How AI can transform climate Investing

• Artificial Intelligence and climate analytics can be a perfect marriage because of the former’s ability to crunch complex datasets
• AI can bring more clarity to ESG data and analytics, which in turn will help advisors craft more customized portfolios

The most difficult and most important data challenge the world faces is arguably in capturing and measuring carbon emissions which are driving catastrophic climate change. Experts widely agree that climate and carbon analytics are challenging due to the major differences in carbon intensive business operations and products.

For investors, who asset allocation decisions will shape the low carbon economy, one of the biggest hurdles in calculating climate analytics is the sheer amount of disparate data sets that need to be curated, collated, and integrated. When analyzing climate profiles for an industry, the basic concept is to gather all the material carbon impacts across the three major scopes of the value chain and calculate the emissions generated per unit of product or output. If the emissions per unit of product or output are going down while the economics of the company are maintained or improved, then that company can rightly claim to be successfully transitioning its business.

This seems simple at a high level, but these “key performance metrics” for different industries often require multiple unique data sets that need to have rules-based adjustments to combine them with broader data sets coming from more traditional sources. Considering there are over two dozen carbon intensive industries requiring these levels of granularity, one can see how this becomes a massive data crunching exercise.

This would require huge resources in terms of time and manpower. But it is a perfect fit for the capabilities of AI.

In order to bring about standardization—which many view as the missing link to useful and credible climate analytics—one must first bring about real transparency. This involves the gathering of all the relevant climate impacting factors a company generates in bringing a product to market along with all the impacts those products produce in their useful lifetimes.

This involves collecting a variety of data—some of which the company itself may not be able to produce. These data sets are often beset with their own issues with gaps and anomalies imbedded within. AI can help cast a wide net of information gathering via scraping bots and can then pull the data into a central repository where it can be normalized with other data sets.

Within that normalization process, AI algos can be geared to craft estimates, interpolations, and extrapolations to cover off data gaps as well as pinpoint and correct errors and outliers. These types of calculations need to be very bespoke to the company or industry in order to be as close to accurate as possible, so they require immense amounts of time that humans may find difficult to do at scale. An engine that captures and combines these data sets will bring us as close to real transparency as possible on a near real-time basis.

With more comprehensive data collection and analytics calculations at hand, more meaningful, quantitatively driven metrics tracking corporate transition will emerge that will stand up to regulatory and investor scrutiny. From these metrics, industry best practices will be surfaced, and the true corporate transition leaders will be recognized. If these advancements come to pass, it will inherently highlight the more effective disclosure requirements needed for these transition tracking metrics and that will, in turn, result in more standardization of data inputs from industries.

As data becomes more standardized, it will bring more legitimacy to analytics (ratings, scores, etc.), which will drive more regulatory agreement and regional alignment which will then drive more required and standardized disclosures. This will effectively enhance transparency across the value and supply chains of businesses. This is the utopic vision of the future long shared by many in the climate analytics space—and AI can help get there.

Not all data sets come from corporate disclosures (or NGOs, non-profs, think tanks etc., for that matter) and not all are disclosed in similar calibrations that enable easy combinations with other data. Rules-based data science comes into play and can be very time consuming to do at scale across different industries. When you consider the amount of different data sets needed to cover off the carbon intensive industries alone, the time and effort it takes to pull them together and incorporate the integrating calculations and techniques needed to combine them, one can see why different analytics providers may take different paths to do that in order to make it more of a repeatable and viable ongoing process.

These different providers then make decisions as to what elements of this curation and integration process they want to take on which ultimately has a lot to do with the differences in their outputs. It is these differences that have led to confusion and criticism of climate analytics and ESG investing in general,  AI will enable the processing of this immense data exercise more tenable and scalable across the different providers. 

AI will particularly help with the complicated problem of collecting and including scope-three data elements across carbon intensive industries. Scope three are often among the most important profile elements in carbon intensive industries—while also being more difficult to collect and combine with scope-one and -two data sets. AI can scan unique content from a variety of sources on a continuous basis creating the most up-to-date content feeds for the calculation engine. One example would be capturing the ever-elusive methane emissions of companies and their products by using AI to constantly track and attribute satellite methane capture to its sources. Satellites and other new emissions capture techniques are yet more examples of collection and calculation intensive exercises that could be scaled to near real-time with an assist from AI.

A further AI data crunching benefit is the related exercise of consolidating and assimilating large amounts of data into coherent narratives focused on a specific task or theme. In this case, the task is to develop scores, ratings, and descriptive analytics for companies and investment funds. The skill set required is similar to those parsed by analysts and junior staff at law firms, consultancies, and I-banks when they craft their briefs and industry overviews—all of which could be “AI-ed” over time to drive more productivity for these businesses, and one can envision the same benefits for  climate scoring analytics.

The newfound, AI-driven ease in collecting and calculating these large data sets will lead to more agreement of company-level climate footprints which are at the core of scores and ratings—particularly as it relates to combining company-level scores into mutual fund scores.  This agreement will drive a homogenization of core climate metrics as a broader group of analytics providers will be able to process equal amounts of data and, though multiple iterations across vendors, the better methodologies will start to rise to the surface and become de facto standards. Thus, ESG scoring and ratings could become aligned across providers who rely on similar materiality approaches in a manner similar to bond ratings agencies with only nuanced differences among them.

As the more accurate analytics begin to surface, the  disclosures needed to calculate them will come more into focus. Firms not disclosing in a fashion that best enables them to tell their ESG and climate stories within these preferred methodologies will start to adjust their disclosure practices.

Perhaps more importantly, regulators will have a clearer understanding of the important disclosure elements for each industry so they, in turn, can craft regulatory guidelines to ensure all companies are disclosing accordingly (i.e. more standardization). If all these evolutions come to pass, then AI can help bring about enhanced credibility and usability to ESG scores and ratings.

Over time, the scoring and ratings providers can all do their calculations from a similar data set, which should lead to more homogeneity of their outputs. Any differences in their outputs from that point on will be more a result of methodological choices designed for differentiated use cases.

Bringing standardized climate analytics into the fundamental tool kit of portfolio managers will enable them to gauge the managerial effectiveness of any given company in articulating and executing against a strategic vision of beneficial carbon transition. They will be able to track whether companies are effectively managing risks—such as regulatory risks or direct climate-impact related risks—as well as taking advantage of opportunities in growth and productivity such as premium pricing, enhanced efficiencies, or the “greenium” they can achieve in capital markets financing. 

All these things combined with traditional fundamental analysis will enable a manager to see which companies are taking leadership in transitioning their businesses to align in a world where climate change continues to shape so many aspects of day-to-day life. Simply sticking to the valuation analytics of the past will be missing the more important parts of any company’s story—namely, how are they positioning their businesses for a changing future?

Along these lines, new factors of performance will likely emerge, similar to factors such as quality, value, size etc. Additionally, new attributes of companies will start to take shape, such as primarily green-driven revenues or carbon neutral value chains etc. These new factors and attributes can then be used to identify positive aspects of a company that investors could incorporate into their stock and mutual fund selection—not just because they are “climate aware” factors, but because they represent strong and forward-looking management teams. It’s a well-known investing axiom that companies with thoughtful, effective management tend to outperform over time. They are also inherently more ready for what’s to come given they are actively managing towards it. This will be the added value that more standardized AI-driven climate analytics can bring to investment managers.