In late April the Nature Action 100 announced the release of their Company Benchmark Indicators - 50 metrics across 17 sub-indicators and 6 overarching indicators. Benchmarks can be notoriously expensive and time consuming to construct hence at Neural Alpha we have developed a number of AI-powered solutions such as our Responsible Capital Disclosure Assistant to automate and scale these kinds of frameworks. To run an ESG assessment on a company typically requires an ESG analyst 3-5 days to sift through typically dozens, sometimes hundreds, of financial and ESG disclosures. Fortunately AI tools such as ours reduce this time from days to hours enabling analysts to focus on higher value tasks such as interpreting results and ultimately making effective decisions. We were keen to see how our AI would perform in automating the construction of this benchmark dataset - we’re pleased to share the results here.

Baselining coverage, volume & variety

Although ‘only’ covering 100 systemically important companies for nature given many of the constituents are leaders in corporate disclosure data volumes would be an immediate challenge for manual, human approaches (cue AI).  For this assessment we ingested over 22k documents with the volume of disclosures varying greatly by sector and company. Glencore PLC (LON:GLEN) lead the way in terms of volumes with over 4.8k documents with many corporates disclosing in the hundreds and a small minority disclosing less than 5 documents.  Once acquired the disclosures were indexed into our Large Language Model (LLM) to enable analysis.

The quality of the disclosures varies greatly by sector and company with many disclosures standards featuring including GRI, SASB / ISSB / IFRS, TNFD and more specialist industry standards to name a few. Refinitiv ESG scores, although not covering the entire list of companies provide a useful proxy for disclosure volumes and maturity and the below chart correlates with the dispersion we see in terms of volumes and disclosure depth across industries.

Delving Deeper: Employing An ESG AI Copilot

Within Responsible Capital we maintain an ESG-fluent AI language model with access to a Knowledge Base of millions of publicly available financial and  sustainability disclosures for the largest global companies - annual reports, company policies, financial statements, ESG reports, etc. There are now dozens of different types of disclosures and with frameworks such as TNFD and ESRS / CSRD being adopted this is ever growing.  This knowledge base can be queried using chat-bot style questioning (for a single company) or by creating a series of assessment questions to run on a watchlist or portfolio of companies (here, the Nature Action 100). To access this knowledge base and run your own ESG assessments, you can visit responsiblecapital.io (psst, demo accounts are free).

For this assessment we employed a binary framework - True or False with every assessment result grounded in both short hand context summaries and referencing the provenance of the assertions to specific disclosures. Recognising the nuanced nature of corporate disclosures using an "Uncertain" category is essential to reflect ambiguous or sometimes contradictory corporate positions on particular nature and biodiversity topics. This result is provided for example when keywords related to the question are identified by the AI, but the answers are ambiguous such that we can't confidently give an unambiguous True / False score. 

So for a question like "Does the company's commitment explicitly extend to the company’s value chain?",  the output means:

True - The company's commitment explicitly extends to its value chain.

False - The company's commitment does not explicitly extend to its value chain.

Uncertain, Likely False - Some evidence of a position or keywords mentioned, but not enough information to definitively answer True or False. A pass or fail here will likely come down to an individual investor's criteria leveraging the contextual information our AI returns or further assessment questions.

This nuanced approach ensures a comprehensive analysis, and highlights assessment areas warranting closer inspection, as different investors will have varying criteria for what constitutes a passing effort. 

We then converted the benchmark indicators into a series of questions, similar to a scorecard. These were framed in such a way to expect True or False answers, avoiding phrasing questions as “To what extent…”, “In what ways…”, or “How often are..”. In a later assessment, we will specify more open ended questions and ask the model to answer with evidence from disclosures. 

For this article, we’ve omitted the disclosure snippets that informed the answers from the results and focused on analysing the best performing, worst performing, and most ambiguous aspects of companies performance with regards to the indicators. 

All in all, the assessment and analysis took one person a few hours to complete - freeing up the rest of our team for their critical tasks.

Unveiling the Results: Leaders and Laggards

Perhaps not surprisingly, the companies with the best results are also some of the largest on the list. This could indicate progressive action on their part to manage the interface of their operations on nature, but may also be due to their relative sizes and abilities to dedicate resources to reporting and disclosure.

Metrics

Best Performing: Metric 6.1d The company engages its end-user consumers in a shift towards products, services, and/or behaviours with lower nature-related impacts and dependencies.

Metric 6.1.d stands out as the best performer, with an impressive 81% of companies achieving a positive rating. This metric underscores companies' efforts to engage their end-user consumers in adopting products, services, and behaviours with reduced nature-related impacts and dependencies.

Worst Performing: Metric 2.1.b The company publicly discloses all locations of assets and activities within the upstream portion of its value chain that are situated in or adjacent to ecologically sensitive locations.

In terms of the weakest performing metrics, results weren't as close. Metric 2.1.b emerges as the weakest performer, with nearly 70% of companies failing to disclose assessments of their material dependencies on nature in the upstream portion of the value chain. This highlights a crucial area where companies need to enhance transparency and reporting practices.

Most Ambiguous: Metric 4.2.e The company publicly discloses that it requires its tier 1 suppliers to recognize and respect the rights of
Indigenous Peoples and local communities and to obtain their free, prior, and informed consent.

Metric 4.2.e presented the most ambiguity in the assessment, with half of the companies on the list receiving an "Uncertain" score. This metric, which pertains to companies' public disclosure of requirements for tier 1 suppliers to recognise and respect the rights of Indigenous Peoples and local communities, underscores the complexity and need for clarity in supplier engagement practices.

In the evaluation of metrics, distinct trends emerge: Metric 6.1.d stands out as the top performer, with an impressive 81% of companies receiving a positive rating. This metric highlights efforts to involve end-user consumers in adopting environmentally-friendly products and behaviours. Conversely, Metric 2.1.b emerges as the weakest performer, with nearly 70% of companies failing to disclose assessments of their material dependencies on nature in the upstream value chain, indicating a need for improved transparency. Metric 4.2.e presents the most ambiguity, with half of the companies receiving an "Uncertain" score regarding supplier engagement practices related to Indigenous Peoples and local communities' rights, emphasising the need for clearer guidelines in this area.

Sub-Indicators

Best Performing: Sub-indicator 3.3 The company publicly discloses its progress toward its targets on an annual basis.

With a staggering 80% of companies disclosing their progress towards annual targets, this commitment to transparent reporting underscores a collective dedication to accountability and progress tracking.

Worst Performing: Sub-indicator 2.1 The company publicly discloses the location of all assets and activities in its direct operations and upstream and downstream value chain that are situated in or adjacent to ecologically sensitive locations.

Conversely, sub-indicator 2.1 faced notable challenges, with 57% of companies receiving a False rating for their disclosures. This sub-indicator, focusing on the public disclosure of assets and activities situated in or adjacent to ecologically sensitive locations, highlights an area ripe for improvement.

Most Ambiguous: Sub-indicator 4.2 The company respects and upholds the rights of Indigenous Peoples and local communities.

Among the sub-indicators, 4.2 emerged as the most ambiguous, eliciting a significant number of "Uncertain" responses. With 34% of companies providing answers that lacked clarity or assurance, addressing indigenous rights and local community engagement presents a complex and nuanced challenge for corporate responsibility frameworks. These results were focused into 4.2b and the aforementioned 4.2e - metrics for which the answers were Uncertain for 49% and 50% of companies.

In the evaluation of sub-indicators, notable trends emerge: Sub-indicator 3.3 showcases strong performance, with 80% of companies transparently disclosing their progress towards annual targets, reflecting a collective commitment to accountability and progress tracking. Conversely, Sub-indicator 2.1 encounters challenges, with 57% of companies receiving a False rating for disclosures regarding assets near ecologically sensitive areas, indicating room for improvement in this area. Sub-indicator 4.2 presents the most ambiguity, with 34% of companies providing uncertain responses, particularly regarding indigenous rights and local community engagement. This underscores the complexity in addressing such issues within corporate responsibility frameworks, highlighting areas requiring further clarity and assurance.

Indicators

Best Performing: Ambition (Indicator 1)

Ambition emerged as the standout performer, with nearly 80% of companies demonstrating robust disclosure, largely driven by their commitments to mitigate key drivers of nature loss and actively engage in the conservation and restoration of ecosystems. This underscores a collective dedication to environmental stewardship throughout operational processes and value chains.

Worst Performing: Governance (Indicator 5)

Conversely, Governance emerged as a focal point for improvement. A notable 35% of all metrics within the Governance indicator received a False rating. Weaknesses were particularly pronounced in metrics 5.1.b and 5.2.b, with 65% of companies failing to meet the criteria for board oversight and expertise to oversee issues pertaining to engagement with Indigenous Peoples and local communities.

Most Ambiguous: Implementation (Indicator 4)‍

The Implementation indicator presented a mixed picture. While 23% of responses featured keywords relevant to the questions posed, a lack of comprehensive information hindered definitive scoring. This underscores the need for greater transparency and clarity in reporting practices to facilitate accurate assessments of implementation efforts, and to avoid the slide into greenwashing.

At the indicator level, the assessment revealed notable patterns: Ambition stands out as companies demonstrate commitments to environmental stewardship, with nearly 80% showcasing robust disclosure, while Governance faces challenges, with 35% receiving a False rating, particularly in board oversight. Implementation presents a varied landscape, with 23% featuring relevant keywords but lacking comprehensive data, suggesting a need for clearer reporting. This analysis underscores both the commendable effort to sustainability and the areas for improvement in governance and transparency. It emphasises the importance of accurate assessments and the prevention of greenwashing through clearer reporting practices.

Conclusion

In conclusion, using an AI copilot to run an assessment of the Nature Action 100 benchmark indicators offers valuable insights into the ongoing efforts of systemically important companies to nature and biodiversity in managing risks, impacts and opportunities. The benchmark provides a solid, sector agnostic foundation for building more detailed sector and issue specific assessments incorporating more contextual aspects such as materiality and 3rd party data sets - features at the heart of our Responsible Capital platform.

While certain metrics have shown commendable performance, such as engaging consumers and disclosing progress towards nature-related targets, others reveal areas in need of attention, notably in governance and supplier engagement. These findings highlight both achievements and areas for improvement across the corporate landscape.

If there is one key message to take away from this rapid, AI-powered, high level assessment, it's that organisations are still relatively early in their nature-positive transitions.  In conjunction with climate transition paths, companies need to build on high level commitments and start to implement concrete action and detailed tracking, particularly with respect to embedded supply chain impacts.

This short exercise has not been commissioned by the Nature Action 100 team and has been undertaken on an independent basis to illustrate the power of AI-powered ESG Assessments.  If you would like a demonstration of the Responsible Capital platform or access to this dataset we welcome feedback - please reach out to us at info@responsiblecapital.io.

Ultimately, the journey towards enhanced corporate responsibility is an ongoing one, requiring collaboration and commitment from all stakeholders. Looking ahead, the future direction of AI-powered ESG analysis could involve integrating a materiality lens, enabling companies to prioritise ESG factors based on their significance to financial performance and stakeholder interests. Moreover, making benchmark assessments more sector-specific would enhance comparability and relevance, tailoring analysis to the unique challenges and opportunities of each industry. Additionally, contrasting company disclosures with data from third-party sources such as news articles could provide deeper insights and validation, offering a more comprehensive understanding of corporate behaviour and its impact on ESG performance. Through collective efforts, we can strive towards a future where sustainability and economic prosperity are mutually reinforcing pillars of corporate practice.

Closer to home, our analysis has also proved to us the power of the Responsible Capital platform. What would take a team of research consultants many days to process was assessed, with consistency and without bias, in a matter of hours by a single analyst, and with detailed provenance for each answer to check for nuances and to dig in to further details. For corporate financiers, asset managers, ESG teams and researchers that need to conduct rapid due diligence, including first pass assessments on frameworks such as NA100, Responsible Capital is a powerful accelerator.  

‍

Links:

Responsible Capital: reponsiblecapital.io

Nature Action 100 Corporate Benchmark Indicators: https://www.natureaction100.org/media/2024/04/Nature-Action-100-Benchmark-Indicators-2024-1.pdf