Questions have already been raised about the ethics of dual pricing. This started out as the practice of offering new customers lower quotes than a renewing customer would get for the same risk. As insurers gain access to more and more data about us, it has developed into a more sophisticated form called ‘price optimisation’. It also raises a number of ethical issues.
I’ve written before about insurance being provided though a market like any other, with insurers offering products that will earn them a return on their investment and consumers free to choose from a wide range of insurance providers. On the face of it, recent market developments like price comparison sites increase the choice available to consumers and spur on insurers to price their offering accordingly. There is however more to this than first meets the eye.
Enter price optimisation. This involves an insurer applying predictive analytics to its bank of big data, to identify those consumers whose online behaviours indicate that they will be less likely to shop around at renewal. Such consumers find their renewal premium edging up year after year until insurers have them paying a price just below what might drive them to start looking elsewhere.
Now you might think that this is just life. Take the effort to shop around and you could be rewarded. Don’t bother and you pay the price. And this is true to some extent, but not always to an extent that legislators are happy with, for insurance is different. And some of the ways in which it is different raises ethical questions about insurers’ use of price optimisation.
The problem is that there are significant information asymmetries in the insurance market between provider and consumer. Policyholders don’t know how their premium is calculated and are often unclear about what they’re buying. It’s not easy for them to work out what they’re covered for when some policies contained more words than famous novels. So there’s a danger that some insurance firms may seek to exploit these market inefficiencies by price optimising those types of renewals.
There’s a danger that predictive analytics could stray the insurer into an ethical minefield. If those predictive models are not carefully monitored, there’s a danger that they will price optimise according to where policyholders live, what they buy and what they read. These could also be indicators of ethnicity and financial literacy. It can sometimes be a short step from differential pricing to discriminatory pricing, but the consequences of that step for the market would be explosive.
Big data analytics is often thought of as the objective sifting of the information we disclose about ourselves. It’s far from the case. Big data is permeated with lots of implicit and explicit views of the world, used to fuel judgements about what to look for, what to include, how to interpret it, how to weigh up its significance and what to do with the outputs.
Insurers need to take care that, in exploiting the many benefits that big data can bring, they don’t forget the need to think ethically about what they are doing and the consequences that might flow from it. Price optimisation would be a good place to start.
In this later post, I go on to look at three sector-wide risks that price optimisation could expose the insurance market to.