23 April 2019
Our Terms of Reference
We were tasked with considering how policyholder responses to an extreme situation might have a significant impact on insurers. We have considered a broad range of potential extreme scenarios, driven both by economic factors and by other factors such as social change or medical advances. Further information is available on the IFoA website.
Past Experience is of Limited Help
There is only a limited amount that can be learned from studying past events. We found little evidence that insolvencies have been triggered by policyholder behaviour, though in some cases this may have exacerbated an existing situation.
Even if a future situation seems superficially similar to a historical one, key aspects will almost certainly unfold quite differently. Both products and regulatory oversight have adapted and changed. Policyholders, insurance management and regulators have all learnt from experience and become more sophisticated. Compliance frameworks impose a requirement to be able to defend advice given and actions taken against a background of a compensation culture. Online transactions make it increasingly easy for policyholders to react. The Internet and social networks mean that they will be better informed – but also more prone to “herd” behaviour.
What, then, can we suggest that might help those involved manage the risks associated with policyholder behaviour in a future extreme situation?
Understanding policyholder psychology
We believe that results from personal psychology can help us understand how policyholders are likely to behave in a wide variety of circumstances. These are supported by considerable research and can help us to understand the psychological factors that underlie personal behaviour.
Personal values, attitudes and beliefs change only slowly over a person’s lifetime – far more slowly than their behaviour patterns. Faced with an extreme situation, people will respond according to their pre-established psychological profile, which will not suddenly change. Therefore, understanding these factors can help us assess how policyholders might react in many types of extreme circumstances.
Three approaches seem particularly useful: buying behaviour models, behavioural economics and the psychology of human motivation. These can complement each other when trying to understand how behaviour patterns may evolve. The first two may already be in use within marketing departments, while HR departments may be familiar with motivation theory, particularly Maslow’s Hierarchy of Needs.
Using behavioural economics, we have been working through a variety of extreme scenarios to see whether this can help to predict how policyholders might respond. By overlaying different value-systems from Maslow’s Hierarchy, we have also tried to assess how different the behaviours of groups with distinct world-views and priorities might be.
We do not claim that our approach can predict with confidence how policyholders will behave. However, it does seem to help to think of likely outcomes other than a continuation of current trends or what might seem to us, as actuaries, as the obvious rational response.
Modelling Behaviour
Given an understanding of the behavioural drivers, modelling can help deepen our understanding of how these reinforce or balance one another, and the financial (and other) consequences of the behaviour. A model can be an effective tool to share insights with a wider audience, such as senior management. A model can also provide a “sandbox” to develop and test possible interventions, since it may no longer be safe simply to rely on our experience of what worked in the past.
However, traditional actuarial modelling approaches (such as stochastic models) may not particularly helpful in this type of situation. Since our extreme scenario represents an event in the statistical “tail”, there is unlikely to be enough relevant past experience on which to build a model. Also, stochastic models tend to be descriptive, rather than explanatory, treating relationships between variables as correlations. In an extreme situation, how can we be sure that these will remain valid?
We are therefore exploring two alternative modelling techniques: Agent Based Modelling; and Systems Dynamics. These modelling techniques may be less familiar to most actuaries, but they seem to have have the power to model behaviour in extreme situations, using insights derived from policyholder psychology.
Collecting and Using Management Information
Access to suitable data in relation to policyholder behaviour can help insurers to identify when normal variations in policyholder behaviour is giving way to a new, hitherto unobserved trend.
MI can help us understand drivers of current policyholder decision-making behaviour. Through this, it can improve confidence in policyholder behaviour modelling, and help to set sensible model parameter values.
We are looking at what sources of information insurers already have that could be useful, and how they might reasonably be able to collect additional information.
David Graham
Chair, Policyholder Behaviour in Extreme Conditions Working Party