[Skip to content]

Sign up for our daily newsletter
The Actuary The magazine of the Institute & Faculty of Actuaries

Extreme insight 

How can behavioural models help predict policyholders’ reactions to unfamiliar situations? The Policyholder Behaviour in Extreme Conditions working party discuss


Ikon Images
Ikon Images

“Sometimes it may be helpful to consider a wider range of possible responses to a particular scenario”

We probably don’t spend a lot of time thinking about how our policyholders behave. But what if they suddenly started to behave unexpectedly? What might trigger such a change? How serious might the impact be? These are the sort of questions that the IFoA’s Policyholder Behaviour in Extreme Conditions working party has been considering. Our aim is to help companies (and regulators) recognise such situations early, understand what might be happening and devise strategies to mitigate potential adverse consequences. 

Turning to psychology

While studying past policyholder behaviour can provide useful insights, we realised that it was unlikely to greatly improve our ability to predict how policyholders might react to future extreme circumstances. There are a wide range of potential future extreme conditions and triggers. 

Often, traditional models of policyholder behaviour are based on neat mathematical formulae with simple drivers such as interest rates, which may not have been properly calibrated or validated but which seem intuitively reasonable. However, the underlying assumptions and relationships in such formulae may break down in extreme conditions. 

We have therefore considered behavioural approaches rooted in individual psychology, since psychological factors are:

  • Particularly relevant to understanding behaviour, allowing us to get beyond economically rational behaviour 
  • Well researched and evidence based
  • Likely to persist through extreme situations and to remain fairly stable during the timescale in which we might expect the unexpected behaviour to unfold. 


When combined with appropriate and timely management information, this can also help make sense of actual behaviours that are observed.

Behavioural models

We identified a number of behavioural models that we felt were relevant to understanding policyholder behaviour. We found behavioural economics and Abraham Maslow’s hierarchy of needs particularly helpful.

Behavioural economics applies psychological insights into ‘hard-wired’ aspects of human behaviour to explain economic decision-making. We grouped behaviours into three broad categories, in line with the approach adopted by the UK Financial Conduct Authority:

  • Preferences and perceptions that set the behavioural context within which decisions are made (eg loss aversion)
  • Biases that can distort the interpretation of information (eg hindsight bias).
  • Decision-making processes through which policyholders actually make decisions (eg rules of thumb).

Maslow described a hierarchy of levels of motivation, ranging from purely physiological through to self-actualisation. Maslow’s theory was that each level, in turn, must be broadly satisfied before an individual can move to the next level. Maslow’s levels can be conveniently grouped into three general levels:

  • Sustenance-driven (physiological, safety, and social needs): characterised by more conservative and traditional values
  • Outer-directed (esteem needs): embraces change, is generally optimistic in outlook about the future, and is quite success-driven
  • Inner-directed (self-actualisation): looks to ‘bigger picture’ environmental and societal issues with a strong desire for fairness, justice and equality.

Figure 1
Figure 1

Using psychological insights 

Using these behavioural models in a structured way can help us to think about how policyholders might respond to an unfamiliar situation. This can help us see beyond what we, as actuaries, might view as the rational response and to consider other plausible outcomes.

For example, we can ask ourselves how behavioural insights could affect policyholders’ attitudes. From this, we can build up a picture of how behavioural factors might interact to influence policyholder behaviour. We have provided a simple example in the boxout overleaf, based on a scenario of medical advances.  

Behavioural economics tends to assume that policyholders will all respond in a similar way. This is clearly an over-simplification, although it may be good enough for many purposes. Sometimes it may be helpful to consider a wider range of possible responses to a particular scenario. Might policyholders with different values, such as the three general levels from Maslow’s hierarchy, respond differently to the same stimulus? A few examples that illustrate possible diverse attitudes in our medical advances scenario are included in Figure 2

Once we have some understanding of these different viewpoints, we are in a position to assess how this might lead to different behaviours and to suggest different responses. What will work best for each company will depend on the profile of its policyholder base. This is something a company may seek to understand through various channels, such as analysis of data related to its interactions with its policyholders, or from information arising from the sales process. A company’s portfolio of policyholders may be spread among different behavioural profiles in a different way to the general population, so care is needed when making this assessment.

Figure 2
Figure 2


Understanding drivers

Considering potential policyholder responses through the lens of behavioural modelling techniques can help us get away from the premise that policyholders will always act rationally in any given situation, or behave in the same way as each other under the same circumstances. As demonstrated by the psychological insights considered above, rationality is often not the yardstick that is applied by policyholders themselves.

Simply relying on traditional actuarial models may lead to unreliable conclusions, sometimes resulting in worse management decisions than if no attempt had been made to model policyholder behaviour at all.

Having gained a deeper understanding of the drivers (or potential drivers) of policyholder behaviour, companies can share insights with different stakeholders across the business, as well as external stakeholders such as regulatory authorities, distributors and customer advocacy groups. This can greatly assist the development of appropriate responses to policyholder behaviour or, better still, pre-emptive action to try to influence the adverse impact of such behaviour.

Medical Advances Scenario 

In this hypothetical scenario, advances in diagnostic testing techniques result in increasingly affordable ways to detect those with a high risk of developing critical illnesses such as the more common cancers and heart disease. Those who screen negative are considered low risk.

For those that screen positive, a preventative regime is available (regular screening, lifestyle changes, preventative medication, etc) that promises significantly improved outcomes. 

Preferences and perceptions

  • Testing will be most important to those who perceive themselves to be at risk, perhaps because of family history
  • Being tested may gradually become the new normal
  • Those who test negative may no longer feel so committed to their existing protection policies. New policyholders may be tempted to take out policies before being tested and to lapse them if the results show they are not at risk
  • Insurers might choose to emphasise social, rather than commercial norms. They could, for example, offer incentives for ‘at risk’ policyholders to follow preventative regimes.


  • Extrapolating from medical science’s recent successes, projection bias may lead policyholders to believe that they will be immune to all types of serious ill health or early death.


  • ‘Narrow bracketing’ may lead policyholders to consider separately the costs of protection insurance and of their preventative regime. This may dilute the effectiveness of offering ‘discounts for compliance’.

These insights can be used to understand how policyholders’ attitudes to protection insurance might evolve.

Policyholder Behaviour in Extreme Conditions working party members

David Graham (chair) is a retired life and health actuary.

Jean Eu is a senior actuary at Correlation Risk Partners, specialising in life and health insurance.

Jeremy Kent is a principal and consulting actuary with Milliman, specialising in life insurance.

Eamonn Phelan is a principal and consulting actuary with Milliman, specialising in life insurance and enterprise risk management.