Trending to the period over which new rates will apply

[Joe Warren]

When we trend our claim amounts/frequencies/severities to the period in which the new rates will apply, do we always trend to the midpoint of the exposure on which the new rates will apply?

What if there was a reporting delay of 6 months - would we still trend to the average date of occurrence, or would we trend to the average date of occurrence + 6 months, as that is when the claim is reported and the reserve will be set?

I presume in any case in a question you could just set an assumption that losses will be trended to the midpoint of exposure.

 

[Darren Michaels]

Typically you would trend from the average date of payment in your historical data to the average date of payment of the claims that are expected to arise from the period of exposure that you calculating the premiums for.

So, if you assume no reporting and no settlement delays then the latter is the same as the average date of occurrence.

 

[Joe Warren]

Thank you for the reply.

That makes sense - two parts of the course notes threw me off.

Firstly, C14 P8 there is a line under 'General Considerations' that says we 'first project historical frequencies and severities in line with assumed trends to current values and then project them to the mid-point of the future exposure period.' I presume this is just a generalisation, and if there were settlement/reporting delays then we would need to project to the average date of payment (accounting for the delays), and not just to the mid-point of the future exposure period?

Secondly, C12 P19 details an example of adjusting a historical risk premium to be representative of a risk premium in a new rating period. It is noted in the question that there is a 10 month settlement delay, but the trending is only performed until the mid-point of the exposure on which the new rates will apply, and not to the average date of payment with the settlement delay accounted for (i.e. it trends from mid-2020 as our risk premium is based on losses occurring in 2020 to the end of 2022 as we are interested in finding a risk premium for business written in 2022). Is this just a generalisation, and if this scenario was in the exam would it be correct to actually trend to the average date of payment (i.e. 10 months after the end of 2022)?

Thanks,
Joe

 

[Joe Warren]

Hi,

One more question,

In order to trend our claims data do we trend from the average date of claim payment in the historical data to the average date of claim payment for claims arising on policies from the period of exposure on which the new rates will apply, for all trending items?

I.e. is that the typical period of adjustment time for changes in things such as: underwriting, cover, frequency and severity trends?

Many thanks in advance,
Joe

 

[Darren Michaels]

Thank you for the reply.

That makes sense - two parts of the course notes threw me off.

Firstly, C14 P8 there is a line under 'General Considerations' that says we 'first project historical frequencies and severities in line with assumed trends to current values and then project them to the mid-point of the future exposure period.' I presume this is just a generalisation, and if there were settlement/reporting delays then we would need to project to the average date of payment (accounting for the delays), and not just to the mid-point of the future exposure period?

Secondly, C12 P19 details an example of adjusting a historical risk premium to be representative of a risk premium in a new rating period. It is noted in the question that there is a 10 month settlement delay, but the trending is only performed until the mid-point of the exposure on which the new rates will apply, and not to the average date of payment with the settlement delay accounted for (i.e. it trends from mid-2020 as our risk premium is based on losses occurring in 2020 to the end of 2022 as we are interested in finding a risk premium for business written in 2022). Is this just a generalisation, and if this scenario was in the exam would it be correct to actually trend to the average date of payment (i.e. 10 months after the end of 2022)?

Thanks,
Joe

Hi Joe

Yes basically in both cases they are generalisations and in fact it depends on exactly what you are assuming. For example, you may assume that there are no reporting and settlement delays or that reporting and settlement delays remain constant over time.

It also depends on how you interpret the data you have been given. For example, in the calculation in Chapter 12, we are effectively assuming that the £267 already allows for the 10 month settlement delay (and so is a number that is actually from the end of April 2021 - 10 months after mid 2020) and so when we project this forward by 2.5 years we have already included an implicit allowance for the same length of settlement delay arising in the future.

You could of course choose to interpret this differently and make an alternative assumption as you have suggested. Generally speaking as long as you state your assumptions clearly and carry out your calculations in a manner consistent with those, you should still get the appropriate credit in the exam.

Note the comment on page 19 of Chapter 12 which alludes to this "For this sort of question there are often many acceptable ways in which a suitable answer could be calculated depending on how you interpret the data given, and depending on any further
assumptions you make. One answer can be derived as follows."

 

[Darren Michaels]

Hi,

One more question,

In order to trend our claims data do we trend from the average date of claim payment in the historical data to the average date of claim payment for claims arising on policies from the period of exposure on which the new rates will apply, for all trending items?

I.e. is that the typical period of adjustment time for changes in things such as: underwriting, cover, frequency and severity trends?

Many thanks in advance,
Joe

Strictly speaking, the start and end points should vary depending on what it is you are adjusting for, eg changes in underwriting standards or cover will need to be adjusted from the mid point of the exposure period that has given rise to the claims in your base data to the mid point of the exposure period for which you are calculating the premium rates for. This is because these factors will depend on the policy terms in force when the claim event happens. However, to allow for claims inflation and other claims cost trends you typically want to adjust from the average date of payment to the average date of payment. This is because these factors depend on when the claim amount is actually paid out.

However, if we assume all periods and delays (eg reporting and settlement delays) are unchanged over time, then this should mean that the difference between the start and end points of these periods would normally be the same length, eg 2.5 years or whatever.

 

[Joe Warren]

Thank you - that is all very clear.