Click here to load reader
Click here to load reader
Jan 21, 2020
A Long-term Antidote? Insurers in numerous countries are looking to telematics as they seek to control claims costs, enhance pricing, increase profitability and differentiate and personalise their products and services. The UK now leads the world market in telematics adoption by both take up per capita and adoption rates by Insurers with both sides of the UK Government encouraging telematics politically. Since the launch of price comparison sites in the UK market a decade ago, multimillion pound advertising campaigns that highlight the benefits to customers of searching for the best price deal online are the norm, and these sites have become increasingly popular with UK consumers. This was layered upon what was already the most competitive motor insurance market place in the world. The effect of intense online competition, together with rising fraud levels and a legal environment that encourages claims farming, has resulted in a cut throat and largely unprofitable market. It is a market where fluctuating prices and uncertainty of returns prevail. UK Insurers have been frantically searching
for the antidote. This search has led to a period of innovation in UK motor insurance pricing techniques. Portfolio pricing, optimisation tools and data enrichment (including leveraging credit scores) have been presented as solutions; however with the majority of the market now using these, the competitive advantage of the early adopters has worn off and the past two years has seen UK Insurers starting to turn to Telematics or Usage Based Insurance (UBI). UBI for Insurers represents a shift away from pricing based on an approximation of individual risk to pricing based on individual risk (pricing on a sample of one). In addition to pricing innovation UBI offers self-selection, deeper customer engagement, more effective channel management and reduced fraud exposure which is partly why it is being presented as a long-term antidote, rather than a shorter term pricing quick-win. Increased customer awareness and significantly lower cost data acquisition thanks to smartphone applications will assist UBI in the UK to spread from the niche high risk areas out into the lower risk mass market from 2014. The UK innovations in telematics will ripple out into international markets with early adopting Insurers and Brokers in these markets benefiting.
THINKING OUTSIDE OF THE BOX For too long now Insurers have been trapped inside the ‘hamster wheel’ of market underwriting cycles, hoping that today’s customers are more profitable than yesterday’s, yet knowing that they can’t be certain of this until tomorrow. Can telematics help us break out of this cycle and what might global Insurers learn from developments in the UK?
Paul Stacy R&D Director
David Neave Non - Executive Director
What is Insurance Telematics?
Vehicle telematics is the technology of recording, sending, receiving and storing information via telecommunication devices in vehicles. Insurance telematics, better known by consumers as ‘Blackbox’ Insurance has many other names within the industry such as; ‘Pay As You Drive’, ‘Pay How You Drive’ and more recently ‘Usage Based Insurance’ (UBI).
Driving Behaviour Scoring For some time Insurers in all major global markets have competed to identify and measure risk characteristics that are predictive of loss propensity. Current pricing entails the application of multivariate statistical techniques to large and ideally clean sets of historical data. Historical data sets need to contain fields representing all risk characteristics included in the pricing plan. In addition, to yield statistically significant results demands sizeable data sets. There are two aspects of the current approach on which telematics can significantly improve. Firstly there is no feedback unless a claim is recorded (this will only happen to a small percentage of the population) and this introduces challenges leading to uncertainty on sample sizes, particularly for smaller Insurers. Secondly, the timeline between incident, notification, estimation and settlement introduces further challenges and uncertainty and pushes back the intervention point for pricing or account management decisions. Figure 1 provides an outline, the aim being to set the price £x equal to the ultimate experience £y. UBI offers the opportunity to radically reengineer this process and significantly improve upon or enrich the existing pricing approach.
Figure 1. Uncertainty Pricing
Pricing Factors Without the benefit of actual driving behaviour data, Insurers have relied upon proxies such as driver age, driving experience, location and historical claims records. Recent experience indicates that directly measuring driving behaviours can dramatically improve pricing accuracy when combined with traditional rating factors. The most predictive driving parameters include: • Verified mileage and garaging • Time of day – night time driving • Day of week • Type of road • Speeding – amount of incidents exceeding road speed
limits • Smoothness – rate of hard braking and cornering
events • Familiarity – driving on known roads
• Excessively high levels of short journeys • Congestion – actual speed relative to road speed limit • Pace – speed relative to other drivers at that point on
the road • Confidence - absence of firm acceleration on joining
slip roads and junctions As just one example to evidence the predictive power of these driving parameters refer to Figure 2, which shows the relative fault claim rates per mile on different UK roads. There are 11.6 times more claims per mile travelled on 40mph roads and below (e.g., 20mph and 30mph) compared to a 70mph motorway.
Figure 2. Relative Fault Claim Rate Per Mile
A driving score is typically a weighted combination of four or more driving parameters. If we’re to assume a scale between 1 and 100 of driving scores (1 representing the worst end of the scale and 100 being the best), the raw behavioural data in the form of time stamped longitudes and latitudes may come from a wired device, ODB2 device or smartphone. This raw data, owned by the customer, needs to be validated and enriched with mapping data before different driving parameters can be determined.
The law and insurance codes of conduct vary between countries and/or territories with regard to the use of GPS data. Regardless of this, however, the criticality of accurate mapping data in delivering the customer proposition cannot be underestimated. For example inaccurate road speed limit data and/or poor validation will lead to inaccuracy in driving scores, customer complaints and a lack of customer confidence in reporting and feedback. It will also then impact the overall success of any telematics product; Figure 3 shows how tricky this can be.
Figure 3. The Importance of Road Speed Limits Accuracy and Validation
UK Road Speed Limit
Percentage Total Miles
Relative Fault Claim Rate
=> 40mph 75645841 42% 11.6
50 - 60mph 53317470 29% 4.8
70mph 52624061 29% 1.0
Personalised Pricing Figure 4 shows a distribution of actual driving scores across a population of drivers, it also highlights the range of forecasted loss ratios against driving score. It is necessary to show a range of loss ratios due to different scoring approaches. The population of drivers achieving a driving score of 25 or below will have a loss ratio of circa 135% versus a loss ratio of circa 38% for the population scoring above 75. The data in this chart is based on our emerging experience to date from a number of our UK partners.
Figure 4. Distribution of Driving Scores and Loss Ratios
This correlation is exciting but poses the question: is driving behaviour a new addition to already existing methods or is driving behaviour already implicitly included in traditional rating factors? To answer this Actuaries would point to results from a generalised linear model which would show the variability in driving score is already taken into consideration (or not) by the changes in one or more of the variables (e.g., traditional rating questions). We have chosen to demonstrate that driving behavior adds new insight against traditional factors visually.
Figure 5 plots a driving score against claims’ premiums. UK data was used in this example and the driving scores were retrospectively calculated and compared to traditionally generated...
Figure 5. Premium Vs Driving Score
...premiums. This shows that despite the strong correlation between driving score and loss ratio, there is almost no correlation between driving score and claims outcomes or premium, thus reinforcing the conclusions from Figure 4 where prices have been set largely using traditional pricing. The lack of any relationship is evidence that driving behaviour adds something substantially new and incremental to traditional pricing models. Even when the comparison was completed against individual traditional factors such as age, no claims bonus, etc, still little correlation to driving score was observed across circa 23,000 records. So driving scores are correlated to fault claims and they offer a new incremental insight in terms of managing an insurance portfolio, but how are these scores distributed and what can this tell us about how insurers can manage their businesses? If scores show a ve