“With insurance no longer a matter of risk pooling, policyholders pay a fairer premium”. So said a director of a well known UK insurer recently. It raises two interesting questions. Does insurance have to be about risk pooling? And what makes a premium fair?
A line of thinking popular amongst insurers at the moment is that the more a policyholder’s premium can be fine tuned into the precise risk presented by their particular policy, the fairer that is for that policyholder. This is because they would be picking up less of a share of the costs of claims made by their accident prone neighbours. The policyholder’s premium would be for their risk, untrammelled by the misdeeds of others.
In presenting such personalisation of risk as fair, is it really the policyholder that insurers are thinking of? It seems unlikely, unless they are more particularly thinking about those policyholders who do not claim. The logic of this line of thinking surely means that ever greater personalisation would lead to ever greater fairness. That argument only holds water if the policyholder was not submitting any claims, for if they did, then their premiums would shoot up in response. And how fair would that then seem?
Personalisation of risk pricing seems to be more about fairness for the underwriting account. Such sentiments seem understandable, but only to a degree. On the one hand, insurance is a business and it is fair that those who invest in it make a reasonable return over time. On the other hand, how that reasonable return is generated is also a matter of fairness. After all, just because a pricing strategy makes a fair return for its underwriting account does not in itself make that pricing strategy a fair one.
Compared with, say, the personal lines market of twenty years ago, it does seem harder now for underwriters to return an even half decent combined ratio. And yes, the market nowadays does seem to be more competitive and more dynamic than the early 1990s.
Hold on a minute though! Weren’t the early 1990s not just a time of more reliable profit, but also a time when rating was far less personalised? Back then, many insurers still used a single national rate for all household buildings cover and motor insurers relied upon only a dozen car groups? Underwriting in the early 1990s relied much more upon risk pooling than it seems to now, yet it didn’t seem to do the combined ratios much harm. Is there a lesson in that for the underwriters of today? Might too great a departure from risk pooling introduce inherent instabilities into an underwriting account?
Personalisation of risk pricing relies upon an insurer having the capacity to handle a vast amount of data. All too often that capacity is measured primarily in terms of technology, when in fact the capacity of most importance is that of the experience and skills to organise and interpret what all those pricing variables tell you.
Some insurers now admit to having so many pricing variables that they are no longer able to fully understand how the rating factors for some risks add up. This sounds like they’re slowing being squashed by ‘big data’. If insurers don’t keep a firm handle on how their risks are being priced, aren’t they in danger of encountering some unintended consequences?
Another question mark over the fairness of personalised risk pricing comes from how broadly it is applied. Much of the personalisation going on in household accounts has been around the risk of flood, with others such as storm getting far less attention. Yet while personalisation can be applied selectively to just this or that part of a policy’s pricing, the same cannot be said about fairness. An insurer cannot pick and choose which parts of a policy’s premium to call fair when the policyholder only sees the end lump sum.
Let’s return to that earlier point about unintended consequences and consider one that could arise for insurers from a combination of those difficulties of handling too many rating variables and a selective approach to personalisation of risk pricing. Together, these two factors risk introducing an inadvertent drift from the differentiation of risk offered by personalised risk pricing, into the nightmare of discriminatory risk pricing created by lack of care and attention. For policyholders, it means their cover and premiums being determined more by socio-demographic factors and less by the physical and moral hazard being insured.
This might perhaps seem too remote a prospect for insurers to worry about, but then, perhaps US insurers thought so too back in the 1960s before they became embroiled in just such a controversy. One sign that it is not that remote a prospect can be found in the socio-demographic segmentation on offer to insurers in the UK, some of which carries labels with distinctly political overtones.
We started out by asking what makes a premium a fair one. It’s clear that questions can be raised about the fairness of personalised risk pricing. Does this mean that personalisation per se could be unfair? The answer, as is often the case, lies somewhere in the middle. One way of exploring that ‘middle ground’ is through how the experts explain what is meant by fairness.
Here is a synopsis. Fairness needs to take into account all members of society. That however does not mean that every person should be treated in exactly the same way. Some proportionality is needed in how everyone is treated and that proportionality is determined by a mix of need and merit. At the same time, every person is of equal worth and should have an equal opportunity to access what is on offer. So fairness involves ensuring equal access to what is on offer, while at the same time distributing it proportionally according to need and according to merit.
What this tells us is that some personalisation of risk pricing is fair, but if that personalisation is taken too far, it becomes unfair because the ‘equal access’ and ‘need’ dimensions of fairness are undermined. So this means it is wrong to apply one dimension of fairness (in this case, the merit one) as justification for an action (in this case, personalisation) without giving equal consideration to other dimensions of fairness (such as equal access and need).
This brings us to the earlier question about whether insurance has to be about risk pooling. Just as risk pooling can be seen as a financial mechanism for spreading risk, it can also be seen as a social mechanism for sharing risk. This sharing mechanism helps bring the benefits of insurance nearer to those who need it. It also helps make it more accessible to all members of society. So some risk pooling is also fair, but not if it is taken too far, for otherwise that ‘merit’ dimension of fairness suffers.
Like many things in business, fairness in insurance is a balancing act that insurers need to pay attention to and keep working at. Keep your eyes too long on only one dimension of fairness (for example, ‘merit’ in the form of personalisation) and that balance is undermined. Let that imbalance become ‘this is the way insurance is done now’ and all sorts of unintended and unpleasant consequences will emerge for the sector.
If insurers view personalised premiums as a portent for the demise of risk pooling, then they risk losing the argument that insurance is something that is fair and of value to society. Insurance will progressively convert to a rather complicated form of savings and loan, relying for its success more on loan rates and less on risk transfer. We’d all become bankers then, fronted by PR specialists having to explain why the large numbers of now uninsurable families are all part of that much heralded fairness. On the radio, Messrs Humphreys and Naughtie will be rolling up their sleeves in readiness of some interesting interviews with insurance chief executives.
Note: this post has been updated since originally published, to include additional material from a subsequent article on personalisation and risk pooling that I wrote for the CII’s Journal.