Taxonomy information was included in all investigated reports. In 47 cases (76%), the taxonomy information was included in the environmental section of the sustainability statement, as required by ESRS 1 (para. 113). The other 15 cases (24%) reveal a variety of practices: presenting the taxonomy information elsewhere in the sustainability statement, or presenting a short introduction or just the narrative information in the (environmental section of the) sustainability statement while presenting the tables as an appendix either to the sustainability statement or to the annual report as a whole.

To describe the volume of taxonomy information we have looked at the number of pages and the number of words in the taxonomy section (or sections), excluding the tables. As can be seen in Table 1, this differs considerably across companies. On average, companies provide around five pages of narrative information to support the mandatory taxonomy tables. Four companies use one page or less; nine companies provide ten pages or more. Although it might be expected that the volume of explanatory material is related to the materiality of the eligible activities, this is not the case. There is no observable relationship between the number of words in the narrative section and the percentage of turnover classified as eligible.⁶

All companies use the required tables to report the KPI’s (turnover, CapEx, OpEx) by activity. One company, which identified only one eligible activity, uses a slightly abbreviated version. Several companies present tables in a self-designed more compact format, apparently to enhance accessibility, while presenting tables in the required format in an appendix, either to the sustainability statement or the annual report as a whole. One company with both financial and non-financial activities presents separate taxonomy information for these two segments, as well as a consolidated group-level KPI in the form of a weighted average for the two segments in a single overview. This reporting format is in line with guidance by the European Commission in the form of frequently asked questions.⁷

In addition to the KPI templates, the Delegated act – disclosure prescribes a template table in which a company must answer yes or no to six questions on whether it carries out activities related to nuclear energy and fossil gas. If one or more questions are answered with yes, additional templates are prescribed for reporting eligible and aligned KPI’s for these activities. The majority of companies (50 or 81%) provide this table. Of the 13 companies that do not, four include a note that this table would be irrelevant or that any such activities are immaterial. The remainder omit the table without comment. For companies that do not have nuclear or fossil gas activities, including a table with only ‘no’s’ may indeed seem redundant. In its July 2025 modifications, the European Commission intends to eliminate this negative confirmation requirement. In our sample, 14 companies do report relevant activities. All but two of these provide the required supplementary tables.

The key objective of the Taxonomy Regulation is to provide insight into the extent to which reporting entities’ activities are sustainable. Table 2 provides a high-level summary, with a more detailed breakdown by industry in Appendix 2. Eligible activities (aligned or not) range from zero to 100% of the applicable denominator (turnover, CapEx, OpEx), with averages around 40%. These numbers are comparable with those found in earlier research by EY (2024) and KPMG (2024). As can be seen from Appendix 2, eligibility is clearly related to industry. This would be true in any taxonomy, but it is probably more pronounced in the EU taxonomy as the Taxonomy Regulation does not claim to be complete in its listing of eligible activities. In our sample, this notably affects the four brewery companies (malt beverages) who report 0% eligible revenue, not because their activity is inherently unsustainable, but because it is not yet covered by the Regulation. In contrast, pharmaceuticals, biological products (in our sample: biomedical) and motor vehicles show high levels of eligibility, at least for turnover (averages of 73% to 88%).

For all three KPIs, alignment is relatively rare. While a few companies report high percentages of aligned activities, the median is zero or close to zero. Again, industry patterns are observable. For the automotive industry, for instance, high eligibility for turnover (average of 88%) translates into low alignment (10%), whereas for electric services the moderate level of eligibility (47%) represents aligned activities to a significant extent (37%).

The identification of eligible activities is clearly an important step in the application of the taxonomy. Table 3 provides high-level descriptives on the number of activities that the 62 companies in our sample have identified. It should be noted that a specific activity may be reported both under ‘aligned’ and under ‘eligible not aligned’ if a company decides that only part of the turnover (CapEx, OpEx) satisfies the criteria for alignment. For this reason, Table 3 does not show the number of unique activities identified. It can be seen from Table 3 that the frequency distribution of identified activities is heavily skewed. While many companies report no or just a few activities, some companies have gone to great lengths to identify activities. The two construction companies in the sample are outliers with between 25 and 43 activities identified as eligible and/or aligned for turnover and similar numbers for CapEx. Another five companies report ten or more activities as either eligible/not aligned or aligned for turnover. In addition to Table 3, it can be reported that 38 companies (61%) report at least one aligned activity for turnover, 47 (76%) at least one aligned activity for CapEx and 35 (56%) for OpEx.

As suggested by Table 3, the identification of activities does not have to be identical across the three KPI’s, and it does in fact vary considerably. It is reasonable to assume that a revenue-generating activity will often also require capital expenditure and operating expenditure, so that considerable overlap in activities across the three KPI’s is expected. In fact, a few companies (especially with just one or two identified activities) do seem to interpret the taxonomy in this way, so that they report CapEx and OpExKPI’s only for those activities for which they also report turnover KPI’s. However, the taxonomy allows for differences, e.g. when companies invest in activities that do not, or do not yet, generate revenue. Most companies have applied the taxonomy in this way, with the result that, in round numbers, for every instance in which both turnover and capital expenditure are reported for the same activity, there are 1.2 other instances of activities for which only one of the two is reported. For turnover/OpEx and CapEx/OpEx the differences are somewhat greater.⁸

A good example of these differences is that 30 companies (48%) report ‘transport by motorbikes, passenger cars and light commercial vehicles’ under CapEx as eligible with respect to the climate change mitigation objective (a combination referred to as CCM 6.5, based on the Delegated acts – environmental and climate).⁹ Only seven (11%) report a turnover KPI for this activity, suggesting that most have thought in terms of transport as a supporting rather than revenue-generating activity. Of course, this raises the question whether the 32 companies that do not report a CapEx KPI really did not invest in this type of transport at all in 2024 and 2023, have applied a materiality judgment (see Section 4.3) or simply haven’t thought about it. Similar observations could be made about CCM 7.7. ‘acquisition and ownership of buildings’. One would expect that most companies own at least some buildings, yet eligible CapEx is reported by 21 companies or 34% (eligible turnover by three or 5%). While it is possible that some companies did indeed not invest at all in buildings during the year, these results do suggest that, apart from materiality, there exist quite fundamental differences of opinion with respect to how this kind of supporting activities should be handled under the taxonomy.

As indicated in Section 2, the six objectives differ in the number of activities for which substantial contribution criteria have been defined: 94 and 100 for climate change mitigation and adaptation, respectively, but 21 for circular economy, 6 each for pollution prevention and water and marine resources, and 2 for biodiversity. It is therefore to be expected that most reported KPI’s relate to the climate change objectives. This is borne out by Table 4, which shows the number of reported KPI’s, across all companies, by sustainability objective.

It is clear from Table 4 that the vast majority of KPI’s are reported in connection with the two climate change objectives. The only other objective that generates a substantial number of KPI’s is circular economy. The water/marine resources and biodiversity objectives are reported so infrequently as to raise the question whether it is meaningful to keep them in the taxonomy at all (other than for assessing no significant harm), if the number of activities for which screening criteria are defined cannot be increased.

However, increasing the number of potentially eligible activities for each objective creates a potential problem of double eligibility (one activity is potentially aligned with two or more objectives). This is handled in principle in the Delegated act – disclosure by the requirement that a company should in such a case mention both objectives in the relevant activity row in the templates (e.g. CCM 7.1 / CE 3.1, both referring to ‘construction of new buildings), allocate the turnover (CapEx, Opex) to the two activities, show the most important activity in bold in the template, and provide a separate table showing aggregated KPI’s by objective. In practice, only a minority of 15 companies (24%) report at least one activity contributing to more than one objective in their templates. One might also raise the question what incentives companies have to assess activities against multiple objectives, as this will not further increase the KPI’s.

As an example, one might, perhaps, expect a significant overlap between climate change mitigation and climate change adaptation. However, the two objectives are rather different. Whereas climate change mitigation refers essentially to the company’s impact on climate change, climate change adaptation refers to the resilience of the company’s activities to the consequences of climate change. This means that for the activity CCM/CCA 4.3, electricity generation from wind power, the substantial contribution criterion for CCM is very simple: it has just to be shown that the activity generates electricity from wind power. For climate change adaptation, however, it would have to be shown that the climate risks associated with the activity have been systematically assessed using science-based best practices, and that appropriate and proportional adaptation solutions have been implemented. One wonders what company would want to do the necessary paperwork if it has already been ascertained that the revenue (CapEx, OpEx) of this activity is aligned for climate change mitigation purposes. While it would be easy to claim simple eligibility for activities covered both by climate change mitigation and climate change adaptation, it would still require some effort to allocate turnover (CapEx, OpEx). These costs without clear benefits would seem to explain why companies mainly report on climate change mitigation and hardly anything on climate change adaptation. In one case did a company report all its eligible activities under the climate change adaptation objective, whereas its industry peers all reported broadly similar activities under climate change mitigation. The explanatory notes of the first company were not extensive enough to see whether this truly reflected a different assessment, or that it was an error.

That materiality considerations do, in practice, seem to have influenced the application of the taxonomy has been suggested above. One company states explicitly that ‘the concept of materiality established for financial reporting’ has been used to determine taxonomy-eligible activities. However, the Delegated act – disclosure, as applicable in 2024, does not include general materiality thresholds for identifying and reporting activities. Disclosure of OpExKPI’s may be omitted, though, if operational expenditure is not material for the business model of a non-financial undertaking (Annex I, 1.1.3.2). The delegated act defines a rather specific definition of operating expenditure as, essentially, non-capitalized research and development costs and non-capitalized expenditure for day-to-day servicing of property, plant and equipment. This is likely to be a relatively small fraction of what many companies and users of financial statements think of as operating expenditure. Nine companies (15%) make use of this materiality exemption for OpEx. There appears to be some diversity among these companies in whether they assess the materiality of taxonomy-defined OpEx for eligible activities relative to a general definition of operating expenditure for these eligible activities, or whether they assess the materiality of taxonomy-defined OpEx for eligible activities relative to taxonomy-defined OpEx for non-eligible activities. In our view, the usefulness of information on this specific definition of OpEx is not evident.

A general impression of the materiality of reported KPI’s is given in Table 5, which lists the total number of activities reported in the templates as either eligible/non-aligned or aligned, and with a positive amount. With some variation between the three categories, the main finding is that around two-thirds of reported numbers are less than 1% of the relevant denominator. Around 12% has a value of 10% or more of the relevant denominator.¹⁰ Companies report not just the currency amounts of the KPI’s but also show them as a percentage of turnover (CapEx, OpEx). Even though some companies present these percentages with one or two decimals, many amounts are so small that, because of rounding, reported percentages of 0% (0.0%, 0.00%) make up about one-third of all reported KPI’s (not tabulated in Table 5). In combination with the observations about diversity of practice regarding supporting activities, this suggests considerable diversity in terms of whether and how companies have applied a materiality threshold. Some clearly have not, whereas others who probably did invest in own buildings or motor transport, may well have done so. The European Commission’s proposal to adopt a 10% threshold will provide more clarity and, as suggested by Table 5, will lead to a very significant reduction in the number of reported datapoints. In our sample, 15 companies (24%) would not have reported any KPI’s across the three categories over 2024 if this threshold had been applied.

Because of the prescribed templates, companies have little discretion over how they report the KPI’s. They have more discretion regarding the accompanying notes, other than presenting summary tables in addition to tables based on the templates (see Section 4.1). The Delegated act – disclosure (Annex I, para. 1.2) requires explanation of accounting policies (determination of the numerators and denominators of the KPI’s and the relation to financial statement numbers); explanation of the assessment of eligibility and alignment; and contextual information explaining the reported KPIs and changes therein. The Delegated act includes quite detailed requirements for these explanations which are not necessarily easy to apply in the absence of a clearly defined materiality threshold. There may be significant and specific judgments underlying each individual reported datapoint, including the allocation of CapEx to activities and the assessment of alignment based on technical screening criteria specific to an activity. Providing meaningful notes may be possible in case of just a few activities, but will be challenging when reporting on ten or more. As reported in Table 1, companies have made very divergent choices in the amount of footnote information they provide, and the same is true for choices with regard to content. This makes it hard to provide meaningful descriptive statistics or best practices that are generalizable across companies. We make a number of qualitative observations.

Most companies have not made apparent attempts to make their taxonomy notes attractive or accessible, the most common format being plain text. Figure 1, in contrast, is an example of using the kind of graphics that are common in the rest of companies’ management reports.

The Delegated act – disclosure (Annex I, para. 1.2.2.1) requires companies to ‘describe the nature of their Taxonomy-eligible and Taxonomy-aligned economic activities’. In our view, users of the information are best served by information that relates entity-specific activity descriptions to the activity descriptions in the taxonomy, including the numerical coding for ease of reference. While this can be done in narrative format, a tabular format may be suitable if there are more than a few activities (Figure 2).

A few companies affected by the incompleteness of the taxonomy include a brief comment, for instance as follows: ‘[N]on-eligible actives under the EU Taxonomy should not be interpreted as an indication of AB InBev’s sustainability performance or ambition. Additionally, the company would expect these key performance indicators to substantially increase if and when the specific economic activity “manufacturing of food products and beverages” is added to the Delegated Regulations.’ (AB InBev, Annual report 2024, p. 160). One company adds three pages of text to discuss the sustainability of activities not yet covered by the taxonomy or that would be hard to include because their interaction with other activities would make it difficult to calculate meaningful KPIs (Amadeus, Non-financial information statement and sustainability information 2024, p. 57–60).

Explaining the extent to which eligible activities are aligned requires in principle an explanation of the three steps: making a substantial contribution, doing no significant harm, and complying with minimal safeguards. Disclosures on this point range from the simple to the elaborate. Companies that report no aligned activities sometimes disclose the reasons summarily. A few companies report that they lack the required data or have not yet finished their assessments and therefore report 0% aligned. Other companies describe generic processes applied across all activities, or limit themselves to a general summary of the taxonomy criteria without specific application to their own activities. The European Commission has published ‘Frequently Asked Questions’ about application of the taxonomy, in particular the technical screening criteria. In our sample, 20 companies (32%) mention these FAQs in their notes, mainly as a general reference.

Some companies use long texts to describe taxonomy assessment, which is inevitable if a company with many different activities wishes to comply fully with the requirements of the Delegated act – disclosures. A problem here is that for some activities the technical screening criteria are concise and for others very complex and technical. This can make it difficult to create a balanced text, as illustrated in Figure 3. This shows how one company reports on the assessment of the substantial contribution of two different activities (out of eight) to climate change mitigation. The first text block shows activity 3.17, manufacture of plastics in primary form, which is an example of detailed criteria in technical language. The disclosure begins by citing the text of one of three alternative criteria listed in the Delegated act, followed by an application to the company’s activities (starting with the sentence ‘In this context ...). The second text block shows activity 4.1. electricity generation using solar photovoltaic technology. Here the criterion is simply that the activity generates electricity using solar photovoltaic technology, and the company needs to do no more than repeat this as an assertion. The example can be seen as best practice in the sense of being specific about each activity and tailoring the specificity of information provided about each activity to the degree of detail in the screening criteria. However, another question is what level of detail is required to make this information most helpful to users.

A different approach is taken by another company. This company, with 17 identified eligible activities (including 3.17 and 4.1 as in Figure 3) grouped the sometimes different substantial contribution criteria with respect to climate change in five broad categories, so that the result of the compliance tests can be shown in tabular form (Figure 4).

As with substantial contribution, disclosures on doing no significant harm differ considerably, both because of variation in the specificity of the requirements in the Delegated act and because of companies’ choices to use specific or generic language. For most activities in the taxonomy, the do no significant harm criterion with respect to water and marine resources is formulated in the same way, so that for most companies it would be possible to report on this criterion across activities. Figure 5 gives an example of how this can be done in a succinct yet entity-specific manner. Without literally repeating the requirements, the disclosure does address them in substance.

Disclosures on the minimum safeguards tend to be limited. This is partly explained by companies omitting further discussion once it has been established that none of their activities meets both the substantial contribution and the do no significant harm criteria. Some companies that do report aligned activities simply assert that minimum safeguards have been complied with. Others provide some more information, for instance by indicating that international standards such as OECD Guidelines have been incorporated in the company’s own code of conduct and/or in contracts with suppliers. We note as best practice that some companies make negative assertions such as the following: ‘Fortum (or senior management) has not been found to have violated labour law, human rights, or competition laws. Fortum has also not been found guilty of tax evasion, corruption or bribery.’ (Fortum, Financials 2024, p. 77).

While not the main focus of this paper, we offer some observations on assurance. For information based on the Taxonomy Regulation, as well as for the sustainability statement as a whole, the CSRD requires limited assurance by an independent assurance provider. All companies include an assurance report on the sustainability statement. All reports were issued by audit firms,¹¹ mainly (58 or 94%) the so-called Big-4.

In most cases, this is a report on limited assurance as required under the CSRD.¹² In nine cases (15%), though, the report is a combination of a limited assurance report including also a reasonable assurance opinion on identified elements of the sustainability statement, typically on specific metrics reported under ESRS. Only in two of these cases does the reasonable assurance opinion also include part or all of the taxonomy information.

All assurance reports mention the taxonomy information separately in the scope paragraph, as well as in the opinion paragraph. Other than that, they differ considerably in the amount of detail provided on taxonomy information. French reports have a separate section on the taxonomy information. Danish reports tend to use highly standardized language across audit firms. Some reports merely include a bullet item on taxonomy information in the ‘work performed’ paragraph. Others also touch on taxonomy information in other paragraphs such as ‘responsibility of management’ or ‘our responsibility’.