APG takes the lead on AI

The €473 billion ($575 billion) APG, Europe’s largest investor, is constantly looking for new ways to innovate that can enhance its processes and decision-making to benefit the 4.5 million members it serves.

For many investors, the sophisticated application of AI, machine learning and big data remains a fantasy, but APG has had data scientists as part of its investment team since 2016.

The company-wide evolution started about three years ago, when chief operating officer of APG AM, Marcel Prins, asked the question: “What will the workplace environment for an investor look like in 2020?”

This led to AI becoming a core part of the conversation at the group level as the fund looked to change its business model.

It was important to the fund that the changes not become just a top-down exercise; technology innovation had to develop into a core function of every department and every decision.

In APG’s 2020 vision, innovation and investment technology will become a key driver for pension fund success. Specifically, the fund’s leadership thinks ‘investech’ will give portfolio managers new and more data, leading to unique insights, better operational efficiency and, ultimately, better returns.

APG head of quantitative equities, Gerben de Zwart, says that from an investment point of view the team looks beyond the mystique of AI and homes in on what it can bring to the client. Much of the fund’s client focus centres on sustainability, and de Zwart says AI can lead directly to better responsible investments.

An impactful acquisition

Last week, APG announced it would take over Deloitte Netherlands’ data analytics activities for sustainable investing. The acquisition gives APG an edge in a number of ways. Firstly, the fund will inherit a mature data science team that is “ready for action”. It will also now have its own infrastructure, and a data engine with a smart algorithm that can extract information and turn it into actionable insights.

APG’s clients have set clear responsible investment goals they want to achieve by 2020, including: lowering the carbon footprint of the portfolio by 25 per cent; being invested only in companies that behave well; and doubling its allocation to sustainable development investments.

“There’s a lack of data on investments as identified by the UN Sustainable Development Goals [SDGs],” de Zwart says. “For example, we all agree with the goal of zero hunger…but it is difficult to quantify which companies are [actively working on] that. Machine learning and AI can help.”

APG and fellow Dutch investor PGGM have been at the forefront of converting the concept of the SDGs into an investible universe. The two funds developed a methodology to identify investment opportunities linked to 13 of the 17 SDGs, and refer to this methodology as the taxonomies. These are available for all investors to use.

“The beauty of this new machine learning is it can take those taxonomies and look for companies where they apply,” de Zwart says. “The Deloitte team can make an important impact on realising those goals and KPIs we have.”

APG also plans to leverage the infrastructure, data and processes of the Deloitte team and apply these to factor investing and other areas over time.

“We are very excited about this acquisition,” de Zwart says. “The acceleration we will realise from this innovation is a multiple-year acceleration, and we expect higher returns via more responsible investment.”

The Deloitte data team and infrastructure are a complement to the machine learning APG already has.

“We have an innovation team and an innovation portfolio, so it’s not one idea that will make a difference, but about 15 different ideas that could all lead to drivers,” de Zwart explains; for example, machine learning has the potential to save time on reading company reports.

The technology should take two analysts’ reports, read them and see if there is sufficient difference in the reports for the portfolio manager (PM) to bother reading both. If there are no sufficient differences, the portfolio manager can read just one report.

In the quant world, de Zwart says, using 10 different data sets, mostly centred on financial statements, has been the norm for 35 years. Now there are 500 different data sets available.

“Now that data is unlimited, it’s turning the daily work of quant PMs 180 degrees,” he says.

Dealing with data overload

But the abundance of data does not come without its challenges.

De Zwart warns other investors that a vision is required to lay out how to deal with innovation and, for example, prioritise which data set to use for what and when.

“How you deal with the overload of data is a question to ask, and it’s a journey,” he says. “We have changed the culture in the company to be open to new ways of doing things and looking for alpha.”

This is an important point to stress, and APG’s experience shows the entire company needs to be involved, not just a single team, in order to profit from collaboration and develop IT systems that get the most out of AI integration.

“Investments have become even more data-driven than ever. There must be a close relationship between the investment and IT departments,” he says. “The IT department will play a pivotal role. They are servicing PMs in a different way than they used to.”