[Skip to content]

Sign up for our daily newsletter
The Actuary The magazine of the Institute & Faculty of Actuaries

Insurance: In case of injury

In October 2007, the UK insurance industry published the biggest-ever analysis of bodily injury trends, involving analysis of more than two million claims and 15 million transactions. Looking at the fourth UK Bodily Injury Awards Study alongside the first one, published ten years ago, brings home just how much actuarial techniques and software have advanced in a short space of time. It is also a tribute to the greater technical capability of the general insurance industry, especially in the areas of data gathering and retrieval.

Historical shortcomings
Back in 1997 the first study analysed some 100,000 claims, enabling the proud claim that it was the biggest exercise of its kind ever undertaken in the UK. The findings were, by the standards of what actuaries can do today, both tentative and basic.

It found that claims had been rising at around 13% p.a. for the previous decade. It did not, however, differentiate between different sizes of claim, other than to say that the larger ones were ‘probably’ going up at a faster rate. It did not identify with any confidence the relative influence of claims severity and claims frequency. Nor did it differentiate between different lines of business, such as comprehensive, non-comprehensive, private, fleet and motorcycle.

And yet, for all its shortcomings, the first study was considered a breakthrough, not least because it involved unprecedented levels of co-operation by individual insurers. There had been a vigorous debate, especially among the largest companies, as to whether it was in their interests to participate at all. Many executives argued that, by handing over their data, they were reducing their competitive advantage.

In the event, sufficient companies agreed to participate to make it a worthwhile exercise. It did, however, reveal another big shortcoming: the way data was presented. Much of it was so poor or difficult to reconcile as to be of no use.

IT Challenges
The second and third studies, published in 1999 and 2003, saw a huge increase in the willingness of insurers to take part, and the number of claims analysed rose to 1.2 million. So, too, did the quality of the data; the value of the studies and the range of findings rose commensurately.

This brings us to the present study where, with just one exception, every insurer of any size agreed to participate. Our sample included 75% of all motor bodily injury claims for the decade to 2006, rising to 90% for the most recent years. This is a tribute to the industry, because it was a difficult decision for some companies. Leaving aside questions of competitive advantage, taking part was time-consuming as well as voluntary.

It placed great demands on IT departments. For the data to be useful, it had to be presented in a format consistent with the template. Some companies went through the whole process twice, as their first offerings were not as they needed to be. One even had to have a third go, illustrating an impressive level of willingness.

There remain areas for potential improvement. Better and more detailed recording of claimant data and claims data split by type of injury, further divided into general and special damages sections, would greatly assist future analysis. Nonetheless, the data was mostly of a high quality, and much better than before.

Once we could be confident of getting the data, a big challenge was how to use it. The International Underwriting Association (IUA)’s actuarial working party was a valuable source of ideas and guidance. However, there was a huge range of opinion as to what questions we should be answering and the methodologies to use. Had we pursued every option, the findings would have taken years to produce. An early priority was, therefore, to get a consensus on a focussed approach, and then to stick to it.

Claims escalation
Another issue to resolve was the definition of ‘claims inflation’. Some people, speaking informally, mean the aggregate amount paid in claims by insurers. This does not take into the account the rising number of vehicles on the road, and therefore of insurance policies. Nor does it distinguish between claims frequency and severity. We prefer to describe the combined effect of these pressures as ‘claims escalation’.

Another option for defining inflation, which would also have been misleading, would have been to use the average size of each claim as a measure of inflation. The biggest increase in frequency was among small claims in the £2k to £5k range. This inevitably pushes down the average award, whereas a more meaningful figure would be to measure claims trends on a like-for-like basis. We ultimately opted to define inflation as the average increase in burning cost per policy.

We also had to define ‘year’ – a concept that is more contentious to some actuaries than to other members of the public. We chose ‘accident year’ to be the main period from which to measure claim inflation. This provides a composite view of inflation on a year-by-year basis, allowing for the collective influence of all issues affecting the claims process.

These questions – and there were several others – give a flavour of the complexity of the exercise. And that, of course, was before we had started the analysis itself, which was technically immensely complicated and time-consuming.

Trends in both claims frequency and severity are affected by internal factors such as changes in claims department practices, in reserving philosophy, claims settlement procedures and notification patterns. Between insurers, there can be significant differences in claims practices. Equally important, there are variations in the mix of business. All this means that the interpretation of data trends can lead to misleading conclusions, unless handled correctly.

Moving on from multivariate chain ladder techniques
In previous studies we used a multivariate chain ladder technique to address such issues. This time around advances in supporting actuarial software made it feasible instead to analyse each contributing portfolio’s experience separately, taking into account trends specific to each company. A range of standard actuarial reserving techniques were applied to recognise and make appropriate allowance for each company’s claims development experience. For each company, trends in claim frequencies, claim severities and overall burning costs were analysed and projected. Analysing each contributing company’s experience in this way allows a more accurate assessment and projection of claims cost than the method used in previous studies.

In broad terms the analyses performed covered the following issues:
>> Trends in bodily injury claims where each claim arising from an “accident year” has been “capped” at £85,000 in current monetary values.
>> Trends by “accident year” in bodily injury claims for claims in excess of the “capping limit”.
>> Trends in bodily injury claims for a range in small claim size bands. This was intended to provide further insight into small claim trends, in particular, to understand at which claim size ranges are affected the most.
>> Trends in bodily injury claims for a range in large claim size bands.
>> Trends in bodily injury claims for the largest claims. This looks at the trends in the top 0.1%, 0.5% and 1% of claims. We have also compiled the top 50 claims in each year which is often of great interest to most people.

Once the analysis was complete, there remained one important challenge. It is of very limited value to produce what you hope will be a technically excellent report if it can only be understood by other actuaries. We had to make it accessible to a wide range of interested parties such as underwriters, claims staff and senior executives.

Apart from the disciplines required to use plain English, we sought to achieve this objective by separating out the main commentary from the detailed analysis. The report itself was written to be comprehensible to the knowledgeable layman. The appendices, by contrast, were created for the benefit of technically-minded individuals. There are enough graphs and charts to keep any actuary occupied: seventy two pages of them.

Insurers put a huge amount of effort into making this study possible, appreciating the value of gaining a detailed understanding of this important subject and providing much more, better data than ever. New, more advanced reserving and modelling software gave us the power and flexibility to make full use of this additional information. The outcome: the most useful UK Bodily Injury Awards Study so far.

Fourth UK Bodily Injury Awards Study – headline findings
>> The total cost of bodily injury claims paid out by UK motor insurers increased at an annual rate of 9.5% between 1996 and 2006. These increases are largely due to legislative and legal changes.

>> The number of claims rose at an annual rate of nearly 3%, despite a 19% drop in the number of people killed or injured on British roads during this period. This increase was especially marked for claims between £2,000 and £5,000

>> The average value of claims increased at an annual rate of 6.5% between 1996 and 2006. The bigger the claim, the bigger the percentage increase; claims inflation at £2m - £5m was 12% rising to approximately 30% for claims of more than £5 million.

>> Legal costs have continued to increase. Insurers pay lawyers 43p for every pound paid in compensation, representing a billion-pound industry.

>> Personal injury claims in 2006 accounted on average for £114 of every private comprehensive Motor premium, £217 for the average non-comprehensive policy.

>> Uninsured driving costs around £500m, funded by the industry through the Motor Insurers’ Bureau, and is equivalent to £30 for every Motor policy.

>> Speed of settlement of claims above £100,000 has increased substantially.

Mike Brockman is a partner at specialist non-life actuaries EMB. The Fourth UK Bodily Injury Awards Study is published by the IUA, price £250. Contact Deborah Finch, Deborah.finch@iua.co.uk