The Ethics of Insurance Pricing – 7 risks underwriters need to have on their radar

Several years ago, underwriters started to make wry comments about no longer knowing how their premiums were being calculated. This could now be an exposure, for what you’re responsible for, you’re also accountable for.

How an underwriter prices risk has ethical implications. And in this post, we’ll be looking at seven pricing practices that underwriters should have on their accountability radar. Some are more critical than others. The first one for example fits perfectly with the UK Prime Minister’s reference in her recent party conference speech to a ‘broken market’: where loyal customers are charged higher prices.

These seven pricing practices lie at the heart of the ethics of insurance. There may be others, but these are the ones every underwriter needs to be watching.

Dual Pricing and Optimisation

Dual pricing is unpopular with both the public and Prime Ministers. It involves new customers receiving introduction discounts and loyal customers paying more as a result. It is however being superseded by more sophisticated optimisation techniques that move pricing from dual to multiple dimensional. These set the premium at what the consumer is thought to be willing to pay. This means not giving away discounts where you don’t need to, and charging more if the customer is prepared to pay it. It’s controversial and has been banned in over 20 US states. The FCA seems to be in two minds about it.

If price optimisation is a feature of an underwriter’s pricing strategy, she needs to know the answers to two questions:

How does my use of price optimisation support the fair treatment of customers expected by the regulator? This is an explanation that needs to convince a regulator.What data is being optimised? If your optimisation picks up on complaints or number of telephone calls made, then this is an ethical risk that you will need to address.
Profit and Value

Value is not the same as price, but there are clear overlaps. It may be the right product for a particular group of customers, but does it then have the value that those customers would expect? Regulators are increasingly judging value through performance ratios for that book of business: too low a ratio and questions are raised about unethical practices. When identity theft insurance was found to have a loss ratio of only a few percent points, it’s days as a viable product in that form were numbered.

There are two important dimensions to value that an underwriter needs to understand:

visibility: can the customer identify where most value in a product lies?bundling: where does that value sit in relation to the policy’s core cover, extensions and add-ons?. The insurance sector is sometimes criticised for products full of extensive but low value frills, while the core cover that customers are most in need of is trimmed to contain costs.
Fairness and Access

The boundaries and inflexion points of the underwriter’s pricing strategy have ethical implications as well. Fairness can be a rather complicated thing, but its starting point is pretty straightforward: everyone is of equal worth and should have an equal opportunity to access cover for their insurable needs. At the same time, insurers often have target markets and so focus their pricing for commercial reasons. That’s fine, to the extent that an underwriter can demonstrate that her choice of target market and her pricing focus are equitable. There are regulations, and indeed laws, that set out clear expectations on this. The regulator’s interest in vulnerable consumers is one current example. Underwriters need to be clear on two things:

the business case for those pricing boundaries and inflexion points,how they are addressing the needs of vulnerable consumers.
Personalisation

There are two sides to the personalisation of insurance pricing. Advocates talk about premiums that reflect policyholders’ claims. Critics talk about a loss of risk sharing and the price volatility that can then emerge. There is also a more general question hanging over personalisation. It relies on automated decision making and as a result, the underwriter needs to be actively managing the outcomes that such automation generates. Here are a couple of examples:

what levels of pricing volatility is your personalisation strategy producing?are there any concentrations of such volatility, and if so, what implications might that have in terms of fairness of access?
Profiling

Insurers differentiate between risks in difference ways and to differing extents. That’s how a competitive market works. There is however a danger that this could drift into discrimination between risks, which is an entirely different thing. And that danger is a real one when that risk differentiation is driven in large part by automated pricing decisions. Automated decisions have been shown to produce discriminatory outcomes and so an underwriter needs to have procedures in place to guard against this happening in her firm. Here are a couple of steps that an underwriter should be taking:

adopt a discrimination prevention strategy;use discrimination detection algorithms to identify any exposures.
Data

All of that automated decision making only works if it is fed with enough data to be effectively trained on. This can raise questions about provenance: do you know enough about the data’s source and quality to be confident that it will inform your pricing decisions with the necessary accuracy? And then there’s relevance. Just because an algorithm can identify a correlation between two variables doesn’t mean that it has pricing implications. There’s a very high correlation between cheese consumption and death by becoming entangled in a bed sheet. A life underwriter should have procedures for weeding out such spurious correlations in her automated pricing decisions.

A third and hugely significant question relates to the inherent bias that can often be found in historical data. The danger is that such bias can result in discriminatory decisions and outcomes. Just because a system is complex doesn’t mean that such bias cannot be actively addressed.

What standards are in place for checking, recording and utilising information about the provenance of the data being used in pricing decisions?Are there procedures in place to oversee the relevance of data sets to insurable risk?What controls are being used to manage the levels of significance being applied by pricing algorithms?Are there systems are in place to test incoming data sets against discriminatory bias?
Transparency

The premium might be expressed as a nice neat number, but there are still ethical questions associated with the transparency of what that number represents. Insurance now comes with financing options, opt-in covers and a welter of renewal and mid-term fees that often add up to a far from clear picture. And with a strong trend towards the convenience of taking out cover, the question of what exactly you’re paying for rises too. This gives us questions such as these…

Is it clear to the customer what they’ve got to pay?is it clear to the consumer what they’re paying for?How are you judging each of the above? Realistically, it needs an independent eye.

Now, you may  be thinking that some of these ethical dimensions to insurance pricing aren’t that relevant to your firm. And that may well be the case, but the accountability provisions in insurance regulations now mean that such an assessment needs to be more than just a hunch. The underwriter needs to be able to show that she’s assessed these different ethical dimensions through a considered process, and a process that can easily to be repeated over time, and replicated across lines of business. Each of the seven raised here are serious enough to warrant inspection rather than just a passing consideration.

Where do you start then? Well, getting this right starts with a proper understanding of the issues involved, and that means training. If you’re going to be held accountable under regulations, it’s best to know something about the issues that this will encompass. This doesn’t mean more technical underwriting training: it means training, and assessments, relating to the ethical risks your function could be generating.

It may be tempting to leave all this to the compliance function. They can certainly help, but in the end, it is an underwriter’s responsibility to understand, manage and oversee their pricing decisions and the consequences that they can have.

One last point worth considering is the standard against which the underwriter will be making these judgements. The danger is that cost alone becomes that standard. That became clearly recently when the CEO of a leading insurer talked about being unable to address one of the above seven issues because it might cost too much. This was despite that firm talking in its code of ethics about never compromising on their integrity, under any circumstances. Yet integrity means doing the right thing even when it’s not in your interests. If a firm has ethical values, it needs to recognise when they apply and how to apply them, or just drop them altogether.

I’m not saying that cost should not be used, but it needs to be given equal weigh with other factors, such as customers needs, ethical values, fairness, trust and reputation.

Insurance pricing is undergoing a revolution, just at the time that underwriters are being held to account like never before. It’s important then for underwriters to know about the ethical issues that that accountability is built upon.