Roundtable: Innovation in pricing: the need to adapt

The discussion began by exploring what specifically is driving insurers to reassess and change their pricing models. Paul Ridge, sales manager for UK insurance at Sas, offered a couple of explanations.

“Some of it is being driven by customer need and the demand for different pricing models – on-demand insurance, for example. Some of it is just coming from attempts to win greater market share.”

Stuart Jeffery, head of pricing at Markerstudy, picked up on the second of these: “It’s certainly driven by a desire to grow. It’s essentially trying to do more with fewer resources within the pricing function itself.

“Whether you build a separate function to carry out your data science or upskill the people you’ve got within your teams –  it’s about developing who you’ve got.”

“I’m not currently sure if new analytics give us greater advantage. There are still a lot of underwriters in the market who have hands-on knowledge of the customer and each of the brokers in our case.”

Shamit Biswas, pricing director at Acromas Insurance, felt that another factor at play is competition between companies, each jostling for any advantage that might be had.

“It’s about keeping up with people,” he said. “When you see Direct Line launching Darwin, companies recruiting analysts or using cloud computing, you’re thinking: how do we get to that?”

Martyn Green, product and pricing director at RSA, questioned if the changes we are seeing today are fundamentally dissimilar to those the industry has experienced in recent decades.

“In the data analytics world, it’s accelerating but it’s not fundamentally any different to what it’s been in the last 20 years,” he said.

“We saw a [generalised linear model] come in and we’ve had a continual stream of advances and new rating factors. It’s just the same really, but with new techniques.

“The fundamental revolution which is happening at the same time is in customer perspective. The regulator said two years ago, ‘we’re not a price regulator’, and it has absolutely turned about-face on that.”

Whether modern data science represents a paradigm shift or just further evolution of past innovations, much of the discussion revolved around the promise of analytics.

Tempting perspective

Carys Lawton-Bryce, head of pricing at ERS, offered one tempering perspective: “It doesn’t matter how clever and sophisticated some of these data science methods become, essentially you still have to be looking at the customer journey, and at the underwriting view of the world.

“It’s not enough to just upskill actuaries. We also have to take the underwriters and the executives along with us, and be able to explain to them and have them understand what these models are telling them and what it means for the customer.”

Gary Jackson, senior data analyst at Gresham Underwriting, raised the issue of the quality of data that underlies new techniques: “Data is vital. A model is only as good as the data underneath it.

“I probably spend 50% of my time on data quality because there’s no option. Otherwise, the model will just churn out rubbish at the other end.”

Another issue was raised by Jeffery: “There’s a tendency across the market to go looking for data when you’ve already got it. You need to look at the data you already own on historical customers and use that to give those customers you’re aware of, in their circumstances, a better price.”

Whether all of it is being harnessed to its full potential or not, with a burgeoning amount of data informing pricing, where do consumers stand? Is the average policyholder clear on how their answers and other data insurers might collect translates to prices?

“It’s not any clearer than if they were going to buy a handbag,” said Amanda Fox, head of partnerships at Co-op. “You can buy one from Primark; you can buy one from Chanel. People have an expectation on brand, that it’s maybe going to be a higher price, but I don’t think there’s any more clarity.”

The discussion suggested that this is an area ripe for improvement. “We have to be very clear on the questions that we ask because if it’s not clear or grey, then we will pay that claim because it’s been answered,” said Jackson. “That can be much more important than the pricing sometimes.”

Lawton-Bryce agreed, bringing up the example of being asked for her occupation when buying her own insurance: “I always think that’s ridiculous. You go down the list and I’ve never had anything close to my actual job title on there.”

Ridge outlined one way in which data insights can be used to have a real, tangible effect on risk and pricing.

“We’re seeing some work that we do in the Internet of Things around fleets of trucks and being able to do predictive maintenance, where you’re able to say ‘we can detect when something is going to break’ and detecting it before it breaks to prevent other damage, reducing the cost of repair.”

Biswas offered another example. “It’s a great way not just of assessing the risk but actually contacting your customer,” he said.

“You can get in touch with them and say, ‘I noticed you were going through this roundabout. Did you know going this way might be a less risky route?’ You get the chance to interact.”

The need for sensitivity and caution when it comes to how people’s personal data might be used was never too far away from the conversation, however.

Fox raised concerns that doing too much could be counter-productive: “It needs to be very specific to the customer, because it’s valid to say we could tell you the safest route to go in your car, but I get really fed up with stuff on my phone that’s perpetually telling me, ‘you’re here’, ‘tell me about this’, ‘you could go here’, ‘why don’t you go here?’, ‘why don’t you get this?’

“We need to be tailored to what our customers’ needs are. If it’s not driving value for that customer, then we need to be really careful about just using it to perpetually bombard them with things.”

Another point, raised by Green, was that companies shouldn’t mistake trends such as the uptake of telematics for an indication that younger customers will always be comfortable about their data being collected.

“Even young drivers are keen to get away from having a telematics box in the car,” he said. “After one year even of driving with a telematics box, there’s a big shift towards a preference for a non-telematics product.”

“I don’t think there is acceptance from this generation that it’s OK to use data for anything.”

As the role of new analytics techniques expands, the industry is also presented with challenges around how they ensure their pricing teams can take advantage of the deluge of new data.

Different approaches

A number of subtly different approaches emerged from the discussion, ranging from training actuaries and underwriters in new data science techniques to hiring tech-savvy outsiders without industry experience.

Jackson gave one example: “We recruited someone recently and he’s got a degree in computer gaming, a first in that. We know we need those skills and we need to start bringing in these people.”

It is an approach that could prove fruitful. Norman Black, Europe, Middle East and Africa insurance industry principal at Sas, argued that shortfalls in actuarial experience may be able to be overcome with the help of technology.

“Technology can be an enabler,” he said. “Recently, we were set a challenge of using junior staff to challenge and beat the more mature and experienced in terms of model predictability and outputs, and it was successful.

“People that maybe didn’t have as much industry experience, weren’t qualified actuaries, could, with the enablement of technology, become much more effective in the quality of their output.”

Other participants highlighted the need for people working in the pricing function today to have a wide set of skills.

“What we’re seeing in recruitment, particularly at the junior levels, is candidates with hybrid skills: people who’ve worked as an actuary for a couple of years and then have decided to do further study in data science,” said Lawton-Bryce.

“We’re getting data scientists who are starting to do their actuarial qualifications. That’s where I see it going. People are just having to pick up a broader range of skills.”

The value of having all-rounders was re-emphasised by Anthony Stevens, president of Victor International: “Every time you add a person to a team, the risk of entropy increases because communication is always inefficient.

“The more you can have full-stack – either full-stack developers or full-stack actuary, data scientists and people who understand the consumer – the more you can put together agile teams to work on particular projects, rather than having to deal with five different departments, all of whom speak different languages.

“We don’t have any differentiation at all between – we just have people who do product, and within that we don’t really differentiate between consumer, programme analytics, underwriting.”

For all that data science may promise, participants were broadly in agreement on the continued importance of underwriters in shaping pricing.

“In actuarial exams, you have to use open-source programming languages such as R and Python now, so you’ll get a new generation of more multi-skilled, multifaceted actuaries,” said Biswas. “What you haven’t got are the subject matter experts, which is why you have underwriters.”

Hui Yu, senior cyber and tech PI underwriter at Allianz Global Corporate & Specialty, agreed: “From an underwriting perspective these data science tools are useful, but it’s still very important for underwriters to carry out their due diligence and ask questions.”

The role of underwriters

Green elaborated on what role underwriters now have. “Mainstream personal lines has been priced rather than underwritten for quite some time,” he said. “Model-driven pricing has been pretty prevalent in the industry since the 1990s, certainly.

“So the underwriter’s role is a bit different, it’s in the policy wordings and control framework piece.”

For Jeffery, the underwriter’s role remains front and centre: “They’ve got the raw skills to do some of these new techniques, but can they deal with the underwriters? Markerstudy is still heavily underwriter-driven. They own the products, not the pricing function, so they’ve got to be able to liaise with underwriters.”

Delegated authority also presents a challenge as longer distribution chains potentially strain the connection between how a capacity provider might approach pricing and meeting customer needs.

Fox said that in such situations, there can be a temptation to over-police. “You can tend to find everybody wants a governance framework for everything, when actually you need one that is not light touch but is appropriate,” she said.

“That way we can justify what we’re doing and we know we’re giving good customer outcomes, without having 100 pages of documentation behind it because it just doesn’t help anybody.”

Lawton-Bryce highlighted the role data is having in increasing scrutiny on delegated business: “What we’re finding now is underwriters requesting pricing analytics and understanding of what’s going on in that delegated book, so that we can have the same challenging conversations we’d have internally, and why shouldn’t we? It goes back to data.”

Reflecting on how Victor International negotiates the constraints of delegated authority arrangements, Stevens said: “Often we’re in a situation where we have a third-party channel that has a particular view of what the customer proposition should be and we have a carrier that has a particular view of what the underwriting approach should be.”

“We tend to start with the channel and the customer view, because in a sense that’s the constraint. Distribution is always the most important thing for a managing general agent, particularly in Europe you’ve got to start with distribution channels and the customer experience.

“Working together with the distribution channel to figure out what is the right pricing approach to optimise the customer experience comes first, and then you’re in a position to have the discussion with the capacity provider about why strategically this might make sense long term, even if in the short term it may not be something which they’re used to.” 

Sponsored