Monday, 27 May 2024

MioTech’s Tu: Our technology enables real-time ESG data for China

5 min read

By Chris Georgiou

Hong Kong-based fintech MioTech is using algorithm-driven AI and machine learning to find and provide structure to ESG data in China.

With 800,000 private and public Chinese companies covered, one-third of MioTech’s revenues now comes from its environmental, social and governance (ESG)-related business activities such as ESG corporate data and ratings, real-time ESG risk-monitoring as well as portfolio analysis.

The Hong Kong-based fintech conducts ESG analysis through its proprietary shareholding, supply chain and investment knowledge graph, which searches for controversies using alternative sources such as social media and news.

Its algorithm-driven artificial intelligence (AI) and machine learning services tap into databases to spot patterns among the swathes of unstructured data to provide over 120 ESG data points from over 12,000 public sources.

It also offers AI-driven screening of prospective wealth management clients for money laundering and other concerns.

Currently, only around 10% of listed companies in China publish sustainability reports. The China Securities Regulatory Commission (CSRC) has decided to shift from voluntary to mandatory ESG disclosure for all A-share listed companies in 2020, although it remains unclear whether there will be a common taxonomy or unified format for disclosing ESG.

Li Ka-shing’s investment company, Horizons Ventures, joined by an undisclosed publicly-listed media company, led a new Series A+ funding round in early January for MioTech, following the previous $7 million funding round in 2017.

With financial advisors often citing concerns over the potential for ESG products to be mis-sold to them, CEO and co-founder Jason Tu explained further in this interview how MioTech can help asset allocators make informed decisions in China.

Chris Georgiou (CG): How does MioTech use artificial intelligence to fill the ESG data gaps in China?

Jason Tu (JT): From the perspective of a technology company, there are two types of data: structured and unstructured. Structured is usually reported in annual reports and corporate social responsibility (CSR), for example, which contain several inherent flaws.

Firstly, it’s infrequent. It’s only published once a year or quarterly and in the form of a post-event summary. Second, it’s very subjective, particularly for a CSR report as there’s no common standard.

So we focus on real-time events by using alternative data such as news, abstracting topics from social media, from government websites, third-party detection agencies and quality-control agencies as well as climate and water usage data. This means we can provide real-time updated objective data.

CG: Only approximately 10% of your data is derived from companies’ self-reporting. How can investors trust your data?

JT: First of all, we disclose all our data sources. There are also a number of verification mechanisms for data from social media or news. This is something that technology can help you solve. We have drawn out a very large technology-based knowledge graph on entities and individuals in China, so when that additional piece of information occurs, we can try and fit what we call the attributes of the data into the existing knowledge graph and see how it merges.

Without getting too technical, there are also other expert-rule-based verification mechanisms. For example, in terms of a carbon penalty, it shouldn’t be larger than the market cap of the company. If we get data that says a company is being fined by the Environment Protection Agency for several billion dollars, it is obviously wrong. This is the simplest example of expert rules which we have set up internally to manage the data.

CG: With so much of the data and media censored in China, especially in relation to scandals, we’ve seen many occasions where posts regarding scandals and corruption have been deleted online or hushed up. How does your system account for that?

JT: That’s a good question. I think from two perspectives. First, there are two types of scandals that I see reported in the media. Political scandals will be deleted instantaneously, or anything related to political personnel. However, based on our understanding, especially climate and environment related news, these are not being deleted. Although we have seen a tendency for official media in China to be delayed. They are not going to give you first-hand information.

For example, we have a major bank which uses our system to monitor ESG risks of some of their portfolio companies. In 2019, one of those companies was involved in a massive food poisoning scandal in northeast China resulting in 200 people getting food poisoning. The first signal we picked up was on social media complaining about this event. There wasn’t any official media report until two weeks after the event.

We will not give up on media as a lot of the cases are reported by the media but with greater nuisances in China such as being delayed or toned down if considered sensitive by the Bureau, or removed completely from the web if very sensitive.

CG: So what percentage of confidence can you give to your ability to rate scandals in China?

JT: We call it controversy. We get a good number of information and sources from the news, so we are fully confident on that. Having said that, we also have a number of sources that we cross-check. And we also collaborate with third-party watchdogs who constantly publish their reports.

CG: In terms of scale, in China there is not yet a common taxonomy. The data is often unstructured in PDFs across several websites. How many companies does MioTech monitor and how does your artificial intelligence get into PDFs?

JT: Currently, we cover around 800,000 companies. We started with client requests, on Chinese public companies (A-Shares) as well as in Hong Kong and the US. We then realised that most of the Chinese shareholding entities are super clean, so you need to go through the shareholding structure of the subsidiary, the supply chain and the customers in order to find whether there are notable events.

That expanded from around 4,000 Chinese-listed companies to around related 400,000 companies. Then we started monitoring large private companies such as Huawei, then the unicorns following a lot of requests from clients. Later, we also worked on municipal bond issuing agencies, as they are one of the hot sectors covered by the credit rating agencies. That’s how we started, by peeling the onion piece by piece to build up our database. 

A lot of people think that the Chinese government and China does not have data – but that’s the totally wrong concept. China has a lot of data and the government mandates that from the state, provincial and municipal level. However, it’s done extremely poorly. Municipal websites can have really messy texts on their websites and that’s the best case. PDF is not the worst. Sometimes you see JPEG photo files of a document.

That’s where we need to do image recognition to abstract the data and then you still need to parse the data to make sense of it, topic abstraction and what we call “entity recognition” to figure what entities are being mentioned, what topic it is, how much money they have been fined or what kind of legal cases are involved. That’s quite a cumbersome job, but thanks to machine learning and AI we can process it much faster.

CG: Which metrics do you use for E?

JT: We have our own methodology, derived from European and American standards. However, China requires a different set of standards. For example, among the common metrics across the globe such as environmental penalties and greenhouse gas emissions, China has its own idiosyncrasies.

For example, some of the state-owned enterprises are “totally different types of animal” here in China. They will have all types of CSR such as tree planting activities that they boast about as well as donations to charity which are actually mandated by the state. These are all the measuring points that you would get from different sources. Social and governance components are even more different in China.

CG: There is a plethora of rating and information providers entering the ESG scene now, including well-established players such as MSCI ESG. Why would potential users choose to use MioTech for their China-based portfolios?

JT: To be honest, MSCI covers about 600 companies in China and have just expanded and we cover over 800,000. Also, we don’t rely on CSR or public reports. I have a lot of respect for MSCI. But for China, especially when it comes to ratings, it needs to have its own ratings methodology that is accredited by industry participants including regulators, listed companies and fund issuing agencies and these different parts need to come to a consensus on a China specific standard.

I think MSCI will eventually need to localise as many other industries have proved that China is a different animal.

We had a customer from Europe who came into our office, with one of their requirements being thermal coal data in China. Their fund mandate does not allow them to invest in anything related to thermal coal. What’s more, they are not allowed to invest in any company which generates more than half its electricity from thermal coal. So far, they have only invested 22 billion out of their 60 billion budget for Asia because they have no way to select their companies.

CG: Can you provide your clients with solutions regarding this issue?

JT: Yes, we have thermal coal and electricity usage data gathered from different sources. And that’s the power of asset allocation. If there is less and less money directed into that industry, from a financial perspective, it forces them to transform.

CG: China is a signatory to the 2015 Paris-Climate Accord. Does MioTech have any mechanisms to ensure that companies are Paris-compliant?

JT: That’s the first time I’ve heard of that concept. It’s very interesting. Perhaps that’s a certificate we could work on developing. I wouldn’t say we have 100% of the data needed but we do have a huge number of datasets.

CG: Do you have any forward-looking solutions or is your data primarily backwards looking? As I mentioned, being compliant with the Paris Accord. For example, even if it is a dirty company in terms of carbon, but it is making the right transition, do you offer any forward guidance in this sense?

JT: So, there are three parts of our business. The first and the largest part involves the data. Most of the larger asset managers would rather have the data so they can make the analysis themselves. The data goes back as far as the 1960s, while the majority is from after 2005.

Second, we have ratings.

The third product we are trying to launch is a non-traditional consultancy service. If we have the data that shows that certain companies are not doing well, we try to dig into the causal affects from the datasets to try and come up to a conclusion. That conclusion can either be communicated to companies that are not doing so well or to interested regulators or relevant agencies.

CG: Who are your customers?

JT: Banks in China that are issuing the green bonds. On the buy-side, that includes sovereign and mutual funds as well as hedge funds and private equity clients. Family offices are also buy-side and we also have quite a number of sell-side customers. Seventy percent of our customers in terms of numbers are from outside China .

The latest Series A+ funding round will go towards hiring more employees and boosting the expansion of Miotech’s services across South East Asia and China as investor demand for ESG standard ratings in the region intensifies.

Leave your Comments
Recent Comments