‘We discovered 1320 new Teslas’

Published on: 25 March 2019

Entis and APG reach one-year mark


In less than twelve months’ time, smart algorithms scanned 10,000 companies for their contribution to a better world. Gerben de Zwart, Director of Quantitative Investment, looks back on the first year of Entis: the team that uses artificial intelligence to generate higher return and boost sustainable investment.


A good pension in a sustainable world. To achieve that goal for ABP and the other pension fund clients and their participants, APG uses both traditional investment methods and innovative technologies like artificial intelligence. These smart algorithms can identify sustainable companies as well as pick up on hidden clues to help investors assess opportunities for return and risks earlier and more accurately. To speed up the pace of innovation, APG acquired Deloitte’s data analysis team last year: thirteen women and men who apply artificial intelligence and big data to sustainable investment. The team continued as a separate APG business unit known as Entis. ‘This is the Ferrari of investment innovations,’ says Gerben de Zwart, whose quantitative investment team works closely with Entis. He tells us just how quickly things have accelerated in a single year.


What has been achieved in the past year? 
‘In just one year, Entis categorized 10,000 listed companies around the world in an SDI classification system: how sustainable are their products, and can they be considered a Sustainable Development Investment (SDI) on this basis? We used the United Nation’s seventeen sustainable development goals as a starting point for this evaluation. As a responsible long-term investor, we want to know whether companies are working on SDGs and if so, how seriously they take them. To find out, Entis scanned all annual reports, websites, and Chamber of Commerce registration details of these 10,000 companies using artificial intelligence and machine learning.’


Was it a lot of work? 
‘It was an enormous job. But if we had had to do this manually, this task would have taken much longer or been impossible altogether. Take the theme ‘affordable and sustainable energy:’ smart algorithms were able to analyze all sources for key words like ‘solar panels,’ ‘carbon emissions,’ and ‘climate change’ at lightning speed. This information was then compared to sales and revenue figures. This allows you to determine the extent to which the revenue from a company’s products and services contributes to the SDGs, to a better and more sustainable world. A business may claim that its activities are green, but sometimes their actions do not back this up. A combination of these algorithms and human supervision can see through such so-called greenwashing. Entis’ SDI classification system helps us as investors to identify front runners when it comes to sustainability. We were looking for a new Tesla, and we ended up finding no less than 1320 of them! This information is unique and cannot be found anywhere else in the world.  This is confirmed by conversations we have had with fellow pension funds: not just in the Netherlands, but also in Japan, Scandinavia, and Australia.’


Any other highlights?
‘Artificial intelligence can also be used to get a more accurate estimate of risk and return. Entis built an infrastructure to scan text and language use in annual reports and other mandatory business reporting. We processed over 400,000 documents in all. For example: by looking for similar products you can cluster companies in a different way, across traditional sector lines. This helps you map out companies’ real competitors. A company like Facebook, for example, competes with both IT companies and media companies, and car manufacturers are facing competition from unexpected places too, like Google’s self-driving car. This insight helps investors assess the performance of companies in this new competitive field as well as the risks more accurately.’


What other applications are there? 
‘Another example is scanning annual reports for textual changes compared to the previous year, for example in the risk paragraph. This typically indicates issues and future share price drops, according to Lazy Prices, a report by Harvard researchers analyzing the annual reports of American listed companies. However, most investors only respond to this type of hidden signs months after the fact. So investors who do keep an eye out for these negative - and sometimes positive - clues can gain an advantage. In the past year, Entis used intelligent algorithms and machine learning to scan the annual reports of over a thousand American companies for textual changes. A third project focused on mapping the diversity of company boards. A study from Tilburg University showed that companies with a high level of diversity on the board achieve a 0.4 percent higher return per month than companies whose board is virtually homogenous. Entis reproduced this study, but did not consider the results sufficiently convincing to include diversity as a criterium in our investment strategy as yet.’


What else has the Entis team worked on in the past year?
‘Aside from generating insights for APG’s investors, the Entis team transferred its proprietary technology platform from Deloitte to an autonomous environment in the Cloud, including the accompanying innovative technology. The technology platform offers important added value. For example, the platform has significantly sped up document scanning: processing time was reduced from weeks to days. The next step is to expand the existing platform to the production environment. In future, this will allow the investment policy for our pension clients and their participants to benefit from Entis’ innovative insights on a daily basis.’


What challenges have you faced in the past year?
‘Using artificial intelligence for successful and responsible investment on the long-term is a new idea. As a result, we constantly have to forge our own path. Moreover, there are enormous amounts of data involved: analyzing this takes a lot of time. Staying focused is another challenge. We selected just a few projects to start off with, but we have thirty innovative ideas on our shelf.’


What tangible results have been achieved so far?
'Knowledge is power in investment. And Entis has been very busy gathering knowledge this past year. Assigning an SDI classification to those 10,000 businesses was an important milestone. Aside from that, we are currently investigating how the patterns identified by the algorithms could improve the return of our investment strategy. This analysis should determine whether we want to use this information in our investments. We want to consider this carefully: after all, we are now talking about a quantitative equities portfolio of around fifty billion euro under management on behalf of our pension clients.’


What can these innovative investment methods offer our pension fund clients and participants?
‘In the near future, we expect artificial intelligence and Big Data to facilitate better investment decision-making, helping us deliver more pension value to the funds. For example, it allows us to identify SDI companies and help achieve our clients’ sustainability ambitions: the 1320 Teslas offer us the opportunity to invest in this type of high-potential business early on. That way, innovations in quantitative investment can help contribute to higher returns and thereby better pensions for the participants: our predictions of when to invest in which companies are getting increasingly accurate, and so is our assessment of risk.’


So Entis was worth the investment?
‘Absolutely. The role of unstructured data in investment is increasing. Our acquisition of Entis has allowed us to skip ahead three years compared to where we would be if we had had to do this alone. With the SDI classification system we now have in place, we can take a leading role in sustainable investment for our pension fund clients across the globe. We are now exploring the possibility of granting other pension funds access to the classifications.’


How will investing look in the future - will humans still be involved, or will all investment decisions be made by algorithms and robots?
‘We are firm believers in the combination of people and computers. Computers can increasingly take over search and analysis tasks. This allows portfolio managers to work more efficiently and frees them up to make investment decisions. They will continue to perform manual checks, and their expertise and experience will be a key factor in the process. So even in a future with artificial intelligence, humans will remain a constant.’