Big Data and technology can play an important role in helping pension fund members reach optimal retirement outcomes. Pension providers and schemes sit on a huge amounts member data that covers fund values, contribution rates, chosen pension options, salary, gender, age and much more.
There is widespread agreement that providers and schemes should drill down further into their large sets of customer data and analyse them computationally to reveal patterns, trends, and associations. Indeed, open finance platform Moneyhub is in discussions with several pension providers about bringing its data aggregation, analytics and tools to savers.
There is a shallow and a deep end of pensions data because different types of pension arrangements hold varying amounts and types, according to Chris Connelly, who represents the Pensions Administration Standards Association on the UK’s Pension Dashboard Programme.
He says: “Pension schemes can learn a lot from a postcode, but they might have to try and dig deeper into someone else’s data to learn a bit more about their members.
“Investment companies that sell pensions products have access to more data, such as transactional history – as they might have sold them other products. They can ook at what they might be doing in their insurance products and see how that might relate to pension, investment or savings products.” Under the revised Payment Services Directive, which facilitated open banking, this information is now available to organisations if the customer grants permission.
Big Data can be used to help encourage members to make different choices – or take it out of their hands and do it for them. For example, it could be used to create default defined contribution (DC) investment strategies that are more tailored to an individual and their likely risk profile.
Intelligent defaults
Richard Butcher, professional pension trustee and managing director at PTL, says: “At the moment, we design everything around the model member – and really can’t understand why we restrict ourselves to that, because w e ha v e the c omput e r sophistication to be much more granular than that. Very few people will be a model member, so we need to design a system that a c c om mo d a t es peo pl e ’ s individuality as best as we can.”
A default strategy will be optimal for an average member but suboptimal to at least some extent for those not fitting this profile. For some, it will be 5 per cent suboptimal but for others at the extremes of the bell curve, it could easily be more than 50 per cent suboptimal, according to Butcher. For example, in a typical default fund, a 52-year-old high earner with a high fund value, who owns a house in Guilford, and commutes into London, will have the same investment strategy as a 23-year- old living in Hull who works in a bakery, he says.
“That’s not right because those two individuals have different investment needs. If we could plug some of that data into an algorithm, we’d end up with different investment strategies for the two of them, and that would be more appropriate for those individuals,” says Butcher. “The output of all of this is rather than one bell curve where you have a suboptimal investment strategy for a lot of people, you would end up with dozens of bell curves so the degree of suboptimality would be minor,” he adds.
Members would be able to opt out and design their own strategy. In the UK, asset managers are talking about data with increasing intensity, especially where they have learnings from countries like the US where intelligent defaults are more prevalent, says Tim Phillips, vice-president of the Pensions Management Institute.
He says: “It’s not an unusual approach [outside the UK] to be running tens of thousands of projections with variations on things like member contribution rates, member salary growth, timing and size of post-retirement withdrawals, and also market environments and black swan type events, and then crunching these different simulations to learn how different asset allocations perform. One of the difficulties of getting it into the UK is the prevalence of the trust structure and role of trustees in that.”
There is some wariness about the use of AI and algorithms in pension defaults, however. Aon head of UK retirement policy Matthew Arends says while technology has a huge role to play in helping individuals, he remains to be persuaded on Big Data and intelligent defaults in which a machine and algorithm decide for an individual where their DC savings ought to be invested. “How you could possibly be sure that the algorithm has complete data on an individual to make a decision for them?” he asks.
The other option is for providers to use information on a subset of a group to give proxy information for the whole. “It might give providers the leverage to come up with cleverer defaults for groups, rather than individuals,” says Connelly.
Arends can see a role for Big Data to provide suggestions for products or services that might be more relevant to an individual based on their circumstances, with the individual ultimately making the decision.
Protecting members
An important question is whether pension scheme members would be adequately protected from digital marketing manipulation.
There are already strong protections in place under the General Data Protection Regulation.
Moneyhub CEO Sam Seaton says: “We just have to be very careful about using the data only for the purposes for which the customer has given you permission to use that data, and be very transparent about what you’re doing. For example, a pension provider wants to use its pension members’ data in aggregate, to understand what other products and services they should be looking at to help their customers.”
Phillips believes it should not be a problem because data would be anonymised: “If you’re looking at it from a Big Data approach to creating intelligent defaults, then you shouldn’t really be needing to have an individual member’s personal information that isn’t anonymised.”
This is different to using individual members’ personal data to, for example, push nudge messages to encourage them to do certain things. As Aon’s Arends says: “One can see that that might be a path that could get abused if there are too many pop-ups, with adverts for invest in this, buy this product, for example. Control could be an issue here, as well as the ability for the consumer to opt out.”
Having a better digital understanding of scheme members will vastly improve workplace pensions for the benefit of savers, by creating intelligent or optimal investment defaults.
Seaton says a shift in attitude is needed in the pensions sector, and that use of Big Data and AI will be a key differentiator among providers going forward: “Part of this is about shifting from a product focus to a financial wellness focus, which is about seeing the customer for who they are, where they are at and what they need, rather than ‘what products have I got in my bag that I might be able to put in their back pocket’.
“There’s so much fear among trustees in general about data and I suspect that’s because they haven’t grown up with data.”
Seaton believes it is more the mindset of people in the industry that is now holding things back as “the technology and data are available” – but she is confident this mindset is shifting.