The dangers of ‘Static vs Continuously updating risk factors’ using car theft examples, Mike Daly

A black Ford built between five and eight years ago and parked in Birmingham looks to top the list according to data from Tesco Insurance in this article published by Which Magazine which also offers tips on how to reduce the possibility of theft.

"Car theft is on the rise in the UK, with figures from AA showing a 25% increase in 2022. Some vehicles are more of a target than others, it appears.

Tesco Bank crunched its own car insurance data and found that 17% of claims for stolen motors were for Ford models. Land Rovers were second, accounting for 11% of claims, followed by Mercedes (10%), BMWs (8%) and Vauxhalls (8%). "

Which Magazine

Whether you are an MGA, broker or carrier it is vital to be able to access and analyse the data from internal underwriting and claims sources plus external sources. MGAs seeking approaching renewal of their books of business need accurate analysis. Too often when reports are prepared the figures don't add up and in a hard market that just won't do. Capacity providers will spot when the sliced and diced analytics do not tie in with the top-line summaries. They will trust those who can combine underwriting and claims data with all those complicated date variations. You know: -

Effective date
Written date
Earning date
Reporting date
Claim and event date
Original inception date
Claim movement date
Sales date.

And it's no use just looking in the rear-view mirror. Ryan McMahon SVP at Cambridge Mobile Telematics gives a great sporting analogy of the dangers of ‘Static vs continuously updating risk factors’.

'I recently posted about CMT’s upcoming presentation at the Casualty Actuarial Society annual meeting in Los Angeles.

Nicholas Hamwey and Lakshmi Shalini are going to be discussing the role of risk factors that are reflexive and represent the condition of the risk in real time.

Using static factors is like looking at Stanford and thinking that they lost at almost any point in their game two weeks ago vs Colorado. The game ended 46-43 but at halftime, Colorado was up 29-0

According to ESPN Stanford’s probability of winning at the half was .02% which is roughly double the chance of being struck by lightning in the US.

Now imagine using risk factors and models that only represent the world as it was (x) months or in some cases years ago vs simply “looking at the score” in real time.

It’s hard to imagine a world where organizations that rely on static risk factors have the ability to accurately measure and predict outcomes.'

From binder inception to the current date today and the fully earned data and binder expiry; it takes deep expertise, hard work, extensive data processing , cleaning and match plus AI powered capabilities to make sense of the data, make the right decisions and and ensure the fully earned and ultimate loss ratios are industry leading.

The expertise of the actuary, underwriter and claims analyst are today augmented by the AI, and new GenAI/LLM powered technology platforms like Giroux.ai for MGAs and Brokers, Cytora and hyperexponential, Aiimi for brokers and carriers. Palantir for Large carriers, brokers and reinsurers.

Plus the service providers like CGI, publicis sapient, Hitachi Vantara, WNS, Genpact,Sutherland, Imformatica, well profiled by HFS in a recently released report HFS Horizons: Generative Enterprise™ Services, 2023

Whilst many insurers like Sabre have managed to implement viable pricing and risk strategies and maintain good combined ratios there have been many well-publicised cases of the opposite. The existing legacy technologies have proved wanting when circumstances change and predictability becomes less certain and reliable. Not just for motor claims of coUrse but home and property, business interruption, cyber and many other lines of business. Sticking with motor for now.

Annual premiums are now soaring as underwriters try to catch up with cost inflation, putting pressure on consumers. EY warned that it expected prices to rise 16 per cent this year and 11 per cent next as the industry strived to “rebalance its books”. Based on discussions with insurers on their pricing strategies and other workings, the consultancy thinks that will be enough for UK motor insurers to generate an underwriting profit for 2024, achieving a net combined ratio of 97.4 per cent. But given the significant rises in prices and costs, it was quite a “narrow runway” for insurers to achieve profitability and there was a significant degree of uncertainty on either side of that forecast, Bonnard said."

I was reading a prescient article by industry leader Rory Yates on just such a topic.

Should insurers prepare for a transportation revolution?

Yates comments:

"The world has become more connected and data-intensive but only marginally smarter.

Cars are a good example. They've been connected for some time and are only now becoming more autonomous and contextually aware, capable of journey and self-management.

In this latest phase, manufacturers treat cars as ecosystems that manage everything from improving journeys to maximizing vehicle uptime and minimizing disruption. As a result, insurers are now faced with many opportunities and choices. With this comes threats and challenges in equal measure.

The question many ask is, as car manufacturers exponentially increase their understanding of their products and the people driving them, will they become the insurer? A better question is, will the insurance industry adapt quickly enough to avoid this outcome?

I hope so. There are huge upsides for insurers and their customers. However, a fundamental shift in the way insurers align their businesses and their technology is required."

Matteo Carbone often laments over the caution with which insurers leverage IoT and Telematics even today so that means it will be even more of a challenge to monetize all the data emanating from vehicles in the future. The demise of Wejo, one of the data brokers expected to be the inevitable source of telematics data for the insurance sector, is a salutary reminder that it takes time, cash, effort, and sound strategy and not just technology to succeed. See "What went wrong at one-time unicorn Wejo?".

Yet Matteo points out the strength of the incumbent's position against both full-stack insurtechs like Lemonade and Root in This evaluation makes no sense. Insurtech startups are doing it again! Again, technology and digital transformation are not enough.

Matteo has also questioned the implied competitive advantage of the auto OEM as they increase the proportion of electric vehicles and

The auto OEM with the greatest ambition and vertical integration from raw materials to insurance is Tesla which, you would have thought, could offer offered the best value to consumers as it designs effective and cheaper repairs into its products, has its own service centres (in USA at least) and has more data than any on electric vehicles (EVs) for underwriting.

To balance that most households have more than a Tesla on the drive, or in the garage, and this mix of EVs, hybrids, and internal combustion engine (ICE) vehicles is where incumbent carriers should excel. So if a carrier can compete strongly against Tesla it should do even better against other auto OEMs tempted to offer insurance.

Yet as Yates observes insightfully: -

" There are already solutions in the claims automation space that handle 'upstream' data like issue detection and address the really difficult stuff like slow collisions. This includes how they are modeled and validated. There are also those that deal with 'downstream data' into repair networks during the claims process itself. All of which support a better ENOL service and claims experience overall.

The problem is whilst the vehicle data has improved dramatically, it is neither simple nor intuitive to analyze. Even if this includes 3D and 4D Light Detection and Ranging (LiDAR), accessing and interpreting the raw data from a vehicle's Event Data Recorder (EDR).

Therefore, insurers often rely on a crash reconstruction specialist to make sense of the data, ranging from $2,000 to $5,000 per claim (making this largely unfeasible). The situation is made worse because there are many different types of EDRs, which can vary greatly in the data they gather from one system to another. Moreover, the nature of a crash can mean that certain data is not recorded, making EDRs an imperfect system even at the best of times.

Another source of intelligence that insurers use when investigating a claim is the First Notice of Loss submissions from the insured parties involved in the crash. These will typically include a brief description of the crash, the resulting damage, police reports and any photo or video evidence. This extends to services now categorized as "estimatics", which embodies tools, data and services that can better estimate the damage and repair requirements in a more efficient and largely automated way.

However, biases and imperfect memory can skew the accuracy of these reports. Even crash reconstruction specialists can differ in opinion when interpreting more subjective information such as road conditions, weather and the activity of third-party vehicles moments before a collision.

So these solutions have a very clearly defined problem space. However, this does mean the data source has to be proven, especially if there is any legal dispute and arbitration involved. This is becoming a major factor in insurance fraud, where things like generative AI will make it easier for people to fake photographic or video evidence.

Another important area to assess is the solutions that sit on tech outside of the car (i.e. via the mobile or another connected device). Insurers must determine their likely uptake with customers, so they can ensure they get the level of data access and ongoing tracking needed to make the service work. Understanding adoption rates of device-level services is vital and a critical business case component for carriers.

Then there are the downsides of tech-driven service provision. The issue that has plagued this area has historically been consumer uptake. The telematics black box, even for those that weren't suspicious of being tracked and "penalized" rather than benefitted, was scarce.

Many don't keep a car for long, and the installation invalidated the manufacturer warranty in some cases, and so on. The truth is "apps" aren't reliable either, what happens when the iOS upgrade invalidates some of the data security protocols? Does the insurer take the risk? How much can they make it a proper usage-based offering?

If the dawn of zero-touch, multi-choice, and seamless car insurance is here, it begs the question, how do you prepare when ignoring it isn't an option anymore? This is a key priority before you decide which "tools" to use.

Anticipate the products vehicle owners, renters, and subscribers need over the next decade and experiment earlier rather than later to ensure you have world-beating products. That is not as easy as it sounds as few incumbent carriers have gone as far as Baloise Group in considering mobility as a marketplace rather than just vehicles and via direct investment and partnerships have established a mobility ecosystem with; you've got it, embedded insurance.

Patrick Wirth, Vice President of Mobility, has harnessed the group's clear commitment to personalising mobility services by investing in and nurturing mobility companies and providing a continuum of services that caters to the evolving needs and resources of customers.

Source: Baloise Group

How many carriers have that continuum of products and services already deployed?

I believe there has been too much of a trend to move the business from a strategy discussion to a 'tools' discussion. Instead if having a clear answer as to 'WHY' technology should be deployed there is a tendency to ask "Which platform or software" should we choose? If I have encouraged that I apologise, but in my defense I always try and ask what strategic outcomes are required before suggesting the best technologies.

So let's say you have a great strategy that puts you way ahead of Baloise Group and those pesky MGAs that are focused just on single products. What then?

Ambition is nothing unless technology providers allow insurers to experiment, fail fast, and apply the lessons they learn to deliver the outcomes to anticipate this transportation revolution. Yet too many such providers are complacent and rely on insurers to pay hundreds of millions of dollars on platforms that might be 'cloud based' but are certainly not modern, micro-services architected and API-rich native-cloud solutions.

The inflexibility and complexity of the preponderance of legacy and much modern-day technology and platforms is an anchor chaining insurers' innovation in a technology prison. Insurers need to deploy not just a technology ecosystem but also the ability to swap applications and platforms in and out as customer requirements, competition, and the transportation revolution evolve.

Rory Yates comes to our aid again.

"What's needed is a new system design built around a strong foundation of ecosystem enablement. A system that, like its e-commerce forebears, can provide the integration + data fluidity + rules-based system (high level of configurability) that ensures the insurer gets the capability needed to easily consume the value of these partners.

A good example is By Miles (recently bought by Direct Line Group in the UK), where the offering of the product is usage-based and uses mobile as well.

All this means utilizing a range of data sources, experimenting with them and creating new, more relevant business models for car insurance from how it's priced. This includes how the policy works, claims, downstream repairs & the settlement process. All of which demands a fundamental change in approach to system design and business model creation.

Right now, the reality for many insurance businesses is any change, no matter how big or small, is expensive and takes significant time. This means that every change needs to be thoroughly thought through. None of this bodes well for a speedy response to a car market that is changing rapidly.

The answer is to have the right foundations to adopt the ecosystem needed to take advantage of this emerging potential.

The car is transforming rapidly, the opportunities to work with the vehicle ecosystem have never been more straightforward, and a lot of the fail points of the past have been overcome. Insurers will need to adopt these capabilities and take advantage of the efficiencies, customer experience benefits and new proposition potential this will undoubtedly bring."

If I still have not persuaded you of the importance of tackling this challenge now rather than later your customers are already looking elsewhere. Nige Walsh highlighted the number of customers 'shocked' at the price of rising premiums. See " UK drivers complain as car insurance renewal costs rise up to 70%"

They are looking elsewhere for insurance cover and churn is a serious blow to profitability. In the USA, customers with bundles of auto and home policies are increasingly unbundling as they see premium increases in one or the other category, or likely both.

As Nigel commented on the article above:

"Interesting read - Customers typically only see or care about the number, not the cover or why the changes have or are occurring. And 70% gets the headline for sure!

As an industry, we can do more to share the Why - we may not like the answer, but if we take into account changes in IPT, Claims Inflation costs and, much more, it's clear why these increases are taking place."

Explaining cover in plain language may help but as carriers have promoted low cost as a key benefit for many years you can expect customers to be skeptical of the various "added-value" components of insurance cover.

This is the time to make sure you have the right product strategy, the best underwriting performance, the best claims settlement KPIs, and the right technology partners to build the vital ecosystems that deliver the service and customer satisfaction to win and keep customers.

InsurTech World

Contributors

InsurTech World Archive

The dangers of ‘Static vs Continuously updating risk factors’ using car theft examples

Further Reading