Companies in China may play an important role in our client portfolios. As a global asset manager, we seek to understand Chinese companies’ environmental, social and governance (ESG) practices when making long-term investment decisions. New rules requiring Chinese companies to disclose more data are a meaningful step in the right direction, toward providing greater transparency for investors.1 Such disclosure is important in helping to identify the companies that are managing their sustainability risks and taking advantage of the opportunities to address the country’s carbon neutrality goals.
Yet data challenges persist. While the rate of climate-related disclosures by Chinese companies has risen in the past two years, they still lag global peers.
To enhance our understanding of our investee companies as active investors, we have created two proprietary machine learning tools that can help discover and distill a wide range of financially material ESG information, drawing on complex data and information sources.
Our proprietary gap-filling tool incorporates a diverse collection of metrics reported by the company, including revenue breakdown. One tool estimates ESG performance for companies that do not disclose it, using a diverse collection of metrics reported by the company.2 J.P. Morgan Asset Management’s Sustainable Investing team further refines the output, producing an ESG score that can fill data gaps using estimates (Exhibit 1).
Machine learning can fill gaps, with input from human experts
Exhibit 1: Our approach to generating estimates of ESG metrics for Chinese companies
Another tool our China-focused data scientists have developed aims to predict corporate controversies that are likely to affect risk-adjusted investment returns. Machine learning parses alternative Chinese data sources, which is complemented by local experts adding their specialized knowledge of the nuances of local language, law and industry sectors.
Controversies alone do not provide a holistic view of a company’s sustainability risks and potential opportunities, however. Strong policies could show a company’s awareness in managing sustainability risks.
Machine learning is not a substitute for company disclosures. That’s where meetings with investee companies come in. As active stewards of our clients’ assets, we engage with a target list of portfolio companies. However, these tools help identify additional timely discussion topics for meetings. In some cases, engagement has been not just informative but has stimulated the disclosure of more material ESG data by the companies.
As further laws and regulations take effect, opening the door to a larger volume of ESG metrics, they will need to be analyzed and understood to be useful. Over time, we anticipate that our tools will help better inform our understanding of companies in China as the volume of ESG disclosures grows.
1 Recent regulatory changes in China, as of February 2022, mandated that certain listed companies, their subsidiaries and certain private companies disclose their carbon emissions and any environmental penalties.
2 This tool uses a popular machine learning model to identify features that can be used to estimate missing company data points, called gradient boosting, a technique that generates predictions from an ensemble of weak prediction models.