A study of gains from applying Generative AI to SFDR reporting
Asset managers are spending increasing time on sustainability reporting, driven by demand from investors for sustainability-related products and growing regulatory reporting expectations. Our research note examines how technology can support asset managers in reporting in a more robust and efficient way, ensuring their time can be spent innovating products that support end investors’ sustainability goals. Specifically, this research note reports on a recent field study of AI-driven automation among asset managers for periodic reporting under SFDR.
Sustainable funds’ interest grows
Investor interest in sustainable funds continues to grow. In the EU, the market share of Article 8 and Article 9 rose further to more than 60% of the funds (with Article 8 funds alone accounting for 57.6%), and reaching almost €6 trillion in value by the second quarter of this year. The Sustainable Finance Disclosure Regulation (SFDR) has been imposing tighter standards on asset managers through more detailed reporting requirements. Financial market participants located in the EU have had to comply with SFDR requirements since March 2021.
In January 2023, the technical standards came into force, requiring firms to disclose more granular data based on standardized templates for disclosures. The reporting process for financial products and entities involves disclosing data using different frameworks such as the EU Taxonomy, the percentage of sustainable investment, and the Principal Adverse Impact indicators (PAIs), contained in the SFDR.
More broadly, ESG regulations appear to be driving the asset allocation decisions of institutional investors across regions. Recent research (2024) has shown that around 60% of institutional investors expected instruments in public markets – especially green and transition bonds – to benefit from the new regulations, including SFDR, over the next five years.
The short-term ebb and flow of asset allocations do not seem to have affected the general direction of travel. Despite market volatility, sustainable funds generated better returns than traditional and exchange-traded funds last year, according to the IEEFA.
The Morgan Stanley Institute for Sustainable Investment reports that more than 70% of investors believe strong ESG practices can lead to higher returns, with more than half saying they plan to increase their allocations to sustainable investments in the next year. In fact, looking to the future, other recent studies indicate that sustainable assets under management are set to reach anywhere up to $40 trillion by 2030.
Demand and reporting standards escalate the complexity
The medium- to long-term positive trend for sustainable funds, combined with tight reporting standards designed to eliminate ‘greenwashing’, is putting pressure on asset managers’ analyst workforce.
According to a KPMG survey, financial services organizations are facing challenges in resourcing and acquiring sustainability talent. Almost 80% of respondents said they plan to restructure teams to better align ESG goals with business strategy, and 73% are or will outsource core sustainability reporting activities in the next three years.
In short, the growing reporting demand is making an impact on expert asset management personnel. Firms, especially small-to-medium asset managers, want to make their workforce more productive, focusing their expertise where it will produce the most value, and reduce their time spent on compliance processes. Dealing with regulatory reporting manually reduces the amount of time expert personnel have to innovate, analyse, enhance client service, and focus on portfolio decision-making.
Artificial Intelligence deployed to automate manual work
The burden of regulatory reporting like disclosures under SFDR can therefore be considerable and on the increase. Even though there should not be a great need for changes in the annual reporting, as the overall strategy of the fund does not change drastically over a short period, it is nevertheless expected that quantified fields will need updating, with similar qualitative
responses.
At the very least, this requires hours of work to track the changes and complete all the questions, especially for multiple funds. In an atmosphere of growing demand for sustainable funds, many asset managers are looking for ways to effectively automate such compliance/reporting tasks so their analysts can focus on more value- creating activities.
Generative AI (GenAI) is already being deployed to make this task automation happen. By acting as an intelligent assistant to the fund team, AI can reduce the time taken to achieve this mandatory reporting. The team only needs to provide the right input to ensure the quality of the output and do a final review before submission.
GenAI for regulatory reporting, powered by Clarity AI
Clarity AI has been deploying GenAI in this manner for regulatory reporting, starting with SFDR. The company already offers an end-to-end solution for Article 8 funds periodic disclosures. Article 8 funds currently represent 55% of assets under management in the EU, over 11,000 funds. The results of Clarity AI’s analysis are described in the remainder of this short research note, along with a calculation of the average time saved by expert staff.
First we can summarise the capabilities of the AI-driven automation of periodic reporting. The GenAI model uses the client’s uploaded periodic report of the previous reporting period, along with their answers to the questionnaire, and updates their data relevant to EU Taxonomy, SFDR PAIs, and Sustainable Investment (using Clarity AI data), to generate answers for the majority of the sections in the Article 8 periodic report. GenAI is able to complete these tasks via unstructured data from any source.
Key tasks performed by GenAI
- Automatic identification of specific metrics mentioned in the client’s previous period’s report, researching current value updates for them, and entering them in the reporting template
- Elimination of human error in the research process, reducing and focusing human time on review
- Conversion of short notes to full sentence inputs following the proper regulatory tone (from questionnaire inputs)
Two typical scenarios
- Update of periodic reports: If the fund’s strategy remains the same, GenAI uses the previous periodic report as a reference, and updates the quantifiable fields with the latest data from Clarity AI’s platform.
Result: No need to manually update the data for each report every reporting period – merely review. - Adaptation of periodic reports: In addition to updating the quantitative fields, if the strategy of the fund changes, GenAI updates the answers within the report to reflect the new strategy using the client’s input from the questionnaire. Result: No manual effort needed to rewrite the qualitative sections and update the data for the current reporting year – merely review
How much time is being saved?
What, then, is the workflow efficiency impact gained by deploying GenAI to the task of Article 8 periodic reporting? Clarity AI conducted a field analysis in mid-2024 to understand the impact of this technology on their current processes.
In each case, the time to complete Article 8 periodic reports was compared between existing workflow (not using GenAI) and new work practices (using GenAI). While experiences vary depending on the automation success rate, the average time taken to complete each Article 8 periodic report was reduced by over 80%. Using this statistic, asset managers examine their own sustainable fund reporting activities and calculate the potential benefits of applying GenAI capabilities to their Article 8 reporting, at the very least.
“I think every asset manager should use these techniques. And this is only for Article 8, if we can use this for all the regulatory reports that keep coming up, it will only provide more value,” Senior Portfolio Manager (€20M in Assets under Management)
Clarity AI’s four-step process for Article 8 funds periodic reporting
- Choose report type and language
- Upload a previous periodic report as an input for the GenAI models
- Once the user uploads a previous periodic report for reference, a short form is required to customise the answers
- Users can see a preview of the answers generated before downloading the report
Further developments are expected in due course to bring similar AI-driven efficiencies to all SFDR reporting requirements. Fund managers can use this new capability as a way to reduce the manual burden of regulatory reporting and dedicate the freed-up time to focus on expert personnel time that really adds value to the firm.
