Introduction

Artificial intelligence tools are helping asset managers fill the gaps in ESG datasets. Satisfying investor demands for green options, meeting the challenge of SFDR and the taxonomy, and avoiding the dreaded accusation of participating in “greenwashing” are tough challenges. A lack of comprehensive, detailed, reliable ESG investing data is the main concern. AI can help bridge this gap, using methodologies which can be explained to clients and regulators.

“Amongst our fund manager clients, there is a widespread perception that unless you include some form of ESG considerations in portfolio construction and investment strategy, you risk seeing investors shun your products,” said Sergio Venti, Head of Client Solutions and Innovation at Pictet Asset Services. He was speaking at a recent industry panel discussion “EU Taxonomy: Quant and AI-based solutions”.

De facto labelling

A consequence of SFDR is that funds have positioned themselves on the spectrum ranging from article 6, though to articles 8 and 9. In so doing, the “light green” article 8 option “has become something of a defensive strategy; the minimum viable product you need to not lose ground,” Mr Venti added. Yet to achieve this, asset managers are working in an unfamiliar environment, where they often lack the regulatory expertise, data and technology they need to keep pace with the green investing revolution.

Applying the taxonomy to the investment strategy is a particular challenge. There are questions of the eligibility of investments, whether these do significant harm other objectives, and, in the near future, there will also be social considerations to take into account.

Models fill the gaps

“It won't surprise anyone that the main challenge is data quantity and data quality,” said Patricia Pina, Head of Product Research & Innovation at the fintech firm Clarity AI. “An answer is to use existing data to build models which can give a fair estimate of the information which is missing,” she said. This information is accompanied with an assessment of the confidence the asset manager can have in each data point.

The challenge is considerable. Ms Pina noted that of the Clarity AI database of 40,000 companies, they estimate about one-third have eligible revenues aligned with the taxonomy, with only a few dozen reporting taxonomy related metrics. “We work to fill these gaps by combining different data sources to give an eligibility assessment for almost the entire database,” she explained.

According to Julián De La Cuesta, Senior Product Specialist at the fund distributor Allfunds, this is the kind of help investors are seeking when making their choices. “We have integrated Clarity AI data with our digital ecosystem Funds Connect, to enable our clients to identify the extent to which their investments are currently aligned with activities defined by the taxonomy,” he said.

Easing the ESG burden

On the fund side, Pictet Asset Services use technology to help their boutique asset management clients align their portfolios with the taxonomy. “We've developed an online assessment tool which shows clients, in a gamified way, the impact of being aligned with article 8 or article 9,” he said. To this are added oversight services for risk management, investment compliance monitoring, and investment restriction monitoring. There are also pre-completed templates, which include calculations around the taxonomy as well as digital visualisation of the portfolio.

The panel was aware that AI models cannot be 100% accurate, but still serve a clear purpose to fill the data void. “The Clarity AI data is presented with levels of probability, with the system enabling existing data to do as much as it can,” said Mr Venti. “Yet nothing can ultimately substitute for the human judgement exercised by the asset manager who will take the final decision,” he said.