Alan O’Loughlin looks at the impact of the pandemic on motor insurance consumer behaviour, and the long-term impact for pricing

After a period in which motor premiums hit record lows, pricing professionals will want to understand how the changes that have occurred during the past two years will be factored into future pricing strategies. Should this period, in which we saw changes to the frequency and severity of claims, and an increase in fraud, be treated as an outlier and ignored for future pricing?
The financial crisis in 2008 taught us a valuable lesson: that historical events should not be treated as ‘anomalous data’ and removed when modelling future risk. A pandemic should not be treated any differently; we just need to make sure the models don’t overfit or over-correct for it. We must find a new way to model for the new norm.
By looking at a combination of data on claims, policy renewal patterns, gaps in policy cover, policy cancellations, and customers exiting the insurance market, motor insurance providers have an opportunity to segment risk based on data surrounding the consumers’ behaviour before, during and after the pandemic.
Claims and fraud
The pandemic and resulting lockdowns caused significant disruption to the number of vehicles on the road, but while UK traffic volumes reduced by roughly 23% in 2020, road fatalities only dropped by 16%, meaning that, relative to traffic volume, the mortality rate increased year-on-year by 9%. Looking at this data in tandem with telematics driving behaviour data for 2020 and 2021 suggests that people’s behaviour became more reckless during lockdown, when the roads were emptier. This behaviour continued into the period when the roads suddenly became much busier, resulting in far more severe crashes.
Fraud was also a bigger challenge for the market. While the overall downward trend in claims matched the number of drivers on the road, the number of fraudulent motor insurance claims in 2020 was 50% higher than in 2019. Some consumers took advantage of the disruption to insurance providers to commit fraud.
These two concerning trends will influence insurance pricing in the mid to long-term.
Policy renewal and cancellation
From 2019 to September 2021, shopping around and switching of insurer and renewal activity remained relatively consistent, with only an initial small fall in renewals in March 2021, and a small uptick in people leaving the market the same month. The number of policy cancellations during the pandemic decreased year on year; this may be a result of the disruption to new car sales, as we know shopping for and buying a new car is often a trigger for cancelling a policy.
We’ve seen that policy cancellations have a direct correlation with insurance claims and are therefore a powerful attribute to help calculate underwriting rules and deliver more accurate pricing. Policy cancellation variables capture a lot of different behaviours, from people switching vehicle regularly and driving cars that they are unfamiliar with, to people being dishonest about their claims history, defaulting on their policy payments or writing off their car after a crash. We see a 7% lower loss cost relativity for people without a mid-term cancellation; in contrast, people who have more than six mid-term cancellations have a 230% higher loss cost relativity when compared to a book average.
However, if a customer cancelled their policy or had a gap in cover during a lockdown period versus outside of that time, insurance providers must consider factoring in this ‘pandemic-induced’ behaviour to help ensure the customer is priced fairly at new business or renewal.
Seeing the whole picture
We’ve seen a disruption to new car sales, but a fall in cancellations overall. Claim volumes dropped, but opportunistic fraud and mortality rates spiked. Diligence is required, and pricing professionals shouldn’t ignore the change in behaviours that occurred over lockdown as an outlier. Look at data that shows consumer policy buying behaviour before and during the pandemic. If you have seen cancellations rise, put that behaviour in context – was it during a lockdown period? Could it be an indication of financial vulnerability? Look at these attributes not in a silo, but alongside additional data sources pertaining to the individual’s insurance history. With more data, it becomes easier to segment and model for behaviours before and after the pandemic.
Have you seen policy behaviour changes? Are you using the right attributes? Greater data enrichment not only benefits pricing and underwriting, it can also support the whole customer journey – from application to claim.
Alan O’Loughlin is senior director of Data Science, International, Insurance at LexisNexis Risk Solutions