- Machine Learning is leveraged for the construction of ESG ratings and socially responsible portfolios
- Quantitative methods can only yield results as good as the data provided
- Climate investing needs the synergies of human insight and leading technology
Overview
Quantitative methods play an important role in how we understand and approach financial markets. Thanks to developments in technology and computing, we have access to increasingly sophisticated analytical tools like Artificial Intelligence (AI) and Machine Learning (ML). AI, and especially ML, have come to dominate industry trends over the past two decades. Today, AI is used in many financial applications, from forecasting and fraud detection to algorithmic trading and banking chatbots. It has even made its way into the niche area of sustainable finance, where ML is leveraged for the construction of ESG (Environmental, Social, Governance) ratings and socially responsible portfolios. The widespread adoption of these tools has put their advantages over traditional approaches in the spotlight, leading many to ask themselves if the time of human analysts is coming to an end. Are machine methods capable of competing with human analysis and can they play a role in climate investing?
Rise of the Machines
Machine learning algorithms can process impressive volumes of data, more than a human ever could and much faster. Moreover, ML algorithms can adapt to changing data. ML shines in its ability to self-adjust through a trial and error process, in contrast to following precise instructions coded by a human. In this manner, ML can produce increasingly accurate results, as more data is fed into the algorithm1. Fraud detection and anti-money laundering have benefitted from this ML feature, which can learn and adjust responses to real and potential security threats.
NLP (Natural Language Processing) is a valuable new accessory in the analytical toolkit of the investment manager. It can be used to analyse the news, financial filings or social media, look for associations within a certain topic of interest and conduct sentiment analysis. Sentiment analysis can leverage news and analyst reports to gauge feelings around the value of a company, supplementing traditional quantitative methodologies, as well as other ML applications, for example high-frequency trading. NLP has also emerged in sustainable finance, to assign ESG ratings by reading news and company reports, build sustainability factors, and collect GHG emission data from company reporting and so on. It is able to fill in the ESG data gaps and track ESG company performance. This is especially valuable, as current ESG scoring methodologies lack standardisation.
DECODING MACHINE JARGON
The terms Artificial Intelligence (AI) and Machine Learning (ML) are often conflated and used interchangeably, it is therefore helpful to disentangle the two. AI is the broader theory and development of computer systems that are able to perform tasks traditionally done by humans. What we call ML is a subset of AI, which involves designing algorithms that optimise automatically through experience and with limited to no human intervention2. We can further distinguish ML algorithms based on the degree of human input into the process and the data labelling. In this way, algorithms fit into categories such as unsupervised learning, supervised learning, Deep Learning (DL) and so on. Each kind of algorithm has its advantages and uses both within and outside the field of finance. DL in particular has recently seen interesting applications in the industry, most notably Natural Language Processing (NLP), often used to read news and company reports.
When it comes to anticipating distress situations, humans can identify patterns that ML algorithms are unable to detect
While ML is very good at finding patterns from large amounts of historical data, it does not adapt well to extreme situations like natural disasters. Neither can it predict events disconnected from past data, such as the 2007 financial crisis3. Hence, when it comes to anticipating distress situations, humans can identify patterns that ML algorithms are unable to detect. In terms of data volume, machines can clearly beat humans, but what about more granular asset and company level analysis, which delivers better value? The answer varies. In cases of higher information asymmetry, for smaller firms with more intangible assets, human analysts do better. This is because understanding such cases requires profound institutional experience. When comparing machines and analysts making successful forecasts over the past five years, the hit split is 50-504. We can therefore conclude that human judgement is still beneficial and that without it, machines cannot deliver value. As of now and in the near future, these tools cannot replace human insight. Rather, value comes when using both approaches together.
Climate Portfolio Selection: Search Engines or Engineers?
What about climate investing, which has its own unique challenges? As outlined in the first climate investing paper in this series, we distinguish between two kinds of companies for climate transition: Mitigators and Enablers. Both play an important role in climate transition. Mitigators contribute to the transition through self-decarbonisation and enablers help by providing solutions that can support other companies on their decarbonisation path. So how can we find mitigators and enablers? We can take two paths. We can go the stock-picking route and try to pinpoint the companies that make positive contributions to climate, based on qualitative factors. Alternatively, we can embrace the quantitative approach and narrow our focus universe using some climate-relevant metrics. Here, the human and machine approach are at odds.
Mitigators: By their nature, mitigators call for a quantitative approach. A mitigating company has to decrease its own emissions in line with the objectives of the Paris Agreement and be a top performer within its respective sector. In addition, this company must belong to a sector or be involved in activities with high impact on climate issues; otherwise, its contribution to climate mitigation is insignificant. With the appropriate data, one can narrow down the investment universe to a list of eligible candidates by filtering out companies in low-impact sectors. For this, one can use the EU Low Carbon Benchmark Regulation definitions of high climate impact sectors based on NACE classification. Then, we can exclude companies that do not report on their GHG emissions or do not have Paris alignment objectives. After that, we can define investment objectives, such as the extent of alignment with the decarbonisation targets of the Paris Agreement, sector ranking in terms of GHG emissions, absolute direct and indirect emissions and so on. For these purposes, quantitative approaches are clearly better suited. Up to this point, the data needed to identify mitigators is numeric and thus quantitative. It is clear that there is no benefit in engaging in manual stock selection for transitioning companies at this level, a quantitative methodology can handle the large data volume much faster and more efficiently, assuming that it is reliable.
Unfortunately, to date, ESG data is heterogeneous and fragmented. With climate-related data, it is the same but worse
The efficacy of quantitative methods relies on data quality. Unfortunately, to date, ESG data is heterogeneous and fragmented. With climate-related 
data, it is the same but worse. Self-reported data is subject to biases and the same is true for forward-looking emissions plans; GHG emission data lags and scope three emissions estimations are often inaccurate. Quantitative methods can only yield results as good as the data provided. One could try to de-noise the data, even using ML tools, but one would have to deal with lagged data. This is not only the case with GHG emissions data. When it comes to company targets for emissions reduction in line with 1.5-degree decarbonisation pathways, company resources are often more up-to-date than any centralised data providers, who update their data on a quarterly or annual basis. Thus, when it comes down to measuring the real contribution of a mitigator to the climate transition, machine methods are not sufficient. The data is decentralised and often non-numeric. Determining whether a company is moving in line with its decarbonisation targets still requires a qualitative assessment, as it involves the direction of management, R&D, CAPEX, and moving away from certain GHG intensive activities towards more adaptive ones. In theory, it is possible to use NLP to collect the relevant indicator information from company reporting or websites, but these kinds of metrics are not reported in any standardised way, in contrast to financial data or even GHG emissions, so there is much room for misreading data. Hence, even though machine approaches are better for finding potential mitigators, human judgement is essential at the validation stage, especially if we are concerned with the relevance and the level of impact of the mitigating companies.
Enablers: Choosing enablers is a different kind of challenge. It cannot be done purely quantitatively, since it requires expertise that goes beyond simple computing. An enabling company has to be involved in an economic activity that can substantially improve the climate performance of other sectors or activities. This impact is difficult to quantify and finding firms that satisfy these criteria is very difficult with data alone. At the same time, it is neither efficient nor feasible from a resource perspective to engage in pure stock picking based on human analysis. The investible universe is far too large. As such, one can apply the same sectoral filters on the back of EU taxonomy to narrow down the pool of candidates, but no numeric targeting or filtering can identify true enablers. As outlined in the general case of the analyst versus machine debate, if understanding the firm requires extensive industry experience, analysts will outperform the machine. Enablers fall under this category of firms. They often are (though not necessarily) smaller in size and have more intangible assets. Much of the enabling technologies are nascent, such as carbon capture or hydrogen-powered aircraft. These companies are usually in the early stages of their business cycle, with a limited track-record. With such firms, one cannot leverage historical data to determine their potential performance; it requires the expertise of a human analyst, not only to determine whether the firm is a good enabler, but also to gauge scalability and profitability. In the case of enablers, it is clear that the human would win against the machine. However, for efficiency reasons, it is still necessary to filter down a potential pool of enablers and this part of the challenge is best approached quantitatively.
Putting It All Together
In the first climate investment paper in this series, we stressed the importance of approaching climate investing from two perspectives: mitigating and enabling the climate transition. Companies in high-impact and high-emission sectors must decarbonise, so that we can envisage staying within range of a 1.5-degree alignment. For this, mitigators need the help of enabling technologies, as otherwise the economy will not be able to decarbonise at a rate consistent with the goals set by the Paris Agreement.
With the current state of climate reporting, its lack of transparency and standardisation, machine-based methods are simply not enough for taking on the challenge of climate investing with the rigour that it requires. For this task, human expertise and judgement are invaluable for choosing both enablers and mitigators. Climate investing needs the synergies of human insight and leading technology. Therefore, to answer the question of human versus machine for climate investing, our answer is a confident, human AND machine.
1Robert C. Pozen and Jonathan Ruane. “What Machine Learning Will Mean for Asset Managers.” Harvard Business Review, December 3, 2019. https://hbr.org/2019/12/what-machine-learning-will-mean-for-asset-managers
2FSB. “Artificial Intelligence and Machine Learning in Financial Services: Market Developments and Financial Stability Implications.” November 1, 2017. https://www.fsb.org/2017/11/artificial-intelligence-and-machine-learning-in-financial-service/
3Robert C. Pozen and Jonathan Ruane. “What Machine Learning Will Mean for Asset Managers.” Harvard Business Review, December 3, 2019. https://hbr.org/2019/12/what-machine-learning-will-mean-for-asset-managers
4Sean S. Cao, Wei Jiang, Junbo L. Wang and Baozhong Yang. “From Man vs. Machine to Man + Machine: The Art and AI of Stock Analyses.” May 5, 2021. Columbia Business School Research Paper. SSRN: https://ssrn.com/abstract=3840538 or http://dx.doi.org/10.2139/ssrn.3840538
Important information
Past performance is no guide to the future, the value of investments, and the income from them change frequently, may fall as well as rise, there is no guarantee that your initial investment will be returned. This document has been prepared for your information only and must not be distributed, published, reproduced or disclosed by recipients to any other person. It is neither directed to, nor intended for distribution or use by, any person or entity who is a citizen or resident of, or domiciled or located in, any locality, state, country or jurisdiction where such distribution, publication, availability or use would be contrary to law or regulation. This is a promotional statement of our investment philosophy and services only in relation to the subject matter of this presentation. It constitutes neither investment advice nor recommendation. This document represents no offer, solicitation or suggestion of suitability to subscribe in the investment vehicles to which it refers. Any such offer to sell or solicitation of an offer to purchase shall be made only by formal offering documents, which include, among others, a confidential offering memorandum, limited partnership agreement (if applicable), investment management agreement (if applicable), operating agreement (if applicable), and related subscription documents (if applicable). Please contact your professional adviser/consultant before making an investment decision.
Where possible we aim to disclose the material risks pertinent to this document, and as such these should be noted on the individual document pages. The views expressed in this document do not purport to be a complete description of the securities, markets and developments referred to in it. Reference to specific securities should not be considered a recommendation to buy or sell. Investors shall conduct their own analysis of the risks (including any legal, regulatory, tax or other consequences) associated with an investment and should seek independent professional advice. Some of the investment strategies described or alluded to herein may be construed as high risk and not readily realisable investments, which may experience substantial and sudden losses including total loss of investment. These are not suitable for all types of investors.
To the extent that this report contains statements about the future, such statements are forward-looking and subject to a number of risks and uncertainties, including, but not limited to, the impact of competitive products, market acceptance risks and other risks. Actual results could differ materially from those in the forward-looking statements. As such, forward looking statements should not be relied upon for future returns. Targeted returns reflect subjective determinations by Unigestion based on a variety of factors, including, among others, internal modeling, investment strategy, prior performance of similar products (if any), volatility measures, risk tolerance and market conditions. Targeted returns are not intended to be actual performance and should not be relied upon as an indication of actual or future performance.
Data and graphical information herein are for information only and may have been derived from third party sources. Unigestion takes reasonable steps to verify, but does not guarantee, the accuracy and completeness of this information. As a result, no representation or warranty, expressed or implied, is or will be made by Unigestion in this respect and no responsibility or liability is or will be accepted. All information provided here is subject to change without notice. It should only be considered current as of the date of publication without regard to the date on which you may access the information. Rates of exchange may cause the value of investments to go up or down. An investment with Unigestion, like all investments, contains risks, including total loss for the investor.
Backtested or simulated performance: Backtested or simulated performance is not an indicator of future actual results. The results reflect performance of a strategy not currently offered to any investor and do not represent returns that any investor actually attained. Backtested results are calculated by the retroactive application of a model constructed on the basis of historical data and based on assumptions integral to the model which may or may not be testable and are subject to losses.
Changes in these assumptions may have a material impact on the backtested returns presented. Certain assumptions have been made for modeling purposes and are unlikely to be realized. No representations and warranties are made as to the reasonableness of the assumptions. This information is provided for illustrative purposes only. Backtested performance is developed with the benefit of hindsight and has inherent limitations. Specifically, backtested results do not reflect actual trading or the effect of material economic and market factors on the decision-making process. Since trades have not actually been executed, results may have under-or over-compensated for the impact, if any, of certain market factors, such as lack of liquidity, and may not reflect the impact that certain economic or market factors may have had on the decision-making process. Further, backtesting allows the security selection methodology to be adjusted until past returns are maximized. Actual performance may differ significantly from backtested performance.
Unigestion (UK) Ltd. is authorised and regulated by the UK Financial Conduct Authority (FCA). It is also registered with the Securities and Exchange Commission (SEC). Unigestion SA is authorised and regulated by the Swiss Financial Market Supervisory Authority (FINMA). Unigestion Asset Management (France) S.A. is authorised and regulated by the French Autorité des Marchés Financiers (AMF). Unigestion Asset Management (Canada) Inc., with offices in Toronto and Montreal, is regulated in Canada by the securities regulatory authorities in Ontario, Quebec, Alberta, Manitoba, Saskatchewan, Nova Scotia, New Brunswick and British Columbia. Its principal regulator is the Ontario Securities Commission.
Document issued April 2022.
