Big data and pensions

12 October 2017

Big data and pensions

The Royal Dutch Actuarial Association hosts a lot of meetings on contemporary topics for its members. On 3 October 2017 they held a networking session for their Pensions Group with “Big Data” as its theme. With more than 100 members turning up for the event, it was clearly an interesting topic. From the perspective of the academic world, Marike Knoef gave a lecture on the concept of “Big Data” and about potential applications in the pensions sector. Ronald Wiekenkamp, a Partner at Cmotions, spoke further about the practical application of Big Data in this domain.

In addition to being an Associate Professor at the University of Leiden, Marike is also an affiliated Member of the Board at Netspar. This organisation aims to improve the possibilities for financing “old age” in the Netherlands. In her introduction, she showed how data-driven systems are rapidly taking over many cognitive tasks. Even when it comes to choosing a life partner, many people seem to favour an algorithm over their own intuition. Big Data is contributing towards this, and the skill is to identify causal relationships. After a certain amount of debate on the concept of “Big Data”, Marike mentioned a number of ways in which Big Data can be applied in the pensions sector:

• Personalisation, especially in pensions communication
• Probability calculation, particularly for mortality
• Investment, e.g. assessing the mood of investors
• Monitoring and enforcement
• Reforms within the sector

The tension between further differentiation and the solidarity principle resulted in an interesting debate, leading to the conclusion that digitisation means more personalisation cannot be stopped. The following verdict ensued: “Digitisation is like asbestos: it mysteriously spreads and then you can’t get rid of it….”

Ronald Wiekenkamp, drawing on the practical experience of Cmotions, made some further observations about the notion of “Big Data”. To make it more applicable, there are other important characteristics beyond the famous 4 or 5 V’s, namely: relevance, linkability and information overlap. He illustrated this with a few practical applications in the pensions sector. The first of these is the segmentation of pension plan participants by persona. By linking pension fund participants’ data to profile data, you can distinguish segments with the help of statistical techniques. Participants that belong to a segment have very similar characteristics, whereas the segments themselves are as different from one another as possible. Linking people to the segments in which they essentially belong results in persona segmentation. This is highly useful for communication with the various groups.

You can take this a step further by asking participants to answer a few questions about their pension awareness and their pension accrual. Their answers can then be used to segment them by these two dimensions. Cmotions distilled these “golden questions” from a long list of questions. Together with a pension insurer, they are now developing an application that allows people to see in one minute how they are faring in terms of their pension risk and pension awareness. This application is being launched at the three-day pension event, the Pensioen3daagse 2017.

Ronald finished with an application for actuaries. When it is known that an individual’s life expectancy is correlated with their social status, Cmotions enriches the files of pension fund participants with this type of profile data. This is being done for, amongst others, Milliman, which is now much better equipped to calculate mortality. When pension funds and pension providers have more data about their participants, they are better equipped to adjust their fund policy accordingly.