Pioneering responsible investor Brunel Pension Partnership is using AI to improve its stewardship ratings and has spent much of the last year trying to remedy the misalignment of interests between asset owners and managers on climate stewardship.
According to the fund's recently released “Responsible Investment and Stewardship Performance Report 2024,” the team used an AI-driven tool called Generative Pre-Trained Transformer (GPT) to analyze and compare the voting guidelines of around 20 asset managers and owners.
The 30.8 billion pound ($38.9 billion) Brunel is using the insights to update its own voting guidelines to ensure they go beyond current practice and expectations.
“This is about understanding the wider changes in stewardship standards and reflecting that in the guidelines,” said Oliver Wright, head of responsible investment.
Brunel has also developed a quarterly report reconciliation tool that uses AI to assess the implementation of voting guidelines, and uses the generated reports as a tool to engage with service providers if any discrepancies are found.
“Given the importance of voting, the ability to automatically verify this information is invaluable,” Wright said.
“This ensures the reliability of our reports and significantly reduces the time and effort previously required for manual reviews, meaning we can devote more resources to engaging with portfolio companies and other core stewardship activities.”
Brunel believes the role of AI in stewardship will only expand, and as the technology advances, Wright expects to see new tools emerge for more effective engagement with companies, improved monitoring of sustainability factors, and even predictive analytics to identify potential governance risks.
The future of stewardship will require the further integration of AI not just to streamline operational tasks, but to enhance the strategic aspects of the work, such as focusing engagement and ensuring meaningful outcomes.
“GPT has already proven to be an asset and its continued development will undoubtedly open up further possibilities to enhance our management operations,” Wright says.
But while AI offers great benefits, Wright also warned about its risks, such as the potential for social bias in AI algorithms: Because AI systems are trained on large datasets that may include historical socially biased data, there are concerns that these systems could reproduce and even amplify existing social biases.
Added to this is the challenge of ensuring the fund’s reliance on AI does not diminish the value of human judgment, and maintaining strict data privacy and security. To mitigate these risks, the team regularly audits AI tools for bias, ensures operational transparency, and maintains a balanced approach that combines the efficiency of AI with the nuanced understanding of experienced professionals.
Addressing inconsistencies
In another notable trend, over the past year Brunel has been working with other asset owners to address the conflicts of interest between asset owners and managers on climate change management.
The 2023 shareholder meeting season showed signs that some asset managers have failed to clearly challenge oil and gas companies that are backtracking on climate change, the report said. This contrasts with the position of large asset owners, who shared the view that climate-related risks, if not addressed through stewardship activities, could lead to investment risk for portfolios and affect long-term unitholder returns.
To address this discrepancy, Brunel has engaged in “active and constructive dialogue” with managers to identify ways in which they can be better supported in delivering on asset owners' climate management strategies.
Brunel's analysis of the discrepancies and his conversations with other asset owners is based on research published by independent academic Professor Andreas Höpner. Using the energy transition in the oil and gas industry as a test case, Professor Höpner and his team assessed the voting records of selected managers from major oil and gas companies.
The survey provided evidence of misalignment. The full survey, released in November 2023, provided insight into trends in misalignment and voting rationale. For example, very few asset managers publicly state that they align their reasoning with asset owners.
Some asset managers perceive proxy voting and ESG engagement as mutually exclusive and are concerned that if the two are not aligned, they may lose access to management.
The study also found a clear divide in the types of engagement processes being used, ranging from sustained, long-term engagement with considerable progress to a “quick fix” or “jump on the bandwagon” style, pointing out issues with consistency and long-term approaches to engagement.
The study also provides some evidence for this gap and highlights the need for further research to explore these issues in more detail.
These include cultural mismatches (specifically between UK-based asset owners and non-UK-based asset managers) and misunderstandings of resource allocation (potential misunderstandings about the importance of stewardship and voting, leading to poor resource allocation).
Other reasons include misunderstanding fiduciary duties, particularly when it comes to managing risks related to climate change and financial conflicts of interest.
In the next phase of the project, asset owner participants will enter into one-to-one bilateral dialogue with managers based on the survey findings. The next phase will also explore how asset owners can express their views on climate management.
