taxonomy's 'do no significant harm' criteria (DNSH), which set requirements for other environmental objectives, will be applied to the entire programme.
At the international level, similar taxonomies are being developed in other countries.The EU has been a driving force behind the International Platform on Sustainable Finance (IPSF), with the aim of sharing good practices and aligning different initiatives globally.A full alignment of the different taxonomies at a global level is desirable, but might be difficult to reach due to different levels of ambition and different beliefs about the technologies that can deliver a climate neutral economy.

Applications and expected impact of the Taxonomy
In principle, there is a multitude of applications for the EU Taxonomy.We categorise these different use cases into two basic types of applications (Figure 2).The first application is for evaluating the climate performance of specific new investments, such as a new power plant, a new production plant or a new building.Examples are potential applications within EU-level or national-level COVID-19 recovery packages, the multi-annual financial framework of the EU or the European Fund for Strategic Investments (EFSI), which is replaced by the InvestEU programme from 2021.The most relevant metric in the case of new investments is the share of Taxonomy-aligned capital expenditure (capex).
A second application is to evaluate the performance of an individual company or a portfolio, e.g. of power plants, car fleets or buildings.Examples are decisions to invest in a company's equity, an infrastructure fund or a real estate fund, to buy bonds issued by a company, or to provide a loan to a company.The relevant metric in this case is the share of Taxonomy-aligned revenue or operating expenses (opex).The Taxonomy-aligned share of opex might also be relevant to judge if climate targets set by individual companies for their respective institutions can be reached.
In principle, different channels are possible through which an increased disclosure of information based on the Taxonomy may support the decarbonisation of the economy (Figure 3).The Taxonomy is first and foremost a classification system, which provides a standardised definition for sustainable activities.There is evidence that the disclosure of nonfinancial environmental information such as greenhouse gas emissions leads to real economic effects such as a reduction of emissions (Tomar 2019;Downar et al. 2020).However, the Taxonomy will only increase transparency if either governments demand information on new investments supported with public funds, or companies report Taxonomy-related information to their stakeholders.Therefore, the Taxonomy is linked to other regulations, such as the non-financial reporting directive (NFRD) for all companies with more than 500 employees and the sustainable finance disclosure regulation (SFDR) for all financial institutions offering financial products.These regulations lead to an increased transparency about the sustainability of companies (with a focus on greenhouse gas intensity) in terms of their goods and services or investment portfolio.
The first and most important link between the Taxonomy and reduced greenhouse gas emissions is a reduction of the cost of debt and equity capital for greener companies (Figure 3).These lower costs of capital may be offered directly by governments or public banks, who offer preferential loans or grants, or who limit public investment programmes to investments that comply with the Taxonomy.Thus, the Taxonomy may be used as a screening tool for public investments or subsidy programmes, in order to support the political goal of a shift towards climate neutrality.Examples of this use of the Taxonomy already exist: The German development bank KfW has launched its SME loan programme 'Klimaschutzoffensive für den Mittelstand' in March 2020, where investments in line with the EU Taxonomy benefit from a 'climate grant' of up to six per cent.Moreover, the COVID-19 recovery programmes financed by the EU's Recovery and Resilience Facility need to comply fully with the do no significant harm criteria of the Taxonomy.Additionally, a climate share of 37% of the total investment of €672.5 billion needs to be reached.For these investments, the climate mitigation thresholds of the Taxonomy need to be applied as screening criteria.Finally, the European Investment Bank has recently pledged to increase the climate share in its financing activities to 50% by 2025, using the EU Taxonomy as a main reference point (EIB 2020).
Firms with a high share of Taxonomy-compatible assets or revenues may also indirectly benefit from lower costs of capital due to a different risk assessment of capital markets.The literature provides some evidence that firms with better environmental performance face lower costs of equity than firms with environmental concerns (El Ghoul et al. 2011, 2018;Chava 2014;Ng and Rezaee 2015;Gupta 2018).While the effects are less clear-cut for debt capital, Kapraun and Scheins (2019) show that credibility is an important factor for preferential financing conditions for green bonds.The EU Taxonomy is likely to provide the necessary credibility.Therefore, it can be expected that companies with a higher share of Taxonomy-aligned activities may benefit from lower costs of capital.Linking the European Green Bond Standard to the Taxonomy can strengthen this channel (Figure 3).
Second, a high share of Taxonomy-aligned activities may also lead to better financial performance and higher firm value, thus incentivizing the management of companies to reduce emissions.Firms with a better environmental performance may improve financial performance via a more efficient use of resources (Sharfman and Fernando 2008).Moreover, Downar et al. (2020) hypothesise that stakeholder pressure plays an important role in driving companies disclosing their carbon emissions to reduce these emissions.Such stakeholder pressure from customers could have consequences for the sales of a company, if customers have a higher willingness to pay for companies perceived as greener, for example as a result of greater customer loyalty.In that sense, improving the environmental performance to achieve a higher share of Taxonomy-aligned activities may also be in the interest of shareholders and positively influence firm value, as Matsumura, Prakash, and Vera-Muñoz (2014) demonstrate for firms with high carbon emissions.This would provide additional incentives to decrease carbon emissions, as management bonuses are often linked to the stock price of a company.
Firms may also use this channel to make credible claims regarding corporate climate commitments, such as a transition to climate neutrality.Initiatives in the field of corporate claims already exist, for example the science-based targets initiative (SBTi).The Taxonomy could be used to make these claims more quantifiable and comparable across firms, for example by reporting the share of Taxonomy-aligned economic activities.
Third, if firms start to measure their absolute or relative CO 2 emissions (currently the main metric in the Taxonomy), this may lead to reductions of these emissions by addressing internal information deficits and providing the basis for firm-level decarbonisation strategies.Other programmes such as energy audits have been shown to have tangible effects by overcoming information barriers, especially in small firms (Kalantzis and Revoltella 2019;Schleich and Fleiter 2019).In the case of the Swedish energy audit programme, more than 50% of the suggested energy efficiency improvements have been implemented by firms (Backlund and Thollander 2015).One reason might be that new profitable energy-saving investments can now be detected.Another reason might be, that based on emissions data, an increasing number of companies are starting to set emission targets, in line with international climate targets (see Giesekam et al. 2021 for an assessment of companies joining the science-based targets initiative) and to link management remuneration and reward systems to reaching these emission targets.Stakeholder pressure to reduce emissions, be it externally (from customers and investors), or internally (employees), may support setting internal management targets.

Sectoral coverage and scope of the Taxonomy
Currently, the Taxonomy for climate mitigation covers seven main economic sectors, namely agriculture and forestry, manufacturing, energy (electricity, gas, steam and airconditioning supply), transport, buildings, water, waste and sewage remediation, and information and communication technologies (ICT).At NACE level 1, 4 the Taxonomy covers economic activities within all major greenhouse gas emitting sectors (approximately 90% of direct greenhouse gas emissions).Within these broad economic sectors, not all economic activities are covered by the Taxonomy.For a better estimate of the share of total emissions covered, it is therefore necessary to consider a more refined NACE level classification.With a more granular sectoral focus (at NACE level 2), the share of emissions of the sectors that include economic activities covered by the Taxonomy declines to 80%, mainly due to the exclusion of certain transport and manufacturing activities.This is still an upper bound, as not all activities at sub-sector level are covered by the Taxonomy, such as air transport (3.6% of direct greenhouse gas emissions) Interestingly, the sectors addressed by the Taxonomy are responsible for a much lower share of employment and gross value added (GVA) than emissions: The same sectors responsible for up to 80% of EU emissions have a share of 20% of employment and 28% of gross value added at NACE 2 level.Labour intensity is thus not a good indicator for carbon intensity.However, the European non-financial reporting directive (NFRD) makes reporting for non-financial information mandatory for companies with more than 500 employees.Adding Taxonomy-related information into the NFRD requirements will therefore add reporting requirements for some companies with low emissions, and exclude other companies with high emissions.
Figure 4 shows the emissions of different economic sectors, as well as their contribution to employment and gross value added.The sectors with the largest share of greenhouse gas emissions (GHG) are electricity, gas, steam and air conditioning supply (D) with 23% of GHG emissions and crop and animal production (A01) with 13% of emissions.Transport and heating activities by households (jointly adding almost 20% of GHG emissions), which are not defined as economic sectors in the NACE classification, are also among the top four sources of emissions. 5Conversely, sectors with a high share of employment and gross value added are typically less emissions-intensive: The sectors with the largest share in employment are wholesale and retail trade (G), human health and social work activities (Q) and education (P).The economic sectors with the largest share in gross value added are wholesale and retail trade (G), real estate activities (L) and human health and social work activities (Q).
Emissions-intensive sectors currently not covered by the Taxonomy are responsible for 12.4% of total emissions (4595 Mt CO 2 e).These sectors include air transport' (H51), manufacture of coke and refined petroleum (C19), wholesale and retail trade (G), mining and quarrying (B), as well as manufacturing of food, beverages and tobacco (C10-C12).

Sustainable finance Taxonomy in light of climate neutrality
The EU has committed to aim for climate neutrality by 2050 (EU 2020).This commitment is based on the Paris Agreement: In order to have a reasonable chance of limiting warming to 1.5°C, global emissions need to reach net zero around mid-century (IPCC 2018).Based on the performance thresholds in the Taxonomy (see Appendix for an overview), this section relates economic activities to climate neutrality and analyses for selected economic activities whether the performance thresholds set out the Taxonomy are compatible with a pathway towards climate neutrality.We focus on the screening criteria regarding the climate change mitigation target and do not explicitly consider the do no significant harm criteria (DNSH) and minimum social safeguards in our analysis.For the purpose of illustration of the basic mechanisms, we analyse three sectors that are responsible for a significant share of greenhouse gas emissions, namely transport (exemplified by passenger cars and commercial vehicles), buildings (buildings renovation and construction of new buildings), as well as the basic materials sector (cement). 6

Classification of economic activities
The EU Taxonomy defines three types of economic activities, 'green', 'enabling' and 'transition'.The first two types either comply with climate neutrality (green) or contribute to this goal indirectly (enabling).For the green activities, this means that if all companies were to use only the technologies or conduct the activities as defined in the Taxonomy, the sector would converge to net zero emissions.The enabling activities, on the other hand, are needed for other sectors to comply with climate neutrality.Finally, the transition activities do not yet comply with climate neutrality: Even if all companies complied with the current threshold, the sector would not be climate neutral.For these sectors, the Taxonomy sets technical performance thresholds that determine whether such an activity can be regarded as sustainable.
Table 1 provides an overview of the different types of economic activities.The activities not covered can be differentiated into those with low climate relevance (where developing thresholds for all the economic activities would be too tedious), and those with high climate relevance (which are relevant to reaching climate neutrality and therefore should be covered).For the economic activities covered by the Taxonomy, Table 1 differentiates between whether such activities need to meet a threshold to comply with the Taxonomy or not.Activities defined as green or enabling are automatically eligible, irrespective of the actual carbon emissions associated to them (though they still need to comply with the DNSH criteria and minimum social safeguards).For transition activities, on the other hand, a threshold or minimum requirement exists that defines whether an activity is regarded as sustainable or not.However, such a threshold is not necessarily compatible with climate neutrality.For some of the transition activities, a second (future) threshold is defined that indicates a pathway towards climate neutrality, while other activities lack such a pathway.In order to illustrate this concept, we assess some of the activities covered by the Taxonomy from those three categories (green/ enabling, transition with pathway, transition without pathway) in more detail in the remainder of Section 4.

Green/enabling
Green and enabling activities can be found in all of the seven sectors addressed by the Taxonomy.All of these activities cause few or no CO 2 emissions, capture CO 2 from the atmosphere or help to reduce emissions in other sectors.
In the agriculture sector, afforestation and reforestation are classified as green activities, as trees capture CO 2 from the atmosphere and hence reduce CO 2 emissions.In the manufacturing sector, only manufacturing of low-carbon technologies is classified as a green activity.All other manufacturing sub-sectors are classified as 'transition activities' (see next section).Within the energy sector, the production of electricity from solar, wind and ocean energy as well as the transmission, distribution and storage of electricity, thermal energy and hydrogen and the utilisation of waste heat is classified as green activity.In the water, sewage and waste sector, eligible activities are anaerobic digestion or composting of sewage sludge or bio-waste, capture and utilisation of landfill gas and CO 2 and the transport of CO 2 .
The collection of economic activities 'manufacture of low-carbon technologies' is a special case, as it does not correspond to a single economic sector (NACE).Instead, it contains a list of technologies regarded as low-carbon.All of these technologies are classified as compatible with the Taxonomy, without having to meet performance thresholds.These activities include components and technologies for (eligible) renewable energy technologies, for vehicles with low or zero tailpipe emissions, for electric appliances rated in the highest energy efficiency class and for energy efficient equipment for buildings.In principal, any technology can be added to the list, if it supports substantial emission reductions.The advantage is that this category allows for the inclusion of new technologies as they emerge.The disadvantage is that it also allows for lobbying by certain industry associations.This is reflected by the high number of 77 public and private organisations from various industries (energy, transport, manufacturing, buildings) that responded to the public consultation, often suggesting to add further activities to the list.

Transition: threshold with pathway towards climate neutrality
For some activities, two thresholds were defined, one current threshold and one future threshold.In these cases, the future threshold typically indicates a pathway towards climate neutrality.An example for such activities in the energy sector is the production of electricity from gas, bioenergy, hydropower and geothermal energy.For these activities, life cycle emissions (LCE) need to be below 100 g CO 2 /kWh (today) and need to decline to zero by 2050. 7In this section, we look at more detail at passenger cars and commercial vehicles.
For passenger cars, the performance thresholds are in line with the EU's Clean Vehicles Directive (2019/1161).Accordingly, all zero tailpipe emission vehicles (incl.hydrogen, fuel cell, electric) are eligible, irrespective of the carbon intensity of the fuel they consume.Cars with tailpipe emission intensity of max 50 g CO 2 /km (vehicle km) are eligible until 2025. 8The underlying rationale is that the energy carriers used are assumed to become low or zero carbon 'in the near future' (TEG 2020). 9The threshold of 50 g CO 2 /km until 2025 is around 50% of the current fleet emissions performance standards of 95 g CO 2 /km for passenger cars (Directive 2019/631). 10In practice, this excludes many current plug-in hybrids from the Taxonomy, since only smaller plug-in hybrid models can meet this threshold (UBA 2019).
In the stakeholder consultation, a split between industry respondents (associations and individual companies) versus NGOs and environmental agencies can be observed (Figure 5).While NGOs and environmental agencies mostly agree that the thresholds are compatible with net-zero emissions, industry representatives take a more critical stance: A typical suggestion from industry is to make the zero-emission vehicle standard mandatory by 2030 instead of 2025, i.e. to extend the interim period where low-emission combustion vehicles are admissible.Thus, almost 40% of respondents argue for a less stringent threshold.This is not surprising as the responses for passenger cars are dominated by companies (38% of respondents) and industry associations (one-third of respondents).Another commonly voiced suggestion by manufacturers is to change the metric to life-cycle emissions or well-to-wheel instead of tailpipe emissions.
The passenger cars sector illustrates the difficulty of using a single threshold for different purposes (cf.Section 3.1).While on the company level of a car manufacturer, an average emissions intensity of the fleet of 50 g CO 2 /km would already be quite ambitious, on a project level an investment into less efficient vehicles (e.g. a new factory producing hybrid cars) may not be necessary from a climate point of view, since zero tailpipe emissions vehicles are already commercially available.

Transition: threshold without pathway towards climate neutrality
In several sectors, activities exist where only one performance threshold was defined, which does not indicate a pathway towards climate neutrality.In this section, we look at two prominent and emissions-intensive examples, namely buildings (Section 4.4.1) and manufacturing of basic materials (Section 4.4.2).

Buildings
We assess the thresholds both for new buildings and retrofits.For either of these activities, the standards for compliance with the Taxonomy differ based on the location of the building.
4.4.1.1.Building renovation.For renovations of existing buildings, there are two ways how an activity can comply with the Taxonomy.First, any 'major renovation', according to the Energy Performance of Buildings Directive (EPBD, 2010/31/EU), qualifies as compatible with the Taxonomy.Second, any renovation that reduces energy consumption by at least 30% is also eligible under the Taxonomy.This needs to be verified by an energy audit and an energy performance certificate (European Commission 2020b).
The definition of a major renovation varies between EU countries.It is either defined relative to the building value (costs of the renovation exceed 25% of the building value) or relative to the building surface (more than 25% of the surface of the building envelope undergoes renovation).Depending on the national efficiency of the building stock, these criteria are estimated to translate into a reduction of final energy demand of 50-80% (Toleikyte et al. 2016).Consequently, the Taxonomy's second criterion of reducing energy demand by at least 30% can be seen as the 'weaker' criterion (i.e. more easily to fulfil).
Depending on the previous renovation status of a building, reducing energy consumption by 30% does not necessarily make a building compatible with a climate-neutral building stock. 11This would holds for a large share of European buildings: In those European countries, where data from energy performance certificates (EPCs) is available, only three per cent of buildings have the highest energy performance class (BPIE 2017).In Germany, residential dwellings with more than one flat consume around 130 kWh/m² annually (temperature-adjusted), corresponding to an energy efficiency class of D to E (Stede, Schütze, and Wietschel 2020).For these buildings, a reduction of more than 60% would be required to reach Germany's efficiency class A, while a 30% reduction of energy consumption would leave such dwellings still in the mid-range of the energy efficiency class C. 12  In the stakeholder consultation, one of the most frequent critiques was that there is no absolute energy efficiency goal as part of the building renovation threshold.Such a threshold could, for example, relate to the energy efficiency classes set in the energy performance certificates.The thresholds of the EPC classes (e.g.classes A to G) vary between EU countries and thus reflect different climatic conditions or national preferences for efficiency standards, which would make them a viable option.Another critique was that no ex-post metering is required, reflecting concerns that building retrofits sometimes do not deliver the promised savings (e.g.Fowlie, Greenstone, and Wolfram 2018).In total, half of the stakeholders responding to the building renovation threshold propose a tightening of the renovation threshold (Figure 6).This is especially interesting since two-thirds of the respondents in this section are companies or industry associations, which reflects that there are also corporate stakeholders with an interest in higher efficiency goals.
4.4.1.2.Construction of new buildings.For the construction of new buildings, the annual primary energy demand (in kWh/m²) must be 20% lower than the national 'nearly zeroenergy buildings' (NZEB) standard.This lower demand can be achieved either by technical efficiency standards, or by the installation of on-site renewables. 13According to the Energy Performance of Buildings Directive, all newly constructed buildings in the EU must be NZEBs from 2021.However, these NZEB standards are determined nationally.Although this is sensible in principle due to different climatic conditions among Member States, in practice there is a very high variation of the NZEB thresholds: While Denmark has the most stringent standard of all EU countries (20 kWh/m²a), this maximum primary energy requirement rises to more than 100 kWh/m²a in countries such as France (Ipsos and Navigant 2019).Consequently, several respondents suggest to solve this issue by introducing an additional pan-European maximum eligible consumption level for new buildings.This would address the perverse incentive that some investors might invest in countries with lower standards to green their portfolios.Other critiques include that fossil fuel consumption is not ruled out for new buildings, and that (similarly as for building retrofits) actual consumption should be metered.

Manufacture of basic materials
With the exception of the manufacture of low-carbon technologies discussed in Section 4.2, all manufacturing activities included in the Taxonomy belong to the basic materials sector.Typically, the threshold corresponds to the value of EU ETS benchmarks (in tonnes CO 2 e/tonne material), which determine the level of free allocations based on the carbon intensity of the best-performing plants. 14Here, we analyse the manufacture of cement in more detail, due to its central importance in global climate policy: Direct and indirect emissions of the manufacture of cement alone are responsible for eight per cent of global emissions (Andrew 2018).
EU ETS benchmarks rely on best available technology (BAT) estimates calculated from historic production data from 2007 to 2008.NGOs and public authorities in particular emphasised that the benchmarks are outdated and do not take into account technological progress during the last decade. 15The clinker threshold of 0.766 tCO 2e /t of clinker was criticised, since studies show that a threshold of 0.7 would be realistic (Favier et al. 2018).This criticism is reflected in a high number of respondents that propose to tighten the threshold for the cement sector (Figure 7).
Another highly disputed performance threshold is the benchmark for cement, which is based on a multiplication of the clinker threshold with the global average clinker to cement ratio of 65% (share of clinker used in the production of cement).Since for the cement sector an average determines the threshold in the Taxonomy (instead of a metric based on the best-performing installations), this approach is criticised as inconsistent with the BAT approach chosen for the other basic materials by some respondents.Moreover, several respondents cite a report from the United Nations Environmental Program, which regards a global average clinker to cement ratio of 60% on average as realistic (up to 50% in some applications), due to substitution effects (Scrivener, John, and Gartner 2018;Rissman et al. 2020).As a comparison, the current ratio in European production is at roughly 80%, and the European cement industry association predicts 70% for 2050 (CEMBUREAU 2018).Two national environmental agencies (Germany and Austria) question if cement should be part of the Taxonomy of sustainable activities at all, since fully decarbonised cement is currently not commercially available, and refer to the missing focus on the possibility to substitute some types of the basic material. 16 There are discrepancies between the logic behind the performance thresholds of the energy-intensive manufacturing processes cement and hydrogen, which can be explained by differences in industry structure.The Austrian environmental agency points out that the cement sector is similar to the hydrogen sector with respect to having to mitigate emissions by technologies that are not yet commercially available.However, while the hydrogen threshold is set below the ETS benchmark, the threshold for the manufacture of cement is set at 100% of the EU ETS benchmark value.This discrepancy may be explained by different applications of the Taxonomy (see discussion in Section 3.1): While there are significant existing cement production capacities (for which the Taxonomy threshold will apply also at company level), the hydrogen threshold will mostly be relevant for greenfield investment into new production capacity (European Commission 2020c).Again, this discrepancy illustrates the challenge of using a single performance threshold for different use cases.

Discussion and conclusion
The EU Taxonomy for sustainable activities is a tool that may play an important role in channelling investments into low-carbon activities in the economy, thus helping to put the EU climate targets into practice.The main purpose of the EU Taxonomy is to provide more transparent and standardised information on the environmental performance of a company or a specific investment.Hence, it does not replace but complement climate policies, such as the EU-ETS or emissions standards for buildings or vehicles.The EU Taxonomy can serve as an international benchmark and blueprint for other classification systems for sustainable finance.Therefore, early harmonisation with taxonomies being developed in other jurisdictions will be important to avoid fragmentation.The International Platform on Sustainable Finance is an important initiative in this regard.
With its focus on sectors that have a high carbon intensity, we demonstrate that the Taxonomy currently covers economic activities in sectors that are responsible for up to 80% of the EU's emissions, but only 20% of employment and 28% of gross value added.While the Taxonomy is still under development, this paper identifies a number of important lessons learned and points out areas for further development.
First, labour-intensity is not a good indicator for carbon-intensity.However, under the European non-financial reporting directive (NFRD), all companies with more than 500 employees need to report non-financial information.Adding taxonomy-related information into NFRD requirements will therefore exclude some emission-intensive companies, while at the same time adding administrative effort for companies with low emissions.To circumvent this issue, an additional metric based on emission-intensity could be added as a requirement for Taxonomy-related reporting under the NFRD.
Second, the Taxonomy does not currently indicate a path towards climate neutrality for several of the economic activities it covers.While activities labelled 'green' or 'enabling' are assumed to contribute directly to reaching the climate goals, for the sectors in 'transition' technical thresholds need to be met for an activity to be Taxonomy-eligible.We show that pathways towards climate neutrality are set out for some of these activities, such as passenger cars, where only fully electric zero-tailpipe emission vehicles are eligible from 2026.However, in other sectors such a pathway does not exist, most notably in building renovation and the basic materials sector (e.g.steel or cement).In the basic materials sector, the thresholds are based on current best available technology.However, typically capital-intensive breakthrough technologies are needed to decarbonise these economic activities.These innovative technologies are currently not incentivised by the Taxonomy, due to its binary structure: Once a company reaches the threshold, there is no further incentive to improve beyond the threshold.
Third, by setting science-based performance standards for different sectors of the economy, the Taxonomy could be used to make corporate sustainability claims, such as a net zero target, more credible.Although not all thresholds for economic activities are currently in line with climate neutrality, the fact that the Taxonomy will be regularly reviewed and performance thresholds will be updated mitigates this shortcoming to some extent.However, the Taxonomy's stringency varies across sectors and there are differences in the extent of coverage of economic activities across sectors.As a result, crosssectoral comparisons of corporate claims based on the Taxonomy will probably remain challenging.Within a sector, on the other hand, the Taxonomy can be a useful additional tool to compare corporate claims.
Fourth, some of the challenges relating to setting adequate performance criteria can be understood by the variety of possible use cases of the Taxonomy.We differentiate between the company level (share of Taxonomy-compatible revenues), where the Taxonomy may be used for evaluating the current average performance of companies, and the project level, where the Taxonomy may be used as a screening tool for new investments.This dichotomy illustrates the dilemma of using one performance threshold for different purposes: In the basic materials sector, for example, basing the eligibility threshold on the best-performing installations using conventional technology may be sensible when the Taxonomy is used to evaluate the existing activities of a company.However, for new investments, only investments into innovative technologies that go beyond the status quo should be incentivised in order to accommodate the long-term investment cycles in these industries.This dilemma of using a single criterion for multiple purposes may persist for sectors in need of investments into breakthrough technologies, even if the thresholds become more stringent over time.
The challenge of using a single criterion for multiple purposes creates the risk that the Taxonomy will incentivise only marginal improvements for new investments, which might create a lock-in into carbon-intensive assets and hinder innovation in the longterm (Mattauch, Creutzig, and Edenhofer 2015;Unruh 2000).It can be solved in two ways in principle.First, multiple thresholds could be defined for a given economic activity, differentiating between investments in newly built projects on the one hand and existing assets at the company level on the other hand.This spirit is already embodied in the various thresholds for buildings-related activities, where criteria for the construction of new buildings are more stringent than those for existing buildings (retrofits, as well as acquisition and ownership of real estate).Second, additional forward-looking indicators, such as a decarbonisation strategy, climate targets or green investment targets, could be mandated by the Taxonomy in order to identify companies that plan to comply with climate neutrality in the future (Monasterolo 2020).
Finally, although a large fraction of overall EU emissions are addressed by the Taxonomy, many economic activities are not covered by the system.These non-covered activities can be divided into three broad categories.First, economic activities which account for a large fraction of gross value added and employment, but for a small share of overall emissions.Excluding these economic activities minimises administrative effort for sectors where emissions reductions are less important in terms of the overall mitigation potential.Second, carbon-intensive economic activities, which cannot be fully substituted and where technological progress is needed to reduce emissions in the future (such as aviation and maritime shipping).For these activities, thresholds should be developed.Third, economic activities that are carbon-intensive but should be phased out since technological alternatives exist, such as the burning of coal and petroleum.
For the third group of activities that are incompatible with a climate-neutral economy, an explicit so-called 'brown taxonomy' could be developed.Alternatively, exclusion criteria could be built on the 'do no significant harm' (DNSH) criteria of the Taxonomy, as already implemented in the EU Recovery and Resilience Facility.Such a brown taxonomy would be a valuable screening tool for sustainable investment funds, such as the ones defined under the EU Ecolabel Directive, as well as governmental subsidies or investments.A brown taxonomy (or exclusion criteria based on the DNSH criteria) would thus create the possibility to explicitly rule out investments into activities deemed unsustainable, while still allowing investments into activities not covered due to their low emissions intensity.It would also provide transparency to investors about the risk of potential stranded assets in their portfolios.

Notes
1.For most activities, there were either no or minor changes between the thresholds in the interim and the final report released in March 2020.Results are publicly available at https://ec.europa.eu/eusurvey/publication/teg-report-taxonomy?surveylanguage=en.2. Since many organisations commented on several activities, the total number of responses is higher than the total number of respondents.3. See https://ec.europa.eu/transparencyregister/public/consultation/statistics.do?action= prepareView&locale=en#en (numbers as of 09 November 2020).4. NACE is the classification of economic activities in the European Union.Level 1 is the first layer of sectors consisting of 21 sectors, indicated by the letters A to U. Level 2 is the second layer of sectors, indicated by two-digit codes from 01 to 99. 5.These sectors are indirectly covered by the taxonomy, because transport activities of households are covered via the threshold for the production of passenger cars.Heating activities by households, on the other hand, are covered via the construction and renovation of buildings activities.6. Transport is responsible for 11% of EU greenhouse gas emissions (see Figure 4).It is the only major economic sector in the EU where emission are still rising and are above 1990 levels.Direct emissions for the heating of residential buildings alone constitutes nine per cent of EU emissions (Figure 4).If indirect emissions (e.g.electricity and district heating) and emissions from the heating of commercial buildings is considered, this share rises significantlyin Germany, direct and indirect emissions of all buildings account for roughly a quarter of total emissions (Stede, Schütze, and Wietschel 2020).The production of basic materials accounts for around 25% of global CO 2 emissions and 16% of European GHG emissions (Neuhoff et al. 2019), of which direct and indirect emissions of cement alone are eight per cent of global emissions (Andrew 2018).7. The explicit reference to a declining value to zero by 2050 has been removed in the draft of the Delegated Acts (European Commission 2020b).8.In contrast to passenger cars, the rest of the transport sector has no clear pathway: Many of the thresholds will be reviewed in 2025, either because relevant European directives are reviewed (such as the freight transport services by road), or because the technology for zero-emission vehicles is not yet commercially available (e.g.interurban scheduled road transport, see TEG 2020).9.This assumption is more justified for vehicles powered by electricity, as renewable electricity is expanded across (Europe, leading to a declining carbon intensity of electricity).Hydrogen may also be produced by electricity through electrolysis ('green' hydrogen), however it may also be produced from natural gas via steam reforming ('grey' hydrogen), which is far more emissions-intensive.10.In practice, the difference is even larger, since the Taxonomy threshold is based on the Worldwide Harmonised Light Vehicles Test procedure (WLTP), while Directive 2019/631 foresees a replacement of the fleet performance thresholds (which were so far based on the New European Test Cycle, NEDC) for 2021.It is expected that the maximum fleet emissions under Directive 2019/631 will rise by around 20% under WLTP (BMU 2020).11.In a recent study for the European Commission, renovations that reduce energy consumption by 30% are on the margin of 'light' renovations (Ipsos and Navigant 2019).12.In Germany, buildings with a consumption of 75-100 kWh/m²a are labelled as energy efficiency class C. Any consumption below 50 kWh/m²a belongs to class A, buildings with a consumption below 30 kWh/m²a are labelled class A+. 13.Any off-site energy generation must be limited to district heating and cooling systems and local renewable energy sources (TEG 2020).14.The manufacture of hydrogen is an exception to this rule.It is set at 5.8 tCO 2 e/t of hydrogen, while the ETS benchmark is at 8.85 tCO 2 e. 15.The EU has also acknowledged the need for more frequent updates of these benchmarks, which will be updated twice in the years 2021-2030 (phase 4 of the EU ETS), in order to avoid windfall profits and reflect technological developments since 2008.https://ec.europa.eu/clima/policies/ets/revision_en(accessed 12 November 2020).16.Approximately two-thirds of the emissions from the cement production stem from process emissions during the clinker production (decomposition of limestone), with the remainder of CO 2 emissions being due to combustion of fuels.These emissions can only be fully avoided by carbon capture and storage or use, although they can be partly offset by substituting clinker by other mineral components in cement and concrete (IEA 2018;Neuhoff et al. 2014).
Appendix: Comparison of technical screening criteria in the Taxonomy.

Figure 1 .
Figure 1.Respondents in the EU consultation.Note: Total of 355 organisations and 1672 responses.

Figure 2 .
Figure 2. Two different applications of the Taxonomy.Source: own depiction.

Figure 3 .
Figure3.Three channels relating the Taxonomy to reducing greenhouse gas emissions.

Figure 4 .
Figure 4. Greenhouse gas emissions, gross value added and employment per NACE sector (level 1 and 2).Notes: Sectors in light blue are (partly) addressed by the Taxonomy, sectors in dark blue are not covered by the Taxonomy.Sectoral emissions are presented at NACE 2 level where available, and at level 1 where no lower level exists (e.g.sectors D and G) or overall emissions of the sector are low.Source: Data for 2017 from Eurostat.

Figure 5 .
Figure 5. Perspectives on the threshold for passenger cars and commercial vehicles.Note: n=40 respondents.

Figure 6 .
Figure 6.Perspectives of stakeholders on the building renovation threshold.Note: n=42 respondents.

Figure 7 .
Figure 7. Perspectives of stakeholders on the threshold for the cement sector.Note: n=17 respondents.
carbon transport (water transport) Infrastructure required for zero direct emissions water transport, supporting renewable energy sector; fleet using infrastructure must meet Infrastructure a) dedicated to operations of zero direct (tailpipe) emissions, b) to the provision of shore-side electric power to vessels at berth, c) to the performance of the ports own in the relevant activity; not dedicated to the transport of fossil fuels operations with zero emissions, d) to transhipping freight between modes; infrastructure is not dedicated to the transport of fossil fuels Inland passenger water transport Zero direct emissions vehicles; advanced biofuels or renewable liquid and gaseous transport fuels of non-biological origin, < 50gCO 2 /pkm, to be reviewed in 2025 Vessels have (a) zero direct (tailpipe) CO 2 emissions or (b) until 31 December 2025, hybrid vessels use at least 50% of zero direct (tailpipe) CO 2 emission fuel mass or plug-in power for their normal operation x 2 2 Inland freight water transport Zero direct emissions vehicles; direct emission <50% of the reference CO 2 emissions in the subgroup 1a) Zero direct emissions vehicles or b) direct emission <50% of the reference CO 2 emissions in the subgroup; 2) Vessels purchased or operated are not dedicated to transport fossil fuels x 2 3 Infrastructure for personal mobility Part of low-carbon transport in the TEG report Personal mobility: pavements, bike lanes and pedestrian zones, electrical charging and hydrogen refuelling installations for personal mobility devices x -Infrastructure for rail transport Part of low-carbon transport in the TEG report Infrastructure a) electrified trackside infrastructure and associated subsystems or trackside infrastructure and associated subsystems where there is a plan for electrification/use zero tailpipe CO 2 emission trains within 10 years, b) the infrastructure and installations are dedicated to transhipping freight between the modes, c) infrastructure and installations are dedicated to the transfer of passengers from other modes to rail x -Low carbon airport infrastructure Not included in TEG report Infrastructure a) dedicated to the operation of aircraft with zero tailpipe CO 2 emissions, b) to the provision of fixed electrical ground power and preconditioned air to stationary aircrafts, c) to the zero direct emissions performance of the airport's own operations: electric charging points, electricity grid connection upgrades, hydrogen refuelling stations; infrastructure is not dedicated to the transport of fossil fuels x -Retrofitting inland water passenger and freight transport Not included in TEG report Until 2025, the retrofitting activity reduces fuel consumption of the vessel by at least 10% expressed in litre of fuel per tonne kilometre, not dedicated to transport fossil fuels x -Sea and coastal freight water transport Not included in TEG report Vessels have zero direct (tailpipe) CO 2 emissions; or (b) until 2025, hybrid vessels use at least x -direct (tailpipe) CO 2 emission fuel mass or plug-in power for their normal operation; or c) until 2025 direct (tailpipe) CO 2 emissions 50% lower than the average reference CO 2 emissions value // Vessels are not dedicated to the transport of fossil fuels Sea and coastal passenger transport Not included in TEG report Vessels have zero direct (tailpipe) CO 2 emissions; or (b) until 2025, hybrid vessels use at least 50% of zero direct (tailpipe) CO 2 emission fuel mass or plug-in power for their normal operation; or c) until 2025 vessels have an attained Energy Efficiency Design Index (EEDI)469 value 10% below the EEDI requirements applicable on 1 January 2022 x -Retrofit of sea and coastal water transport Not included in TEG report Until 31 December 2025, the retrofitting activity reduces fuel consumption of the vessel by at least 10% expressed in grams of fuel per deadweight tons per nautical mile, not dedicated to transport of fossil centre implements the European Code of Conduct for Data Centre Energy Efficiency or in CEN-CENELEC document CLC TR50600-99-1; 2) The global warming potential (GWP) weighted average for the mixture of refrigerants used in the data centre cooling system does not exceed activity researches, develops or provides innovation for technologies, products or other solutions that are dedicated to enable one or more economic activities for which the technical screening criteria have been set out in this Annex, with the exception of activities x -(Continued) JOURNAL OF SUSTAINABLE FINANCE & INVESTMENT

Table 1 .
Expanded classification of economic activities in the EU Taxonomy and their allignment with climate neutrality.