Introduction

Sustainability concerns are making noise in a post-COVID world, but the difficulties involved in extracting and analysing the non-financial information to facilitate informed decision making in ESG investing and sustainable finance present a challenge.

Artificial intelligence algorithms offer a solution by improving the efficiency and effectiveness of the data extraction processes, but the challenges of ESG reporting implies the element of human judgement is not going away soon. This article takes a closer look at this intersection of artificial intelligence and sustainable finance.

With the discourse on sustainable finance and ESG investing accelerating following Covid-19, the discussion surrounding the challenges to extract and analyse non-financial material data has also picked pace.

This is where artificial intelligence, a branch of computer technology that leverages data to automate complex tasks, holds tremendous potential to improve the efficiency and effectiveness of data mining and analysis.

Artificial intelligence (AI) has seen significant impetus in recent years as part of the 4^th Industrial Revolution to leverage digital mechanisms for socio-economic development; and it is only opportune to utilize its merits to deepen sustainable finance and ESG investing.

But how does AI revolutionize the way financial companies deal with data?

First, AI can make data extraction more efficient and most companies are utilizing AI in their ESG data research. Reporting frameworks for sustainability or non-financial data vary from each other, both in content and in their applicability per jurisdictions.

And since most disclosures are unaudited, the consistency of the reported data varies between companies, further complicating the extraction process and making intra-sectoral comparisons unreliable.

Moreover, a lot of non-financial information is often found on third-party sources or unstructured data sources, which adds to the time taken to identify and compile the information.

AI technologies can help scrape these multiple sources of information for keywords and data, thus extracting granular information at a far rapid pace as compared to human analysts.

That would also aid the comparability and consistency of the extracted information. Once the algorithm’s process is set, the AI bot can repeat it multiple times at fast speeds, despite the sources having large volumes of irrelevant information.

A part of this pertains to small companies, be it small-caps in ESG investing or smaller-issuers in sustainable finance fundraising. Most large companies anyway publish sustainability reports as per the industry’s reporting frameworks, often to boost their branding as a corporate citizen.

However, the small companies who often do not report material non-financial information present a challenge; and this is where the ability of AI to quickly extract relevant data and keywords from third-party sources becomes even more relevant.

Why so? Because as the interest around ESG investing and sustainable finance deepens, the financial sector will have to expand beyond the large-caps and large issuers to the smaller companies, just like small-cap funds are launched in traditional investing once the investor demand pushes the valuations of the large-caps to the roof or their fundraising needs are met.

Next, AI can make data analysis more effective. It is capable of qualitative and quantitative analysis.

Qualitative because it can correlate the financial data with the captured non-financial information to help the investing decision process, can make informed investment decisions based on this analysis.

Quantitative because it can power quants-based funds and indices entirely; AI is anyway being used heavily by the hedge fund industry for quants-based strategies.

Last, AI can enable profitable investing decisions. A secret sauce of making profitable investments is timing! Using AI will not only help the financial sector screen companies for non-financial information faster with more accuracy than fatigued human analysts.

This combination of speed and accuracy would lead to better timing in investments, thus translating into improved profitability!

Despite its merits, there is a need to exercise caution.

First of all, technology cannot be used in isolation. Given the inconsistencies between the data reported, disclosure methods and sources of information, we cannot just rely on technology.

Some human intervention is still needed for judgement and analysis in order to understand the accuracy of the information.

An element of engagement with the company reports is also needed to dig deeper and reaffirm the methodology behind the reported data.

Perhaps as AI technologies evolves further, the human need will be eliminated. But we are still some time away from that.

A follow-up argument to this human vs. technology debate also connects with cost. The US and European financial sectors have seen significant mid-office (analyst-level) work offshored to emerging markets like India, Philippines and Sri Lanka which offer a cost advantage in terms of the analyst workforce.

This needs to be compared with the probable cost-saving the sector can gain by using AI technology for data extraction instead of human analysts, however this will make sense only with extremely large universe. This arithmetic cannot be forgotten for sure.

Second, AI technologies that exist today for ESG data extraction are still unable to solve the issue of greenwashing, an eternal criticism the ESG space faces given the inconsistencies in reporting frameworks and the voluntary nature of the disclosures.

Most non-financial reporting are unaudited, fuelling further the fire of greenwashing. AI technologies that can tackle this issue will be the real game-changers in this business.

Next, AI algorithms are strengthened as more and more historical data is inputted for that algorithm to learn from (called machine learning in technology parlance), which reduces its error-rate overtime.

Since ESG and sustainable finance is still evolving in most geographies, there may be a dearth of historical data; hence the output of the algorithm would have an error-rate. That also necessitates human supervision. Also, simply depending on historical data may not suffice.

Also, many regulatory changes (ergo regulatory risks) on sustainability are still anticipated. That necessitates the fund managers to take a forward-looking view as well, not just use technology to scrape historical data.

Finally, there is still a talent gap at the intersection of technology, statistics, finance, sustainability, and social sciences, all being critical components to build a sustainable future.

Even many Board members are still not fully conversant with ESG guidelines. That makes the human intervention tough as it is.

But despite these immediate headwinds in depending excessively on AI technology, the long-term advantages offered by AI offer a clear rationale to weave it within our sustainable finance and ESG investing initiatives.

As the abilities of automation and algorithms improve, the day may not be far when AI even becomes capable to not only identify and extract the relevant information, but also audit the material information and deliver critical insights without any human intervention. And with the post-COVID economic slowdown reorienting the focus of companies and governments towards lean-cost solutions, the long-term cost advantages offered by AI might also offer an economic rationale to do so.