Using Python to apply climate change objectives of EU Taxonomy

The result is a Pandas dataframe. Here is an example of revenue data for Kering group:

3. Calculate the segment revenue as a percentage

The model is designed to segregate business segment revenues which are stemming from the classification of economic activities, that substantially contribute to the climate change mitigation and adaptation. In order to do that we need to combine two outputs and:

1. Calculate segment revenues as a percentage
2. Find out how many NAICS codes per segment match with the EU classification and substantially contribute to the climate change initiative

The first output is a simple calculation, where if there is a NAICs code available for the segment, then divide this figure with the segment total for the company. The code SEGMTL is the total revenue figure for the segment.

Using this information, the model is then able to calculate the segment revenue percentage of the segmental total. In the implementation, we decided to use segment total versus the consolidated total because the later would include other revenues at a company level, which cannot be identified from a classification point of view. These revenues can be Realized investment gains (losses), Intersegment Eliminations or Corporate Expense/Other etc, which are not part of segment revenue.

The code snippet which performs this calculation is contained in the function getTaxoForRic():

4. Convert NAICS code to Refinitiv business classification equivalent

In order to identify which business segment NAICS codes match with the EU Taxonomy, we must convert the raw NAICS 2007 codes into either NACE or TRBC codes.

• NACE is Nomenclature Statistique des Activités économiques dans la Communauté Européenne. Or translated into English The Statistical classification of economic activities in the European Community.
• TRBC stands for Refinitiv Business Classification (formerly Thomson Reuters Business Classification).

The script reads and loads the Excel file called database.xlsx as a pandas dataframe. This database contains multiple sheets and the one titled NAICS>TRBC contains the conversion mapping table. The NAICS codes are based on the NAICS 2007 version and TRBC is using the latest 2020 updates which include new economic activities such as Solar Electric Utilities.

One limitation of this model is that since the NAICS>TRBC mapping table has a one to many relationships e.g. for one NAICS code you can have multiple equivalent TRBC codes. The model will only return the first TRBC code it finds. This means in the example below concerning Soybean Farming, the first TRBC code for Fishing & Farming is used. Fishing and Farming is included in the draft list of eligible economic activities as of April 2021. However it requires further testing whereas Organic farming does not. Users may want to configure the order or TRBC codes relative to NAICS.

The code snippet which standardizes and performs this conversion is:

5. Matching TRBC codes to the EU climate change taxonomy

Now that the model has found an equivalent TRBC code, the next step is to see if any of these match against the list of climate change economic activities. The worksheet titled EU Taxonomy in the database spreadsheet contains the TRBC code groups. Any TRBC code not found in the EU Taxonomy table is not in scope. These will show up as na or not available. This doesn’t mean that future EU taxonomy objectives won’t include these at a future stage. If the model does find a match, then a value of Yes or No will be returned in the generated report. So, what do these mean?

• Yes = Additional testing needed – for many industries this means a series of screening criteria tests need to be performed in order to determine if the activity is aligned. The targets and thresholds for these industry activities were made publicly available in the Excel tool provided by TEG. We will provide further details about this in the next couple of steps.
• No = No further testing needed – revenue stemming from these activities are considered not to require further testing, and therefore are aligned.

6. Technical screening criteria

The next step of the model is to check if any of the industry codes mapped to each business segment are part of the technical screening criteria. The worksheet Testing Metrics in the database contains the Refinitiv ESG measure and the thresholds for each TRBC activity. For the testing metrics the model only includes 9 NACE activities out of the 72 in total outlined in Mitigation Full Data worksheet of the Excel tool. The model does not incorporate screening criteria tests for Adaptation.

Why only 9 NACE activities?

As of the fourth quarter of 2020, content research teams within Refinitiv performed a data coverage check against the Mitigation Full Data screening criteria. Many of the technical criteria and thresholds did not exist within either fundamental operating metrics database or the ESG database. Several results meant only partial matches, and these were excluded from the model. Only direct matches were included, and these pertained to two ESG datapoints. Cement CO2 Emissions Equivalents and Fleet CO2 Emissions.

Once a positive match has been found with an industry activity the first step is to identify what the technical screening is. For the script this is one of two options – Fleet CO2 emissions or Cement CO2 emissions equivalents and is read from database.

7. Retrieving company reported performance metrics

If a successful, the script will have identified which Refinitiv data measure to use to perform the screening criteria test. The script will show the standardized company reported value into the same units of measure as required by the technical screening criteria. For example, the CO2 equivalent emissions in g/km. Currently the ESG data is predominantly reported at an organizational level only. Even though the ESG values are reported at an organizational level, the testing is being performed at business segment level based on business classification codes available.

8. Passing the technical screening criteria tests

We simplify the stated testing metrics and thresholds from database spreadsheet Testing Metrics into single ESG data point for the purposes of this model. For example:

Original full description:

In the case of the cement industry activity, Refinitiv has data pertaining to total CO2 and CO2 equivalents emission in tonnes per tonne of cement produced. So, section A of the technical screening criteria relating to cement clinker production can’t be calculated as of April 2021. However technical screening can be performed against section B and the target of 0.498 tCO2e/t is used to benchmark against company reported data. To pass the threshold test, the emissions stated by a company need to be lower than 0.498 tCO2e/t. The model compares the company reported figure against the threshold and will return one of three outcomes within the Excel generated report:

9. Calculating segment level taxonomy results

To calculate the result for the whole business, the model first counts how many TRBC/NAICS codes there are per business segment (there can be a max of 4) and then equally weights each of these. Refinitiv does have details about which of the NAICS codes are primary – i.e. the industry activity that best describes the primary products or services the segment is providing. However as of the time of writing this article, there are no methods to expose this level of detail and therefore all NAICS codes are treated with equal importance.

The next step is to then calculate what percentage of segment revenues fall into each of the following segment taxonomy categories:

• Business Segment Level - Eligible %
• Business Segment Level - Eligible - but not aligned %
• Business Segment Level - Not in scope %
• Business Segment Level - Aligned by industry activity %
• Business Segment Level - Aligned passed technical screening criteria %
• Business Segment Level - Aligned total %
• Business Segment Level - Additional testing needed %

To calculate the following formula is used: Segment Revenue % * (Count of EU taxonomy results * Weight of segment code)

10. Calculating organization level taxonomy results

To calculate the organization level results, the model sums up the segment level data. For example, to calculate the percentage of revenues that are not in scope with the EU Taxonomy at an organization level, we simply sum all the percentages not in scope for each business segment. Below is an example illustrating Not in Scope and Additional Testing Required.

11. Using organization TRBC activity code when no business segment revenue available

Since many companies don’t report revenues at a segment level, that can often leave gaps in the reporting of the overall portfolio. To overcome this limitation, the model uses the organization level TRBC activity code if no business segment revenue data is available. It checks to see if the code is available in the EU taxonomy. Since there’s no distinction from a segment revenue perspective, the script takes a binary approach and assigns 100% of revenues to either eligible or not in scope. There’s no further breakdown e.g. additional testing needed or % revenues aligned etc.

Do No Significant Harm & Minimum Social Safeguards

Understanding the percentage of a company’s revenues are eligible or align with the EU taxonomy is one part. In addition to this an assessment of Do no significant harm and Minimum social safeguard also need to be factored in. The model does this by downloading for all the constituents in the portfolio (and where ESG data exists) environmental and social controversies covering the latest fiscal year period. Recent controversies are updated weekly. Below is a list of all the controversies being captured, the first two focus on environmental controversies whereas all the remaining controversies are grouped under the Social umbrella.

Please refer to ESG Scores for description and additional details about these metrics.

DNSH (Do No Significant Harm)

To determine DNSH, the model looks at whether the company has made a public statement around the following criteria:

If a company has both a TRUE value next to any of the above metrics, and has an environmental controversy, then the model classes this as a red flag, which is shown in the generated excel report in the Organization Summary worksheet.

Minimum Social Safeguards

To capture minimum social safeguards, the model shows the sum of all social themed controversies per company in the Organization Summary worksheet report. Values for each controversy data point can be found in the worksheet titled DNSH data within the auto-generated report. The output for each data point is a count of how many controversies there were for a given topic, these are manually processed by a team of trained ESG controversy analysts.

References

• EU taxonomy of sustainable activities regulation
• TEG Excel tool
• Eikon Data API
• ESG Scores Description
• Pandas Dataframe