Simply sign up to the ESG investing myFT Digest -- delivered directly to your inbox.
There are two questions that every wealth manager will ask a client: What is your sensitivity to risk and do you prefer performance or ethical investments?
Whereas risk can be assessed on a 10-point scale, the issue of ethics has never been simple. Artificial intelligence can be used to facilitate investments that have a near-perfect match to the values of investors, but only if we stop trying to place environmental, social and corporate governance (ESG) into neat boxes.
To use AI, it is vital to accept that the values of an individual are too complex to build a taxonomy around. We also need to look at the ethical “fingerprint” of organisations we are considering investing in, and not simply tag them as ESG or not ESG, or lump them together in groups based on an arbitrary cut-off point on a scale.
The use of AI has taken off across in the investment industry. Robo-advisory services are widespread, while AI can be harnessed for back-office automation by processing complex data sets that would otherwise be uneconomical for humans to analyse.
For more than a decade, I have used AI and advanced data analytics techniques in the wealth and asset management industry, building and then selling a data-science consultancy.
We have the AI technology to model investor values and match these to the ethics of organisations in their portfolio, but there is too much emphasis on building uniform ESG models rather than embracing the natural complexity that the data gives us.
To truly personalise investments, we need to feed our engine with two main data sets: the individual values profile of the investor and the ethics profile of the organisation. The values of an individual are fairly easy to assess, but how do you measure the ethics of an organisation?
While I was the head of AI at asset manager Fidelity, we explored new approaches to measuring the ethics of technology companies. EthicsGrade, an ESG rating agency that I founded, is building on this idea using AI-driven models that create a more complete picture of the ESG of organisations.
Unlike classical approaches to scoring ESG, EthicsGrade is not bound by a set of fixed rules or criteria. By harnessing Natural Language Processing (NLP) — an AI tool — we can automate the analysis of huge data sets, such as tracking controversial topics, and evidence of governance and stakeholder engagement strategies in organisations’ public statements.
This gives us a score based on our assessment of what “good” looks like, but equally we can offer a personalised score based on an investor’s values.
At the heart of ESG should be the question of individual ethics. As an investor, I might have a preference for environmental sustainability, but be ambivalent towards a company that produces, say, motors for military uses. If your investments are in cloud computing, how do you know that the websites they host align with your values? Today’s simple ESG scoring does not help us match these preferences.
Current ESG ratings vary in all but one area: they all seek to take complex information and simplify it. There are many attempts in the market to create a more uniform rating system, but this is fraught with danger.
If the rating activity is on an arbitrary standard, designed by a predominantly white, male and middle-class elite, it will not reflect the diversity and individuality of investors, nor will it evolve organically with society.
Asset managers must recognise the advantages of the rich complexity of the ESG data they receive. They should use technology to match the complexity of investors with the complexity of ESG, rather than diluting ESG to the meaningless simplicity of stakeholder capitalism metrics. This will be driven by investor pressure as ESG comes under fire for not accurately reflecting investor values.
I look forward to the day where capital flows to organisations that best match the collective values of society, and evolves with them. But if ESG is thought of as simply good or bad, or scored as one to 10 against fixed criteria, then it will soon lose its appeal and miss benefits it can bring to business and society.
Charles Radclyffe is founder and chief executive of EthicsGrade, an ESG rating agency. He is a former head of AI at Fidelity International and a visiting fellow at Bristol University.