By Dr Pooja Khosla, VP Client Development, Entelligent

 

 

Monitoring and evaluation of investment risk is an integral part of the job of financial intermediaries such as banks and insurance companies. The global economy is set to lose up to 18 per cent in GDP due to climate change if there is lack of action from regulatory authorities and market stakeholders. There is no denying that we are currently experiencing a shift in priorities as global investors increasingly wish to see better climate risk disclosure. In parallel, however, investors and risk evaluators are concerned these disclosure metrics are only part of the story.

 

We see asset managers, from Acadian Asset Management ($110bn AUM) to BlackRock ($8.676tr AUM), building powerful artificial intelligence tools with a strong capability to screen public companies for climate risk exposures and resiliency. These tools may have the ability to empower engagement with firms suspected of failing to live up to their promises.

 

We also see new innovative investment products, in both ETF and annuity markets, which are designed based on the climate resiliency themes of modern risk evaluation tools. Amundi MSCI Climate Change Index ETF Series, Lyxor S&P 500 Paris Aligned ETF Series, and Society Générale’s SG Entelligent Agile 6% VT index (Fixed Indexed Annuity offering by Investors Heritage) are some examples that give us hope for more sustainable investment solutions.

 

But demand for products will require looking much more broadly at climate risk than the evaluation market has yet dared to. Asset managers must look at not only physical risk, but also transition, reputational and business risks. This broader scope of observation will require a wider and deeper analytical framework, and such a framework will inevitably generate inconsistencies based on the application of underlying metrics.

 

Quantifying climate change risk is not just about measuring a company’s carbon footprint; companies impact the climate, so does climate change impact company performance. It is about looking at how climate change impacts a company and how that company is likely to respond to future risk. In this light, inconsistencies are just bits of information we have not yet learned to extract.

 

Moreover, the effects of climate on the economy are not a smooth force. It is difficult to predict exactly how businesses will respond to episodic events such as fires, drought, floods, heatwaves, energy price shifts, as well as how markets are likely to perceive policy changes. The variation we see is a mixture of both, ‘real’ signals – that of companies responding to their unique combination of management response and exposure to climate change – and some ‘noise’ signals from extraneous factors within the wider economy.

 

Ideally, green investing should have the potential to integrate the Return on Investment (ROI) with Return on Environment (including both climate and social aspects) as one offering. With the existing demand from markets (through shareholder advocacy and engagement) and socio-political developments (TCFD, SASB and Green Taxonomy), has created a need to develop multiple contextually-sensitive metrics and reporting standards. Today, there are over 100 providers of ESG data serving institutional and retail investors. More information is a boon, yet often also a cause of confusion about applications in the space.

 

In this context, scientific tools such as reinforcement learning, artificial intelligence, and natural language processing, have a promising role to play.

 

With further innovation and technological sophistication, the contribution of AI and ML (machine learning) within the ESG space would enable us to actionably address the data gaps which make it difficult for asset owners or managers to assess long-term risks and rewards. It would allow for better integration, customisation, and scaling of ESG data for meaningful measurement. Ultimately, what can be measured can be managed.

 

Such technologies also have the potential to revolutionise ‘rational’ policy and reallocation of capital to businesses with good environmental governance. Detecting defaults, next-best action, and forecast analysis, are among the many challenges that machine learning can help solve.

 

Both the quality and quantity of input data matters a lot in AI and ML models, which are only as good as the data fed to them. For better models, we require access to better data at a fundamental level. Investors should keep the pressure on and continue demanding more comprehensive climate disclosures from corporations. That is the only means of securing the effectiveness of risk models.

 

Qualitative and quantitative analyses using modern technology work best when blended together. Qualitative research is always the beginning. It allows us to understand the problem broadly and in terms of the correct ESG objectives. This is followed by a switch to quantitative methods that help us understand how to actionably apply and scale this information into solutions. Finally, AI and ML methods help in the aggregation of this information. Ultimately, I envision AI/ML ESG solutions emerging as the core science behind qualitative and fundamental data.

 

With these unique and powerful climate risk analytics, we may be able to prevent wildfires, mitigate environmental risks, and continue down a low-carbon trajectory, harnessing the potential of capital markets to make our planet more liveable and resilient for decades to come.