By Charles Sutton, Financial Services Industry Lead, EMEA and Prabhu Ramamoorthy, Developer & Partner Relations Manager, Financial Services and Technology Team, NVIDIA
Learn how AI is helping investors make more sustainable financial decisions by gathering and analyzing ESG metrics.
Sustainable finance, which involves considering ESG (Environmental, Social and Corporate Governance) factors when making investment decisions, has quickly become a trillion dollar catalyst in the financial industry.
With the ESG industry now as large as 40 trillion dollars, various financial institutions, such as investment banks, asset management organizations, mergers & acquisition firms, countries, and regulators are being influenced by its growing size and importance to take ESG metrics into account in their operations.
The purpose of measuring an organization, or country, on its Environmental, Social and Governance metrics is to understand how advanced it is with sustainability for socially conscious investors. For example, if someone were looking to invest in an electric vehicle or battery firm, taking an ESG approach would mean investing in a sustainable, climate friendly company over comparable automotive firms that don’t have the same environmental credentials.
Gathering Data with Natural Language Processing
Closely linked with behavioral finance, ESG data metrics are not the same as market data metrics. Equities and options, for example, would usually be provided by market infrastructure data providers such as Bloomberg. Unlike market data, the industry has not standardized the way ESG data is shared. The way it is presented depends on each individual organization, and as such alternate data analysis is required.
This is where artificial intelligence can help, more specifically with a process called Natural Language Processing (NLP). NLP is a form of AI that enables computers to process text, and interpret the meaning of words, sentences, and paragraphs contextually. For ESG, NLP can be used to read, analyze and interpret data from various sources to present data on ESG factors.
If we look back to the example of searching for an electric vehicle or battery firm to invest in, NLP could be used to mine data based on different keywords, such as electric vehicle, battery technology, solar, green, democracy, and governance. These terms can identify key ESG information about companies in the automotive industry. The information for the data search could come from any text source, such as quarterly or annual filings, and both private and public news, either from a particular batch of data or by using real time, evolving data.
Financial analysts will be familiar with the collection of public, non-public, and non-material information about a company to determine value and make recommendations. The same collection of information needs to be done for ESG data, which is where NLP’s automated analysis techniques can be used to collect and identify data against ESG metrics.
Compiling ESG Metrics with AI
As the data mined for understanding ESG metrics comes from multiple sources, it is compiled as alternate, or unstructured data. Leveraging this unstructured data using NLP is the key to success for analyzing and understanding ESG data.
Not all organizations will present their information in the same way, especially relating to sustainability. In our electric vehicle example, an organization dedicated to building new energy vehicles might have significantly more data available about their battery technology, in various formats, than an organization who isn’t as advanced.
NLP techniques for identifying ESG information often require large amounts of AI compute power to process the vast amounts of data gathered. Building an NLP algorithm requires a pipeline of AI models for named entity recognition and knowledge graphs, along with techniques such as sentiment analysis, semantic search, and summarization. Deep learning NLP algorithms such as BERT, GPT-3 and other algorithms can be used for NLP ESG analysis, with BERT and GPT-3 rising in popularity in the financial industry.
By using GPU-powered NLP models, keyword and sentiment analysis can be carried out on the unstructured data to pull together structured data as the ESG metric output, which can then be used by downstream financial models for financial decision making.
Driving Sustainable Finance with NLP
ESG is still a relatively new trend in financial services, along with AI, natural language processing and deep learning.
But this growing trend of ESG is changing the financial industry. Today, companies and financial trends are being evaluated from an ESG perspective for financial decisions. Using NLP can speed up the process of gathering important, relevant information for analyzing important metrics.
For organizations looking to implement their own algorithms, the ease of doing so will depend on its AI infrastructure and ability to manage the data. Top investment banks and market makers are already working with unstructured data and NLP to drive their ESG initiatives.
As the world becomes ever more aware of the challenges with sustainability, organizations are increasingly realizing that they need to do their part by making sustainable investment decisions that benefit the world. AI is accelerating almost every industry and combating the world’s greatest challenge – a sustainable future.
