Green-leader or green-washer? Using AI to prioritise company engagements

Introduction: provide a short overview of the practice, process or product that is being proposed for the award

Acadian’s investment universe comprises more than 40,000 securities globally. The size of the universe requires the firm to adopt a nuanced approach to company engagement simply because meeting company management isn’t feasible at large scale. This raises the question as to how a systematic investment manager can use engagement to offer differentiated insights to companies and enhance shareholder value to clients.

Acadian’s approach to engagement draws upon the firm’s strength in understanding data and building predictive models. ‘Structured’ (hard) data includes a company’s CO₂ emissions, statistics on gender diversity and ESG vendors’ policy scores that capture whether a company has a particular governance policy in place. By contrast, ‘unstructured’ (soft) data includes free-form text published in company reports, press releases, regulatory filings and AGM and earnings call transcripts.

By combining these datasets with machine learning models, Acadian has developed a tool that compares what companies say about sustainability versus what they actually do. The underlying premise is that a company’s actions speak louder than its words. The tool is used to prioritise engagement and offers portfolio managers unique insights that can be shared directly with companies.

Process, practice or tool: Provide a description of the innovative approach to ESG incorporation, its coverage within your firm, why you decided to undertake this approach and the value it provided preferably using a practical example of how you have applied your approach to an investment (security/issuer/sector/asset class/portfolio)

Acadian has built an Artificial Intelligence screening tool, called ENGAGER (Engaging on Non-Green company Actions Gathered by Evaluating management Rhetoric). The tool identifies and prioritises company engagements, distinguishing between ‘green-leaders’ and potential ‘green-washers’ associated with the firm’s conviction themes, namely the energy transition, employee well-being and management long-termism.

ENGAGER mines company documents and applies computational linguistic techniques to compare companies’ announced sustainability commitments associated with each theme. Commitments are assessed over time to determine whether a company’s rhetoric is supported with evidence via hard data points.

Specifically, ENGAGER scores companies’ sustainability plans to evaluate consistency, depth of messaging and whether a company is ultimately true to its word as measured by changes in tangible targets.

To take the example of Acadian’s energy transition theme, ENGAGER identifies whether a company’s plans are aligned to the EU Taxonomy and assesses the credibility of its plans based upon recommendations outlined by the Taskforce for Climate-related Financial Disclosures (TCFD) – namely, governance, strategy, risk management, metrics and targets. For instance, when mining earnings call transcripts, ENGAGER assesses the extent to which company management directly answers sellside analysts’ sustainability questions, whether they avoid difficult questions and whether the answers offer sufficient granularity to inform investment decisions. By combining multiple sources of information, ENGAGER cross-validates company statements at a large scale. This includes, for example, comparing companies’ environmental goals with actual carbon emissions.

Similarly, during Acadian’s engagements on employee well-being, ENGAGER identified an influx of companies adding sentences post-COVID to their regulatory filings emphasising their concerns for their employees’ physical health and mental wellbeing. By cross-validating companies’ statements with employees’ perceptions obtained from social media reviews, ENGAGER identified inconsistencies and selected companies to engage with where employee morale appeared low.

Outcomes, benefits, challenges and next steps: provide an example of the outcomes, outline the benefits and challenges associated with the introduction of this initiative and what you have learned from this approach that can be applied more broadly. How might you intend to develop the process or practice?

In Acadian’s view, the ability to cross-check companies’ sustainability discussions offers a novel way to identify both risks and opportunities within a portfolio. Its climate-aware funds, for example, penalise companies that score relatively poorly on carbon metrics and tilt towards companies that offer transition alignment solutions.

As companies increasingly announce net-zero targets, the challenge for investors is to assess the credibility of those commitments. The ENGAGER tool provides a systematic solution to do so. One example of ENGAGER’s ability to quantify the impact of greenwashing is illustrated by the detection of boilerplate statements in regulatory filings. Its sustainability research finds that, while a number of companies acknowledge climate change risk in their regulatory disclosures, many stop short of describing plans to mitigate or adapt to climate risk. By drawing comparisons across filings, ENGAGER can quantify the perceived impact of granular versus boilerplate disclosures on a firm’s cost of equity. This information offers portfolio managers a framework to directly engage with companies and to explain the importance of disclosures in valuation terms.

Most recently, the firm applied ENGAGER to analyse human rights issues in China. In particular, the tool mapped out customer/supplier linkages for a European car manufacturer. Acadian shared its insights directly with the company and asked it to cross-check its suppliers’ involvement in the Uyghur region. Acadian was told that ENGAGER’s analysis offered differentiated insights and that the outputs were of great interest to the company’s purchasing department.

One of the greatest benefits of the ENGAGER tool lies in its ability to mine information time-efficiently, uncovering issues that may be material for investment decisions.

One challenge is that it has proved harder to engage with companies in Asia compared with those in Europe and North America. This is largely because the contact details of companies in the region are often missing from their websites. For this reason, Acadian advocates combining direct engagement efforts with collaborative and third-party initiatives to further the advancement of firms’ disclosures.

In Acadian’s view, the ability to combine structured and unstructured datasets offers a way to generate novel insights. Currently, it is investigating ways to further quantify ENGAGER’s insights for firm valuation and it intends to develop the tool in this direction to enhance its dialogues with companies.