10th February 2022
Richard James, Civica Pensions Product Director, explores the untapped potential of machine learning and how it can drive positive change in the pensions industry.
Machine learning (ML) is not a new concept. We’ve been using it since the 1970s. With the availability of more and more data, increased computing power and easier access to ML tools, there’s never been a better time to tap into this technology than today. If 86% of local government leaders think machine learning has the potential to benefit their sector, why aren’t we seeing it being implemented more?
Many organisations are reluctant to use the term ML with words such as ‘predictive analytics’ or ‘behavioural insights’ often used instead. So, why the hesitation? Maybe it’s because, as with every new technology, people can fear the unknown. Done wrong, when fraught with bias (the ‘dark side’), ML can deliver negative outcomes for customers, eroding their trust, and damage your organisation’s reputation. But done right, it can improve efficiency, help prioritise resources, and gain valuable insights we might have never known.
Local Government Association (LGA)
The most important benefit to come from using predictive analytics [machine learning] is that it enables councils to deliver better outcomes.
Despite the level of work that can be involved in managing and reporting on pension data, good data management is vital to the success of any pension fund. Imagine then, the advantage funds would have if ML could deliver extra, crucial data about members and employers: could member movement analysis be solved by machine learning to help understand short term liability horizons between valuations? Could funds be more aware of trends, and make predictions to help them make better business decisions? Are external payroll providers focussing enough on member data quality? These are just a few examples of questions that can be answered now by a biological machine (i.e., humans like you and me). But what could ML bring to the table?
Machine Learning can leverage digital technology, allowing funds to decrease fraudulent risk, whilst reducing costs, and making better data driven decisions.
ML offers many positive opportunities, which I like to refer to as the 3 Ps for pensions.
ML could drastically improve the efficiency of pension funds by mapping out what is happening at work, compared to what should be happening. This could allow funds to identify their pain points and improve on time constraints, and efficiency savings (or lack thereof!). In an uncertain world this would be paramount in the quality of decision making, particularly when we think about the fluctuations that funds face with the impact of COVID-19. In the current climate, it’s more important than ever to meet the needs and expectations of funds and their members.
With blended working now the norm, staff need to be ready to react anytime, anywhere. ML systems comprise algorithms that, supplied with large datasets, ‘learn’ how to perform a set task by identifying patterns, connections, and common indicators, which means longer tasks can be done in a seamless manner, with less resource and time required.
As with any smart technology, ML relies fundamentally on high quality, reliable data. Data allows funds to identify patterns, behaviours and break down barriers. Predicting these risks, peaks and issues would make it easier for pension funds to be successful.
Predicting the behaviour of members could optimise fund and member interactions, enabling them to better understand what members would like to achieve via self-service. You’ll know how members move and when.
Wouldn’t it be valuable to use the analytics to assess past trends, and over time, find out if there are seasonal variations at work?
If ML can help improve productivity, and predict trends and threats, couldn’t it also help prevent threats? This answer is yes. ML could help identify the behaviours common to fraudsters and point to current claimants acting in similar ways, triggering additional checks and validations. Our industry is exposed to the fraudulent risk associated with typical ID&V processes more than others. But with sophisticated identity verification checks, that use biometric technology and quality data, organisations can be confident that they know the identity of the person and any risks associated.
Fed up with repeatedly answering the same enquiries face-to-face or on the telephone? ML can enhance chatbots and help quickly settle service users’ most common problems – reducing the need for front desk staff and call handlers and supporting social distancing.
With better understanding of the person and their identity, it’s easier to prevent fraud, while making it more efficient to quickly let the ‘good’ people in.
With machine learning market projected to reach a value of $17bn by 2027, ML is undoubtedly going to be at the heart of Civica’s digital future. For the pension industry, it offers real opportunities to enhance productivity, predictability, and prevention, driving positive change, learning from data to create models that humans wouldn’t be able to build otherwise. Pension funds can then focus on the member experience and best use of staff, whilst constantly raising the bar to meet new expectations.
For many years, computers have been saying yes, helping us improve pension services and outcomes for our communities. With the right data, tools, skills, and accountability we can help funds get the best result. With more understanding of the benefits of ML and the proper application of data science skills, trust in this technology can be built further, and perspectives can shift. So, maybe now is the time for pension funds to say “yes” too.