State Super uses machine learning to combat negative cashflow challenges
State Super, a closed fund managing A$37 billion ($24 billion) in assets, is deploying machine learning to manage the rise of significant negative cashflow as the majority of its members enter retirement phase.
As a closed fund, which precludes new members from joining the fund, with regular pension payments flowing out to retirees, the fund cannot rely on traditional "ride-it-out" investment approaches, according to Chief Executive Officer John Livanas.
"We have a smallish team, and because we have significant negative cashflow, we cannot invest through cycles," Livanas told AsianInvestor.
The majority of State Super members are now in retirement phase.
"Our machine learning models absorb market information and attempt to identify where there are changes in correlations, changes in relative values, or breaks in trends," the chief executive explained.
The fund's approach represents a significant shift in how superannuation funds are deploying artificial intelligence (AI) and moving beyond basic automation to sophisticated market analysis tools.
NATURE-INSPIRED TECHNOLOGY
Livanas drew a parallel between the fund's machine learning approach and systems found in the natural world.
State Super
"An analogue from nature is the way compound eyes focus light in insects work to excite the nerve cells when there are changes in shapes or light between the compounds," he said. "They're nowhere near as accurate as human retina, but they are facilitating faster reactions."
ALSO READ: State Super CEO sees human touch as key to Australian pension reforms
The biological comparison underscores the fund's focus on rapid response capabilities rather than perfect prediction accuracy. The system generates signals that prompt the investment team to evaluate potential changes to key factors.
"Many times, the signals are not useful. But where they are, they lead to a discussion about changing our risk profile, looking at our derivative overlays, or watching our liquidity," Livanas said.
PROVEN RESILIENCE
State Super's technology-driven approach has already demonstrated its effectiveness during periods of market stress.
"During Covid-19 the fund famously never went negative on a year-to-date basis," Livanas said.
The use of protections came with a careful consideration of costs.
"Options premiums are expensive, and we have a discipline to pay for these by ensuring additional risk is taken, creating an asymmetrical payoff process," he added.
The approach is part of a broader trend in Australia's A$4 trillion superannuation industry, where funds have increasingly turned to technology to manage complex market conditions and regulatory requirements.
EVOLVING PRIORITIES
While State Super's machine learning capabilities have matured in the investment space, the fund is now focused on applying technology to its operations.
"I think our machine learning in investments will now just evolve. It's pretty good now. I'm not sure whether huge additional investment will provide a payoff," Livanas said. "We're focusing on robustness and governance. Our AI focus will switch to process improvements and customer technical support."
Other Australian superannuation funds have also taken steps to leverage technology for member services and operational efficiency.
According to JP Morgan's 2024 Future of Superannuation survey, improving member engagement and communication strategies ranked as a top priority for funds.
INDUSTRY CONTEXT
Increasing complexity in investment markets and rising regulatory requirements have put pressure on Australia's superannuation sector.
State Super’s adoption of technology-driven investment decisions is also notable given the challenging market conditions facing pension funds globally.
With interest rates at elevated levels and geopolitical tensions creating market uncertainty, funds have increasingly looked to technology to provide early warning signals of market shifts.
"We can't predict the future," Livanas noted. "And while machine learning helps us move much faster in protecting the funds, we also have an extensive, layered downside protection framework that protects in deep downturns."
ALSO READ: AI won’t ever make our investment decisions: State Super CEO
The fund's recent successful navigation of market volatility suggests that its technology-first approach could serve as a model for other pension funds facing similar cashflow challenges.
Looking ahead, State Super's strategic pivot toward customer-facing AI applications while maintaining its existing investment technology infrastructure could signal a new phase in how superannuation funds balance investment technology with member services.
"As we get into a more highly regulated environment, we need to be able to interpret regulations and policies in our operations," Livanas added.