The «gold rush» moment of ESG in finance reveals that data models used by banks and insurers to gauge progress simply aren’t ready.

Mark Twain said «during the gold rush its a good time to be in the pick and shovel business.» There is no greater gold rush in finance today than ESG, or environmental, social, and governance criteria.

Federal Reserve governor Lael Brainard, who wants the U.S. to follow other central banks into action on climate change recently noted this will require «substantial work to address current data gaps.»

No wonder data vendors see ESG emerging in finance as their pick-and-shovel moment. However, banks and insurers need to think differently about the solutions they adopt.

Past Performance Mantra

There is unlikely to be one golden source of ESG data but more a tapestry of threads. Clarity on definitions from regulators will be important, but ESG involves a qualitative element as well.

Financial and software models will need to work hand in glove with human industry expertise and judgment. A crucial challenge is the need for banks and insurers to focus on forward-looking data.

In fund management, the mantra that past performance is not necessarily an indicator of future performance is enshrined into regulatory disclosures. By contrast, lending and underwriting decisions are heavily weighted towards extrapolating historical trends.

Stranded, Uninsurable Assets
 
At some banks, «early warning systems and procedures for assessing borrowers’ unlikeliness to pay are overly reliant on ineffective indicators, outdated ratings and backward-looking information,» European Central Bank supervisory member Kerstin af Jochnick said last month.
 
Disruption caused by ESG, technology and supply chain changes are reducing the meaning of historical reference points and segmentations. They are creating so-called stranded assets, such as companies with an increased probability of default and properties near wildfire and flood zones that are becoming uninsurable.

Social inflation has made predicting the outcome of legal interpretations in insurance claims especially difficult in the U.S.

Challenging Collection Of Data
 
Despite industry leaders warning of record natural catastrophe losses in 2021, few insurers are assuming that the increased scale and frequency of such events continues in their pricing and capital models. The impact of surging inflation on re-build costs will only add to the complexity of these calculations.
 
Banks and insurers can no longer lean on internal data from counterparty histories or a small number of external sources. Collecting and aggregating data and developing comprehensive risk assessments is now more challenging.

In the U.K., the Financial Reporting Council has criticized companies for not sufficiently incorporating ESG metrics in annual reports. In reality, a wide number of online unstructured sources are necessary to understand the whole ESG value chain from suppliers through to partners and buyers.

Machine Learning Vs Rules-Based
 
Conventional model techniques used by banks and insurers look for a linear cause and effect from one piece of data. But as the ECB noted when presenting the results of an extensive climate change stress test of European banks exposures, such risks have a tendency of increasing «in a non-linear fashion».

Whether it is the 2008/09 crisis or Covid-19, recent history shows sudden shocks cause the greatest disruptions. Advanced machine learning-type algorithms can be better suited to extracting signals than rigid rules-based algorithms are.

Correct Signals In Noise
 
New technology and more processing power can help parse through the huge amount of unstructured data online, but there are no magic bullets. Armies of in-house data scientists and technologists has typically failed to solve the data dilemma effectively – or economically.
 
Just like the hedge fund managers in the movie «The Big Short» were able to piece together the relevant data points to predict the financial crisis, banks and insurers will still need to extract the correct signals from all the noise. And their ability to do so and constantly fine-tune their risk profiles will be a competitive advantage.
 
There are always multiple variables at play in driving outcomes and proving causation rather than correlation is as much an art as a science. After all, real life is messy. This will be even more so in ESG. 

Rupak Ghose is the operating chief of Galytix, a British fintech specialized in integrated data ecosystems for banks and insurers. He began his career on CSFB's trading floor in 1998 before moving into equity research. From 2003 until 2011, Ghose was Credit Suisse's research analyst for diversified financials, during which he was one of the sector's top-ranked buy-side analysts. After leaving the Swiss bank, Ghose was head of corporate strategy – including mergers and acquisitions – for Icap from 2013 until 2016 and then for Nex, now part of the CME Group, until two years ago.