On June 12th, 2018, FSI was delighted to host three expert panelists to talk about how artificial intelligence can help with the integration of ESG factors into investment processes and decisions. The panelists were Jérôme Basdevant, CTO and Co-founder of Datamaran, Sebastian Brinkmann, Europeen Executive Director of TruValue Labs and Valérie Cecchini, Vice-President, Portfolio Manager at Mackenzie Investments. The session was moderated by Milla Craig, President of the FSI Board of Directors, and President and Founder of Millani.
Jérôme and Sebastian introduced the subject by describing how their companies’ products use natural language processing to analyse free-form text, turning it into datasets that can then be mined for insight about the value of a company, in the case of TruValue Labs, or what is material to a company and its investors, in the case of Datamaran. Valérie explained how frameworks such as SASB’s are essential to structure the enquiry, using a common language to focus attention on the material issues for specific sectors of the economy. The SASB materiality matrix was developed after extensive consultation with investors and industry, as well as other stakeholders.
Jérôme underscored that it was not only artificial intelligence, but also recently emerged technologies such as cloud computing and the availability of big data, that has allowed this new generation of analysis tools to emerge. The question of materiality is central to all decisions and needs to be firmly embedded in analysis tools. Sebastian confirmed its importance in explaining the valuation of a company, despite information asymmetry in the data sets.
Another growing issue is the problem of the potential poor quality of the information analysed: datasets that are too small, gaps in information filled with proxies, errors in company reporting. Sebastian underlined the importance of due process in collecting information and the need for transparency around accuracy of results. And of course, more disclosure by companies around ESG. It is a process requiring continual improvement.
Valérie considers there to be a risk when a company choses not to disclose certain information. The lack of disclosure itself can be interpreted as negative by a third party and not reflect actual practice. On the other hand, only 80% of data disclosed is considered relevant, the rest not being financially material. And this despite the common understanding that only half of the value of a company is explained by the information presented in the company’s financial statements.
Jérôme and Sebastian agree it is vital to maintain the confidence of clients and all partners. The question of transparency around the use of AI is often brought to the table. Data providers try to give access to raw data as much as possible, and explain their methodologies to clarify the limitations of the processes and contextualize the results.
Valérie added that ESG analysis helps put a value on the intangible assets of a company. That in itself can justify the use of active fund management over passive management where electronically traded funds follow an index. But ESG data are not yet formalized; there is a need introduce standardization to disclosure, while including the particular context of each company. The aim is to improve the way data can provide insight to help analysts make better investment decisions. But, as Valérie suggested, everyone has their own definition of a good investment in line with their fund purpose and investment strategy. Another weakness is the lack of historical information. At best, ESG data go back five years.
A recurring theme is the perception that AI is going to replace financial analysts. Sebastian indicated that IA is rather another tool in the financial analyst’s toolbox, enabling more potential to be unlocked. The technology is evolving at a fast pace and will be very different from what it is today. The next innovation is likely around voice recognition, so that information could be extracted from investor calls or other dialogue. All of the panelists are convinced that the use of AI will be common place in the near future.